MH survey analysis
Abbreviations used in the markdown/comments of this script
- LR: Logistic regression
- FLR: Factorial Logistic regreesion
- OLR: Ordinal logistic regression
- KW: Kruskal-Wallis test
- OR: Odds ratio
- CI: Confidence interval (two-sided 95% confidence)
- glm: Generalised linear model
- CS: career stage
- PNTA: Prefer not to answer
A quick introduction to the statistical tests and choices
This section intends to give a high level explanation to the code that follows. The information provided in this section does NOT supercede the methods section of the report - please always regard to the information provided in the report as final.
Independent variables
We checked mainly for the effect of 2 independent variables on the responses: - Gender - men vs women (variable name: All_Gender_clean) - Career stages - PhDs vs Postdocs vs Early career PIs vs mid-career PIs vs late-career PIs (variable: Supporter_CareerStage_clean) We had respondents of other genders and career stages, but for this report, we chose to only analyse these groups because they had a reasonable and comparable number of samples. (If you’d like to look into other groups specifically, you’re more than welcomed!)
For some questions we also checked the effect of whether someone is a supporter on their response (variable name: Provided_Support_clean). Supporter status is determined based on their response to the statemnet “I have provided support to someone who was doing research and who was struggling with their mental health”. We compared respondents who answered “No” and one of the “yes” options to this statement. For some questions we also wanted to see if early career PI stands up from other career stages. For this, we run either (1) Kruskal-Wallis test between early PI and others pooled together; or (2) pairwise comparison between early PIs and all other groups. In the second case, Sidak correction was applied posthoc. ### Dependent variables, i.e. the response A wide variety of questions were asked - we focussed our statistical analyses on two types of responses (or responses that could be logically recoded to one of the following two types): - Categorical, and in most cases, binary, i.e. Yes/No - Ordinal, i.e. Likert scale type responses
For categorical responses, we modelled response using (binomial) logistic regression. We ran 3 seperate models - one with genders as predictors, the second with career stages (CS) as a predictors, and the final with both as predictors. For gender, men is always the baseline indicator variable, and for CS, PhD is always the baseline indicator variable. The first two models provide the reported p-values, odds ratios (OR, obtained with exp(coef()) and confidence intervals of the OR (two-sided 95% confidenece, obtained with exp(confint())). This final factorial logistic regression allows us to check for interactions between gender and CS.
For ordinal responses, we modelled responses using ordinal logistic regression (OLR). We ran 2 seperate models - one with genders as predictors, the second with career stages (CS) as a predictors. For gender, men is always the baseline indicator variable, and for CS, PhD is always the baseline indicator variable. These give the ORs and CIs reported. As the ordinal responses we’ve obtained are usually not normally distributed (code not reported here, but normality was tested using Shapiro.test()), we chose to use the Kruskal-Wallis test to test for the effect of gender/CS on the ordinal responses. To check for any interactions between the two independent variables we also split the data into groups by one independent variable and used the chi-square test to check for the effect of the other independent variable (i.e. within all women, is there an effect of CS, etc).
Initialising
Load and inspect data structure
data=read.csv("../mh-data/cleandata2604 REDACTED.csv")
# colnames(data)##Filtering the data Looking at only 2 independent variables:
Gender (label: “All_Gender_clean”): Man -1, Woman -3
Career stage (label “Supporter_CareerStage_clean”):
- A PhD student-3;
- A postdoctoral researcher (postdoc) - 4;
- A group leader (less than 5 years’ experience)- 6
- A group leader (5 to 10 years’ experience) -7
- A group leader (10 or more years’ experience) -8
genderCS=data[data$All_Gender_clean %in% c(1,3) & data$Supporter_CareerStage_clean %in% c(3,4,6,7,8), ]
genderCS$All_Gender_clean=as.factor(genderCS$All_Gender_clean)
genderCS$Supporter_CareerStage_clean=as.factor(genderCS$Supporter_CareerStage_clean)
dim(genderCS)## [1] 1255 196levels(genderCS$All_Gender_clean)=c("Men", "Women")
levels(genderCS$Supporter_CareerStage_clean)=c("PhD students", "Postdocs", "Group leaders (<5yr)", "Group leaders (5-10yr)", "Group leaders (>10yr)")the eLife colour palette
eGreen="#346A2D"
eLime="#7DB441"
eBlue="#06589C"
eSky="#2997D4"
ePurple="#881350"
eFuschia="#D81F62"
eGrey="#666B6E"
ePalette=c(eGreen, eLime, eGrey, eSky, eBlue)Q28xQ04 Gender x Career stage (CS) (Section 2, Panel 1)
chi-square
gender=genderCS[,c("Supporter_CareerStage_clean","All_Gender_clean")]
dim(gender)[1] #this gives N## [1] 1255table(gender)## All_Gender_clean
## Supporter_CareerStage_clean Men Women
## PhD students 160 374
## Postdocs 123 233
## Group leaders (<5yr) 66 89
## Group leaders (5-10yr) 48 43
## Group leaders (>10yr) 77 42chisq.test(table(gender))##
## Pearson's Chi-squared test
##
## data: table(gender)
## X-squared = 62.364, df = 4, p-value = 9.236e-13Q28xQ04 Gender x Career stage (CS) (Graph 2.1.2)
gender$Supporter_CareerStage_clean=factor(gender$Supporter_CareerStage_clean) #be careful when rerunning this line
# This als all other eps images will be output to your code directory inside the images sub-folder
#eps("images/CareerStage_gender.eps", width=1000, height=578)
graphdata = gender %>%
group_by(Supporter_CareerStage_clean, All_Gender_clean) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'Supporter_CareerStage_clean' (override with `.groups` argument)ggplot(graphdata, aes(x=Supporter_CareerStage_clean, y=perc, fill=All_Gender_clean)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=20))+
geom_text(aes(label=round(perc, digit=1)), size=8, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", size=3, title="During my supporting role, I was:") +
guides(fill=guide_legend(reverse=TRUE)) +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=c(eFuschia, ePurple))
ggsave("2-1-2.png")## Saving 7 x 5 in imageQ01 Providing support
Dependent variable: Provided_Support
- 1- No
- 2 - Yes, 1 individual
- 3 - Yes, 1-5
- 4- Yes, >5
Treat Dependent variable as ordinal
Provided_Support=genderCS[,c("Provided_Support", "Supporter_CareerStage_clean", "All_Gender_clean")]
Provided_Support=Provided_Support[Provided_Support$Provided_Support %in% seq(1,4), ] #remove NAs and some choices
dim(Provided_Support)[1] #N## [1] 1254Provided_Support$Provided_Support=as.factor(Provided_Support$Provided_Support)
table(Provided_Support)## , , All_Gender_clean = Men
##
## Supporter_CareerStage_clean
## Provided_Support PhD students Postdocs Group leaders (<5yr)
## 1 0 0 0
## 2 35 25 17
## 3 91 78 43
## 4 34 20 6
## Supporter_CareerStage_clean
## Provided_Support Group leaders (5-10yr) Group leaders (>10yr)
## 1 0 0
## 2 7 10
## 3 39 52
## 4 2 15
##
## , , All_Gender_clean = Women
##
## Supporter_CareerStage_clean
## Provided_Support PhD students Postdocs Group leaders (<5yr)
## 1 1 0 0
## 2 62 37 10
## 3 245 147 67
## 4 66 48 12
## Supporter_CareerStage_clean
## Provided_Support Group leaders (5-10yr) Group leaders (>10yr)
## 1 0 0
## 2 9 8
## 3 27 18
## 4 7 16Providing support x gender
exp(coef(polr(Provided_Support~All_Gender_clean, data=Provided_Support, Hess=TRUE, method="logistic")))## All_Gender_cleanWomen
## 1.24383exp(confint(polr(Provided_Support~All_Gender_clean, data=Provided_Support, Hess=TRUE, method="logistic")))## Waiting for profiling to be done...## 2.5 % 97.5 %
## 0.9852717 1.5713866kruskal.test(Provided_Support ~ All_Gender_clean, data=Provided_Support)##
## Kruskal-Wallis rank sum test
##
## data: Provided_Support by All_Gender_clean
## Kruskal-Wallis chi-squared = 3.3578, df = 1, p-value = 0.06689Providing support x Career stage
exp(coef(polr(Provided_Support~Supporter_CareerStage_clean, data=Provided_Support, Hess=TRUE, method="logistic")))## Supporter_CareerStage_cleanPostdocs
## 1.0478760
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 0.8164628
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 0.7695144
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 1.4508723exp(confint(polr(Provided_Support~Supporter_CareerStage_clean, data=Provided_Support, Hess=TRUE, method="logistic")))## Waiting for profiling to be done...## 2.5 % 97.5 %
## Supporter_CareerStage_cleanPostdocs 0.7957483 1.379945
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.5699034 1.170022
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.4942601 1.199812
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.9633699 2.181784kruskal.test(Provided_Support~Supporter_CareerStage_clean, data=Provided_Support)##
## Kruskal-Wallis rank sum test
##
## data: Provided_Support by Supporter_CareerStage_clean
## Kruskal-Wallis chi-squared = 7.2295, df = 4, p-value = 0.1242chisq for intersections [NOT REPORTED – EMMY]
#split by career stages, gender effect
chisq.test(table(Provided_Support[Provided_Support$Supporter_CareerStage_clean==levels(Provided_Support$Supporter_CareerStage_clean)[1],c("Provided_Support","All_Gender_clean")]))## Warning in
## chisq.test(table(Provided_Support[Provided_Support$Supporter_CareerStage_clean
## == : Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: table(Provided_Support[Provided_Support$Supporter_CareerStage_clean == levels(Provided_Support$Supporter_CareerStage_clean)[1], c("Provided_Support", "All_Gender_clean")])
## X-squared = 4.2632, df = 3, p-value = 0.2344chisq.test(table(Provided_Support[Provided_Support$Supporter_CareerStage_clean==levels(Provided_Support$Supporter_CareerStage_clean)[2],c("Provided_Support","All_Gender_clean")]))## Warning in
## chisq.test(table(Provided_Support[Provided_Support$Supporter_CareerStage_clean
## == : Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: table(Provided_Support[Provided_Support$Supporter_CareerStage_clean == levels(Provided_Support$Supporter_CareerStage_clean)[2], c("Provided_Support", "All_Gender_clean")])
## X-squared = NaN, df = 3, p-value = NAchisq.test(table(Provided_Support[Provided_Support$Supporter_CareerStage_clean==levels(Provided_Support$Supporter_CareerStage_clean)[3],c("Provided_Support","All_Gender_clean")]))## Warning in
## chisq.test(table(Provided_Support[Provided_Support$Supporter_CareerStage_clean
## == : Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: table(Provided_Support[Provided_Support$Supporter_CareerStage_clean == levels(Provided_Support$Supporter_CareerStage_clean)[3], c("Provided_Support", "All_Gender_clean")])
## X-squared = NaN, df = 3, p-value = NAchisq.test(table(Provided_Support[Provided_Support$Supporter_CareerStage_clean==levels(Provided_Support$Supporter_CareerStage_clean)[4],c("Provided_Support","All_Gender_clean")]))## Warning in
## chisq.test(table(Provided_Support[Provided_Support$Supporter_CareerStage_clean
## == : Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: table(Provided_Support[Provided_Support$Supporter_CareerStage_clean == levels(Provided_Support$Supporter_CareerStage_clean)[4], c("Provided_Support", "All_Gender_clean")])
## X-squared = NaN, df = 3, p-value = NAchisq.test(table(Provided_Support[Provided_Support$Supporter_CareerStage_clean==levels(Provided_Support$Supporter_CareerStage_clean)[5],c("Provided_Support","All_Gender_clean")]))## Warning in
## chisq.test(table(Provided_Support[Provided_Support$Supporter_CareerStage_clean
## == : Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: table(Provided_Support[Provided_Support$Supporter_CareerStage_clean == levels(Provided_Support$Supporter_CareerStage_clean)[5], c("Provided_Support", "All_Gender_clean")])
## X-squared = NaN, df = 3, p-value = NA#split by gender, career effect (same as KW above)
chisq.test(table(Provided_Support[Provided_Support$All_Gender_clean==levels(Provided_Support$All_Gender_clean)[1],c("Provided_Support","Supporter_CareerStage_clean")]))## Warning in chisq.test(table(Provided_Support[Provided_Support$All_Gender_clean
## == : Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: table(Provided_Support[Provided_Support$All_Gender_clean == levels(Provided_Support$All_Gender_clean)[1], c("Provided_Support", "Supporter_CareerStage_clean")])
## X-squared = NaN, df = 12, p-value = NAchisq.test(table(Provided_Support[Provided_Support$All_Gender_clean==levels(Provided_Support$All_Gender_clean)[2],c("Provided_Support","Supporter_CareerStage_clean")]))## Warning in chisq.test(table(Provided_Support[Provided_Support$All_Gender_clean
## == : Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: table(Provided_Support[Provided_Support$All_Gender_clean == levels(Provided_Support$All_Gender_clean)[2], c("Provided_Support", "Supporter_CareerStage_clean")])
## X-squared = 18.362, df = 12, p-value = 0.1051Q02 Not_ProvidedSupport (Section 2, Panel 2)
Dependent variable:Not_ProvidedSupport
- 1: No
- 2: Yes, once
- 3: Yes, a few times
- 4: Yes, often
- 5: I’m not sure -removed
- 6: PNTA -removed
NotProvidedSupport=genderCS[,c("Not_ProvidedSupport", "All_Gender_clean", "Supporter_CareerStage_clean")]
summary(NotProvidedSupport) #there are some NAs## Not_ProvidedSupport All_Gender_clean Supporter_CareerStage_clean
## Min. :1.000 Men :474 PhD students :534
## 1st Qu.:2.000 Women:781 Postdocs :356
## Median :3.000 Group leaders (<5yr) :155
## Mean :2.655 Group leaders (5-10yr): 91
## 3rd Qu.:3.000 Group leaders (>10yr) :119
## Max. :6.000#clean
NotProvidedSupport=NotProvidedSupport[(NotProvidedSupport$Not_ProvidedSupport %in% seq(1:4)),]
dim(NotProvidedSupport)[1] #this gives N## [1] 1131#factorise
NotProvidedSupport[NotProvidedSupport$Not_ProvidedSupport==1, "Not_ProvidedSupport"]="No"
NotProvidedSupport[NotProvidedSupport$Not_ProvidedSupport==2, "Not_ProvidedSupport"]="Yes, once"
NotProvidedSupport[NotProvidedSupport$Not_ProvidedSupport==3, "Not_ProvidedSupport"]="Yes, a few times"
NotProvidedSupport[NotProvidedSupport$Not_ProvidedSupport==4, "Not_ProvidedSupport"]="Yes, often"
NotProvidedSupport$Not_ProvidedSupport=factor(NotProvidedSupport$Not_ProvidedSupport, levels=c("Yes, often", "Yes, a few times", "Yes, once", "No"))Not provided support x gender graph
graphdata = NotProvidedSupport %>%
group_by(All_Gender_clean, Not_ProvidedSupport) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'All_Gender_clean' (override with `.groups` argument)#eps("images/NotProvidedSupport_gender.eps", width=1000, height=578) #[NOT REPORTED]
ggplot(graphdata, aes(x=All_Gender_clean, y=perc, fill=Not_ProvidedSupport)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=20))+
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="I have been in a situation when I thought someone in academia needed help, but I couldn’t or didn’t provide
support") +
scale_x_discrete(limits = rev(levels(graphdata$Supporter_CareerStage_clean))) +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=ePalette)## Warning: Unknown or uninitialised column: `Supporter_CareerStage_clean`.
dev.off()## null device
## 1Not provided support x CS graph (Graph 2.2.2)
graphdata = NotProvidedSupport %>%
group_by(Supporter_CareerStage_clean,Not_ProvidedSupport) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'Supporter_CareerStage_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/NotProvidedSupport_CS.eps", width=1000, height=578)
ggplot(graphdata, aes(x=Supporter_CareerStage_clean, y=perc, fill=Not_ProvidedSupport)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=10), legend.position="bottom", legend.text = element_text(size=7))+
geom_text(aes(label=round(perc, digit=1)), size=2.5, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="I have been in a situation when I thought someone in academia needed help, \nbut I couldn’t or didn’t provide
support") +
guides(fill=guide_legend(reverse=TRUE)) +
#coord_flip() +
scale_fill_manual(name="", values=ePalette)
ggsave(dpi=700, "2-2-2-v3.png")## Saving 7 x 5 in imagedev.off()## null device
## 1Stats (Section 2, panel 2)
recode to Yes - 1 / No-0 Model: FLR
# recoding response to binary variable
NotProvidedSupport$Not_ProvidedSupport_binary=NotProvidedSupport$Not_ProvidedSupport
levels(NotProvidedSupport$Not_ProvidedSupport_binary)=c(1, 1, 1, 0)
NotProvidedSupport$Not_ProvidedSupport_binary=as.numeric(NotProvidedSupport$Not_ProvidedSupport_binary)
NotProvidedSupport[NotProvidedSupport$Not_ProvidedSupport_binary=="2", "Not_ProvidedSupport_binary"]=0
# LR intersection between gender and CS
NotProvidedSupport_glm=glm(Not_ProvidedSupport_binary~Supporter_CareerStage_clean*All_Gender_clean, data=NotProvidedSupport, family="binomial")
summary(NotProvidedSupport_glm)##
## Call:
## glm(formula = Not_ProvidedSupport_binary ~ Supporter_CareerStage_clean *
## All_Gender_clean, family = "binomial", data = NotProvidedSupport)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.8757 -1.1894 0.7257 0.7352 1.1655
##
## Coefficients:
## Estimate
## (Intercept) 1.57022
## Supporter_CareerStage_cleanPostdocs -0.42159
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.82300
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -0.94161
## Supporter_CareerStage_cleanGroup leaders (>10yr) -1.54205
## All_Gender_cleanWomen -0.40015
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.45123
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.26912
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -0.01715
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 1.89180
## Std. Error
## (Intercept) 0.21555
## Supporter_CareerStage_cleanPostdocs 0.31164
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.35824
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.37722
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.32064
## All_Gender_cleanWomen 0.25028
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.37548
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.44894
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.51470
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.54138
## z value
## (Intercept) 7.285
## Supporter_CareerStage_cleanPostdocs -1.353
## Supporter_CareerStage_cleanGroup leaders (<5yr) -2.297
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -2.496
## Supporter_CareerStage_cleanGroup leaders (>10yr) -4.809
## All_Gender_cleanWomen -1.599
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 1.202
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.599
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -0.033
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 3.494
## Pr(>|z|)
## (Intercept) 3.22e-13
## Supporter_CareerStage_cleanPostdocs 0.176108
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.021597
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.012554
## Supporter_CareerStage_cleanGroup leaders (>10yr) 1.51e-06
## All_Gender_cleanWomen 0.109862
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.229457
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.548872
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.973413
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.000475
##
## (Intercept) ***
## Supporter_CareerStage_cleanPostdocs
## Supporter_CareerStage_cleanGroup leaders (<5yr) *
## Supporter_CareerStage_cleanGroup leaders (5-10yr) *
## Supporter_CareerStage_cleanGroup leaders (>10yr) ***
## All_Gender_cleanWomen
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1308.5 on 1130 degrees of freedom
## Residual deviance: 1269.0 on 1121 degrees of freedom
## AIC: 1289
##
## Number of Fisher Scoring iterations: 4# LR gender
summary(glm(Not_ProvidedSupport_binary~All_Gender_clean, data=NotProvidedSupport, family="binomial")) #glm summary with p-values##
## Call:
## glm(formula = Not_ProvidedSupport_binary ~ All_Gender_clean,
## family = "binomial", data = NotProvidedSupport)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.6506 -1.5954 0.7692 0.8107 0.8107
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.9440 0.1071 8.811 <2e-16 ***
## All_Gender_cleanWomen 0.1224 0.1378 0.889 0.374
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1308.5 on 1130 degrees of freedom
## Residual deviance: 1307.7 on 1129 degrees of freedom
## AIC: 1311.7
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Not_ProvidedSupport_binary~All_Gender_clean, data=NotProvidedSupport, family="binomial"))) #OR (men as baseline)## (Intercept) All_Gender_cleanWomen
## 2.570248 1.130252exp(confint(glm(Not_ProvidedSupport_binary~All_Gender_clean, data=NotProvidedSupport, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 2.0897373 3.181804
## All_Gender_cleanWomen 0.8616905 1.479460# LR Career stage
summary(glm(Not_ProvidedSupport_binary~Supporter_CareerStage_clean, data=NotProvidedSupport, family="binomial")) #glm summary with p-values##
## Call:
## glm(formula = Not_ProvidedSupport_binary ~ Supporter_CareerStage_clean,
## family = "binomial", data = NotProvidedSupport)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.7480 -1.3670 0.6995 0.7315 0.9990
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) 1.2830 0.1093 11.743
## Supporter_CareerStage_cleanPostdocs -0.1012 0.1727 -0.586
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.6123 0.2134 -2.870
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -0.8477 0.2487 -3.409
## Supporter_CareerStage_cleanGroup leaders (>10yr) -0.8012 0.2246 -3.567
## Pr(>|z|)
## (Intercept) < 2e-16 ***
## Supporter_CareerStage_cleanPostdocs 0.557970
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.004105 **
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.000653 ***
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.000361 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1308.5 on 1130 degrees of freedom
## Residual deviance: 1283.8 on 1126 degrees of freedom
## AIC: 1293.8
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Not_ProvidedSupport_binary~Supporter_CareerStage_clean, data=NotProvidedSupport, family="binomial"))) #OR (PhD as baseline)## (Intercept)
## 3.6074766
## Supporter_CareerStage_cleanPostdocs
## 0.9037547
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 0.5420841
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 0.4284032
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 0.4488034exp(confint(glm(Not_ProvidedSupport_binary~Supporter_CareerStage_clean, data=NotProvidedSupport, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 2.9240898 4.4893393
## Supporter_CareerStage_cleanPostdocs 0.6451819 1.2708518
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.3579850 0.8275075
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.2640113 0.7017920
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.2897130 0.7000910Q05 How often do they support someone who is officially their responsibility (Section 3, Panel 1)
Dependent variable: Supporter_Officially
- 1: No - not my official job
- 2: Yes - my official job
(remove 3- not sure, and 4 - PNTA)
Model: First individual factors (Gender, CS), then Factorial logistical regression Supporter_Officially ~ Supporter_CareerStage * All_Gender_clean) to check for interactions
#subsetting data and recording response to binary
Officially=genderCS[,c("Supporter_Officially", "Supporter_CareerStage_clean", "All_Gender_clean")]
Officially=Officially[Officially$Supporter_Officially %in% seq(1,2), ] #remove NAs and some choices
dim(Officially)[1]## [1] 1207Officially[Officially$Supporter_Officially==1, "Supporter_Officially"]=0 #recoding No->0 and Yes->1 for binom modelling
Officially[Officially$Supporter_Officially==2, "Supporter_Officially"]=1
# LR gender
summary(glm(Supporter_Officially~All_Gender_clean, data=Officially, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Supporter_Officially ~ All_Gender_clean, family = "binomial",
## data = Officially)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.7837 -0.7837 -0.5795 -0.5795 1.9325
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.0230 0.1069 -9.571 < 2e-16 ***
## All_Gender_cleanWomen -0.6763 0.1467 -4.609 4.05e-06 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1192.8 on 1206 degrees of freedom
## Residual deviance: 1171.7 on 1205 degrees of freedom
## AIC: 1175.7
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Supporter_Officially~All_Gender_clean, data=Officially, family="binomial"))) #OR (men as baseline)## (Intercept) All_Gender_cleanWomen
## 0.3595166 0.5084953exp(confint(glm(Supporter_Officially~All_Gender_clean, data=Officially, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.2905011 0.4418647
## All_Gender_cleanWomen 0.3811990 0.6779330# LR CS
summary(glm(Supporter_Officially~Supporter_CareerStage_clean, data=Officially, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Supporter_Officially ~ Supporter_CareerStage_clean,
## family = "binomial", data = Officially)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.4509 -0.3619 -0.2476 -0.2476 2.6459
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) -3.4696 0.2539 -13.667
## Supporter_CareerStage_cleanPostdocs 0.7769 0.3361 2.311
## Supporter_CareerStage_cleanGroup leaders (<5yr) 3.7912 0.3069 12.353
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 3.7505 0.3345 11.213
## Supporter_CareerStage_cleanGroup leaders (>10yr) 4.0928 0.3255 12.573
## Pr(>|z|)
## (Intercept) <2e-16 ***
## Supporter_CareerStage_cleanPostdocs 0.0208 *
## Supporter_CareerStage_cleanGroup leaders (<5yr) <2e-16 ***
## Supporter_CareerStage_cleanGroup leaders (5-10yr) <2e-16 ***
## Supporter_CareerStage_cleanGroup leaders (>10yr) <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1192.85 on 1206 degrees of freedom
## Residual deviance: 749.92 on 1202 degrees of freedom
## AIC: 759.92
##
## Number of Fisher Scoring iterations: 6exp(coef(glm(Supporter_Officially~Supporter_CareerStage_clean, data=Officially, family="binomial"))) #OR (PhD as baseline)## (Intercept)
## 0.0311284
## Supporter_CareerStage_cleanPostdocs
## 2.1746154
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 44.3103448
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 42.5439189
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 59.9087838exp(confint(glm(Supporter_Officially~Supporter_CareerStage_clean, data=Officially, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.01814131 0.04940077
## Supporter_CareerStage_cleanPostdocs 1.13130138 4.26963302
## Supporter_CareerStage_cleanGroup leaders (<5yr) 24.89038704 83.38081578
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 22.56652408 84.20421287
## Supporter_CareerStage_cleanGroup leaders (>10yr) 32.43059228 116.79151129# FLR gender-CS interactions
summary(glm(Supporter_Officially~Supporter_CareerStage_clean*All_Gender_clean, data=Officially, family="binomial"))##
## Call:
## glm(formula = Supporter_Officially ~ Supporter_CareerStage_clean *
## All_Gender_clean, family = "binomial", data = Officially)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.4632 -0.3706 -0.2792 -0.2331 2.6904
##
## Coefficients:
## Estimate
## (Intercept) -3.22552
## Supporter_CareerStage_cleanPostdocs 0.43523
## Supporter_CareerStage_cleanGroup leaders (<5yr) 3.71800
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 3.40784
## Supporter_CareerStage_cleanGroup leaders (>10yr) 3.87611
## All_Gender_cleanWomen -0.36630
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.51278
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.07449
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.56964
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.28625
## Std. Error
## (Intercept) 0.41628
## Supporter_CareerStage_cleanPostdocs 0.57002
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.49651
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.51474
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.48651
## All_Gender_cleanWomen 0.52540
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.70645
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.63228
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.68303
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.67812
## z value
## (Intercept) -7.748
## Supporter_CareerStage_cleanPostdocs 0.764
## Supporter_CareerStage_cleanGroup leaders (<5yr) 7.488
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 6.621
## Supporter_CareerStage_cleanGroup leaders (>10yr) 7.967
## All_Gender_cleanWomen -0.697
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.726
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.118
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.834
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.422
## Pr(>|z|)
## (Intercept) 9.30e-15
## Supporter_CareerStage_cleanPostdocs 0.445
## Supporter_CareerStage_cleanGroup leaders (<5yr) 6.98e-14
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 3.58e-11
## Supporter_CareerStage_cleanGroup leaders (>10yr) 1.62e-15
## All_Gender_cleanWomen 0.486
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.468
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.906
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.404
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.673
##
## (Intercept) ***
## Supporter_CareerStage_cleanPostdocs
## Supporter_CareerStage_cleanGroup leaders (<5yr) ***
## Supporter_CareerStage_cleanGroup leaders (5-10yr) ***
## Supporter_CareerStage_cleanGroup leaders (>10yr) ***
## All_Gender_cleanWomen
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1192.85 on 1206 degrees of freedom
## Residual deviance: 748.41 on 1197 degrees of freedom
## AIC: 768.41
##
## Number of Fisher Scoring iterations: 6How often do they support someone who is officially their responsibility x gender graph (Graph 3.1.2)
Officially[Officially$Supporter_Officially==0, "Supporter_Officially"]="No"
Officially[Officially$Supporter_Officially==1, "Supporter_Officially"]="Yes"
Officially$Supporter_Officially=factor(Officially$Supporter_Officially)
#ImpactPersonal$Impact_Personal=factor(ImpactPersonal$Impact_Personal, levels=c("No","Yes"))
graphdata =Officially %>%
group_by(All_Gender_clean, Supporter_Officially) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'All_Gender_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/Officially_gender.eps", width=1000, height=578)
ggplot(graphdata, aes(x=All_Gender_clean, y=perc, fill=Supporter_Officially)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=12), axis.title.x = element_text(size = 8), axis.title.y = element_text(size = 16))+
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="It was part of my official job description to be in charge of the person I helped") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=c(eBlue, eGreen))
ggsave(dpi=1000, "3-1-2.png", limitsize = FALSE)## Saving 7 x 5 in imagedev.off()## null device
## 1How often do they support someone who is officially their responsibility x career graph (Graph 3.1.1)
graphdata = Officially %>%
group_by(Supporter_CareerStage_clean, Supporter_Officially) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'Supporter_CareerStage_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/Officially_CS.eps", width=1000, height=578)
ggplot(graphdata, aes(x=Supporter_CareerStage_clean, y=perc, fill=Supporter_Officially,)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=12), axis.title.x = element_text(size = 8), axis.title.y = element_text(size = 16))+
geom_text(aes(label=round(perc, digit=1)), size=3.5, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="It was part of my official job description \nto be in charge of the person I helped") +
guides(fill=guide_legend(reverse=TRUE)) +
scale_x_discrete(limits = rev(levels(graphdata$Supporter_CareerStage_clean))) +
coord_flip() +
scale_fill_manual(name="", values=c(eBlue, eGreen))
ggsave(dpi=1000, "3-1-1.png", limitsize = FALSE)## Saving 7 x 5 in imagedev.off()## null device
## 1Q06 MH at the time of support (Section 2, Panel 3)
Response (label: “Supporter_MH”): No-1 , Yes-2 (Not sure-3, and PNTA-4 are removed for binomial modelling)
Model: Factorial logistical regression Supporter_MH ~ Supporter_CareerStage * All_Gender_clean)
# Subsetting data and recoding response to binary
MH=genderCS[,c("Supporter_MH", "Supporter_CareerStage_clean", "All_Gender_clean")]
MH=MH[MH$Supporter_MH %in% seq(1,2), ] #remove NAs and some choices
dim(MH)[1] #this gives N## [1] 1125MH[MH$Supporter_MH==1, "Supporter_MH"]=0 #recoding No->0 and Yes->1 for binom modelling
MH[MH$Supporter_MH==2, "Supporter_MH"]=1
MH_glm=glm(Supporter_MH~Supporter_CareerStage_clean*All_Gender_clean, data=MH, family="binomial")
summary(MH_glm)##
## Call:
## glm(formula = Supporter_MH ~ Supporter_CareerStage_clean * All_Gender_clean,
## family = "binomial", data = MH)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -2.0022 -0.9647 0.5380 0.8109 1.7866
##
## Coefficients:
## Estimate
## (Intercept) 1.2256
## Supporter_CareerStage_cleanPostdocs -0.6125
## Supporter_CareerStage_cleanGroup leaders (<5yr) -1.6029
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -1.7489
## Supporter_CareerStage_cleanGroup leaders (>10yr) -2.5951
## All_Gender_cleanWomen 0.6341
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -0.3038
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.1542
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -0.5809
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen -0.0756
## Std. Error
## (Intercept) 0.2011
## Supporter_CareerStage_cleanPostdocs 0.2828
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.3327
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.3741
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.3522
## All_Gender_cleanWomen 0.2571
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.3600
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.4333
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.5234
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.5197
## z value
## (Intercept) 6.096
## Supporter_CareerStage_cleanPostdocs -2.166
## Supporter_CareerStage_cleanGroup leaders (<5yr) -4.818
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -4.675
## Supporter_CareerStage_cleanGroup leaders (>10yr) -7.368
## All_Gender_cleanWomen 2.466
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -0.844
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.356
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -1.110
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen -0.145
## Pr(>|z|)
## (Intercept) 1.09e-09
## Supporter_CareerStage_cleanPostdocs 0.0303
## Supporter_CareerStage_cleanGroup leaders (<5yr) 1.45e-06
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 2.94e-06
## Supporter_CareerStage_cleanGroup leaders (>10yr) 1.73e-13
## All_Gender_cleanWomen 0.0136
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.3988
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.7219
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.2671
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.8843
##
## (Intercept) ***
## Supporter_CareerStage_cleanPostdocs *
## Supporter_CareerStage_cleanGroup leaders (<5yr) ***
## Supporter_CareerStage_cleanGroup leaders (5-10yr) ***
## Supporter_CareerStage_cleanGroup leaders (>10yr) ***
## All_Gender_cleanWomen *
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1443.0 on 1124 degrees of freedom
## Residual deviance: 1223.7 on 1115 degrees of freedom
## AIC: 1243.7
##
## Number of Fisher Scoring iterations: 4# LR gender
summary(glm(Supporter_MH~All_Gender_clean, data=MH, family="binomial")) #glm and p-values##
## Call:
## glm(formula = Supporter_MH ~ All_Gender_clean, family = "binomial",
## data = MH)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.6090 -1.2662 0.8003 0.8003 1.0911
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.20634 0.09719 2.123 0.0337 *
## All_Gender_cleanWomen 0.76793 0.12906 5.950 2.68e-09 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1443.0 on 1124 degrees of freedom
## Residual deviance: 1407.4 on 1123 degrees of freedom
## AIC: 1411.4
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Supporter_MH~All_Gender_clean, data=MH, family="binomial"))) #OR (men as baseline)## (Intercept) All_Gender_cleanWomen
## 1.229167 2.155293exp(confint(glm(Supporter_MH~All_Gender_clean, data=MH, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 1.016462 1.488235
## All_Gender_cleanWomen 1.674328 2.777481# LR CS
summary(glm(Supporter_MH~Supporter_CareerStage_clean, data=MH, family="binomial")) #glm and p-values##
## Call:
## glm(formula = Supporter_MH ~ Supporter_CareerStage_clean, family = "binomial",
## data = MH)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.9076 -0.9746 0.5948 0.8531 1.6921
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) 1.6426 0.1245 13.194
## Supporter_CareerStage_cleanPostdocs -0.8192 0.1742 -4.704
## Supporter_CareerStage_cleanGroup leaders (<5yr) -1.7449 0.2116 -8.246
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -2.1405 0.2595 -8.247
## Supporter_CareerStage_cleanGroup leaders (>10yr) -2.8012 0.2533 -11.058
## Pr(>|z|)
## (Intercept) < 2e-16 ***
## Supporter_CareerStage_cleanPostdocs 2.56e-06 ***
## Supporter_CareerStage_cleanGroup leaders (<5yr) < 2e-16 ***
## Supporter_CareerStage_cleanGroup leaders (5-10yr) < 2e-16 ***
## Supporter_CareerStage_cleanGroup leaders (>10yr) < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1443.0 on 1124 degrees of freedom
## Residual deviance: 1234.7 on 1120 degrees of freedom
## AIC: 1244.7
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Supporter_MH~Supporter_CareerStage_clean, data=MH, family="binomial"))) #OR (PhDs as baseline)## (Intercept)
## 5.16883117
## Supporter_CareerStage_cleanPostdocs
## 0.44078641
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 0.17465801
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 0.11759779
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 0.06073975exp(confint(glm(Supporter_MH~Supporter_CareerStage_clean, data=MH, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 4.07557055 6.64410256
## Supporter_CareerStage_cleanPostdocs 0.31268313 0.61936110
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.11496592 0.26378398
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.07006469 0.19431557
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.03640658 0.09852935MH at the time of support x gender graph (graph 2.3.3)
# This will save the image to your local code folder
#eps("images/SupporterMH_gender.eps", width=1000, height=578)
MH[MH$Supporter_MH==0, "Supporter_MH"]="No"
MH[MH$Supporter_MH==1, "Supporter_MH"]="Yes"
MH$Supporter_MH=factor(MH$Supporter_MH)
graphdata = MH %>%
group_by(All_Gender_clean, Supporter_MH) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'All_Gender_clean' (override with `.groups` argument)ggplot(graphdata, aes(x=All_Gender_clean, y=perc, fill=Supporter_MH)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=12), axis.title.x = element_text(size = 8), axis.title.y = element_text(size = 16))+
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="As I was providing support, I was struggling with my own mental health:") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=c(eBlue, eGreen))
ggsave(dpi=1000, "2-3-3-v2.png")## Saving 7 x 5 in imagedev.off()## null device
## 1MH at the time of support x CS graph (Graph 2.3.5)
# This will save the image to your local code folder
#eps("images/SupporterMH_CS.eps", width=1000, height=578)
graphdata = MH %>%
group_by(Supporter_CareerStage_clean, Supporter_MH) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'Supporter_CareerStage_clean' (override with `.groups` argument)ggplot(graphdata, aes(x=Supporter_CareerStage_clean, y=perc, fill=Supporter_MH)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=12), axis.title.x = element_text(size = 8), axis.title.y = element_text(size = 16))+
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="As I was providing support, I was struggling with my own mental health") +
guides(fill=guide_legend(reverse=TRUE)) +
scale_x_discrete(limits = rev(levels(graphdata$Supporter_CareerStage_clean))) +
coord_flip() +
scale_fill_manual(name="", values=c(eBlue, eGreen))
ggsave(dpi=1000, "2-3-5-v1.png")## Saving 7 x 5 in imagedev.off()## null device
## 1Q07 Helping more than one person at the time (Section 2, Panel 3)
Dependent variable: Supporter_NumberReceivers
- 1: No (just one person)
- 2: Yes (more than one person)
Model: First individual factors (Gender, CS), then Factorial logistical regression Supporter_NumberReceivers ~ Supporter_CareerStage * All_Gender_clean) to check for interactions
# Subsetting data and recoding response to binary
NumberReceivers=genderCS[,c("Supporter_NumberReceivers", "Supporter_CareerStage_clean", "All_Gender_clean")]
NumberReceivers=NumberReceivers[NumberReceivers$Supporter_NumberReceivers %in% seq(1,2), ] #remove NAs and some choices
dim(NumberReceivers)[1] #N## [1] 1183NumberReceivers[NumberReceivers$Supporter_NumberReceivers==1, "Supporter_NumberReceivers"]=0 #recoding No->0 and Yes->1 for binom modelling
NumberReceivers[NumberReceivers$Supporter_NumberReceivers==2, "Supporter_NumberReceivers"]=1
#LR gender
summary(glm(Supporter_NumberReceivers~All_Gender_clean, data=NumberReceivers, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Supporter_NumberReceivers ~ All_Gender_clean, family = "binomial",
## data = NumberReceivers)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.6950 -1.3821 0.7369 0.7369 0.9857
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.46928 0.09721 4.827 1.38e-06 ***
## All_Gender_cleanWomen 0.69566 0.13018 5.344 9.10e-08 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1431.7 on 1182 degrees of freedom
## Residual deviance: 1403.1 on 1181 degrees of freedom
## AIC: 1407.1
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Supporter_NumberReceivers~All_Gender_clean, data=NumberReceivers, family="binomial"))) #OR (men as baseline)## (Intercept) All_Gender_cleanWomen
## 1.598837 2.005029exp(confint(glm(Supporter_NumberReceivers~All_Gender_clean, data=NumberReceivers, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 1.323121 1.937416
## All_Gender_cleanWomen 1.553825 2.588992#LR CS
summary(glm(Supporter_NumberReceivers~Supporter_CareerStage_clean, data=NumberReceivers, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Supporter_NumberReceivers ~ Supporter_CareerStage_clean,
## family = "binomial", data = NumberReceivers)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.6712 -1.3349 0.7541 0.7952 1.0525
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) 1.1121 0.1039 10.699
## Supporter_CareerStage_cleanPostdocs -0.1229 0.1614 -0.762
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.3888 0.2028 -1.917
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -0.8110 0.2405 -3.372
## Supporter_CareerStage_cleanGroup leaders (>10yr) -0.7492 0.2148 -3.488
## Pr(>|z|)
## (Intercept) < 2e-16 ***
## Supporter_CareerStage_cleanPostdocs 0.446196
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.055192 .
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.000746 ***
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.000487 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1431.7 on 1182 degrees of freedom
## Residual deviance: 1410.8 on 1178 degrees of freedom
## AIC: 1420.8
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Supporter_NumberReceivers~Supporter_CareerStage_clean, data=NumberReceivers, family="binomial"))) #OR (PhD as baseline)## (Intercept)
## 3.0406504
## Supporter_CareerStage_cleanPostdocs
## 0.8843137
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 0.6778893
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 0.4444284
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 0.4727607exp(confint(glm(Supporter_NumberReceivers~Supporter_CareerStage_clean, data=NumberReceivers, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 2.4884495 3.7415024
## Supporter_CareerStage_cleanPostdocs 0.6450783 1.2151196
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.4571324 1.0134795
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.2778975 0.7150942
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.3108053 0.7224488#FLR gender-CS interactions
summary(glm(Supporter_NumberReceivers~Supporter_CareerStage_clean*All_Gender_clean, data=NumberReceivers, family="binomial"))##
## Call:
## glm(formula = Supporter_NumberReceivers ~ Supporter_CareerStage_clean *
## All_Gender_clean, family = "binomial", data = NumberReceivers)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.7570 -1.2346 0.7148 0.8412 1.1330
##
## Coefficients:
## Estimate
## (Intercept) 0.85678
## Supporter_CareerStage_cleanPostdocs -0.38387
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.56909
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -0.72325
## Supporter_CareerStage_cleanGroup leaders (>10yr) -0.75142
## All_Gender_cleanWomen 0.37750
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.45274
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.41816
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -0.02553
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.39953
## Std. Error
## (Intercept) 0.17792
## Supporter_CareerStage_cleanPostdocs 0.26345
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.31060
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.34777
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.29058
## All_Gender_cleanWomen 0.21956
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.33617
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.41697
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.48832
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.46776
## z value
## (Intercept) 4.815
## Supporter_CareerStage_cleanPostdocs -1.457
## Supporter_CareerStage_cleanGroup leaders (<5yr) -1.832
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -2.080
## Supporter_CareerStage_cleanGroup leaders (>10yr) -2.586
## All_Gender_cleanWomen 1.719
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 1.347
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 1.003
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -0.052
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.854
## Pr(>|z|)
## (Intercept) 1.47e-06
## Supporter_CareerStage_cleanPostdocs 0.14509
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.06691
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.03755
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.00971
## All_Gender_cleanWomen 0.08555
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.17807
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.31593
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.95831
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.39304
##
## (Intercept) ***
## Supporter_CareerStage_cleanPostdocs
## Supporter_CareerStage_cleanGroup leaders (<5yr) .
## Supporter_CareerStage_cleanGroup leaders (5-10yr) *
## Supporter_CareerStage_cleanGroup leaders (>10yr) **
## All_Gender_cleanWomen .
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1431.7 on 1182 degrees of freedom
## Residual deviance: 1387.9 on 1173 degrees of freedom
## AIC: 1407.9
##
## Number of Fisher Scoring iterations: 4graph for Helping more than one person at the time x gender (Graph 2.3.4)
NumberReceivers[NumberReceivers$Supporter_NumberReceivers==0, "Supporter_NumberReceivers"]="No"
NumberReceivers[NumberReceivers$Supporter_NumberReceivers==1, "Supporter_NumberReceivers"]="Yes"
NumberReceivers$Supporter_NumberReceivers=factor(NumberReceivers$Supporter_NumberReceivers)
#ImpactPersonal$Impact_Personal=factor(ImpactPersonal$Impact_Personal, levels=c("No","Yes"))
# This will save the image to your local code folder
#eps("images/NumberReceivers_gender.eps", width=1000, height=578)
graphdata = NumberReceivers %>%
group_by(All_Gender_clean, Supporter_NumberReceivers) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'All_Gender_clean' (override with `.groups` argument)ggplot(graphdata, aes(x=All_Gender_clean, y=perc, fill=Supporter_NumberReceivers)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=12), axis.title.x = element_text(size = 8), axis.title.y = element_text(size = 16))+
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="I was helping more than one person \nat the same time (in my professional and personal life)") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=c(eBlue, eGreen))
ggsave(dpi=1000, "2-3-4.png", limitsize=FALSE)## Saving 7 x 5 in imagedev.off()## null device
## 1graph for Helping more than one person at the time x CS (Graph 2.3.6)
graphdata = NumberReceivers %>%
group_by(Supporter_CareerStage_clean, Supporter_NumberReceivers) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'Supporter_CareerStage_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/NumberReceivers_CS.eps", width=1000, height=578)
ggplot(graphdata, aes(x=Supporter_CareerStage_clean, y=perc, fill=Supporter_NumberReceivers)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=12), axis.title.x = element_text(size = 8), axis.title.y = element_text(size = 16))+
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="I was helping more than one person at the same time \n(in my professional and personal life)") +
guides(fill=guide_legend(reverse=TRUE)) +
scale_x_discrete(limits = rev(levels(graphdata$Supporter_CareerStage_clean))) +
coord_flip() +
scale_fill_manual(name="", values=c(eBlue, eGreen))
ggsave(dpi=1000, "2-3-6.png", limitsize = FALSE)## Saving 7 x 5 in imagedev.off()## null device
## 1Q13 Missing resources for person supported (Section 3, Panel 4)
Filter down to people who picked at least 1 of the 6 options:
- Receiver_Missing_Personal
- Receiver_Missing_Institutions
- Receiver_Missing_Peers
- Receiver_Missing_Managers
- Receiver_Missing_Pro
- Receiver_Missing_OutPro
#subsetting data and reocoding variables
Receiver_Missing=genderCS[(genderCS$Receiver_Missing_Personal==1 | genderCS$Receiver_Missing_Institutions==1 | genderCS$Receiver_Missing_Peers==1 | genderCS$ Receiver_Missing_Managers==1 | genderCS$Receiver_Missing_Pro==1 | genderCS$Receiver_Missing_OutPro==1), ]
Receiver_Missing=genderCS[,c("Receiver_Missing_Personal", "Receiver_Missing_Institutions", "Receiver_Missing_Peers", "Receiver_Missing_Managers", "Receiver_Missing_Pro", "Receiver_Missing_OutPro", "Supporter_CareerStage_clean","All_Gender_clean")]
dim(Receiver_Missing)[1]## [1] 1255#fill all na with 0
Receiver_Missing[is.na(Receiver_Missing)]=0Receiver_Missing_Personal (Yes/No) xCS xGender xCSxGender
Dependent variable -
- 0 - didn’t select this option
- 1 - did select this option - i.e. Missing personal support
Model: Factorial logistical regression
summary(glm(Receiver_Missing_Personal~All_Gender_clean, data=Receiver_Missing, family="binomial")) #LR gender##
## Call:
## glm(formula = Receiver_Missing_Personal ~ All_Gender_clean, family = "binomial",
## data = Receiver_Missing)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.7077 -0.7077 -0.6913 -0.6913 1.7599
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.2568 0.1106 -11.363 <2e-16 ***
## All_Gender_cleanWomen -0.0528 0.1410 -0.374 0.708
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1309.5 on 1254 degrees of freedom
## Residual deviance: 1309.4 on 1253 degrees of freedom
## AIC: 1313.4
##
## Number of Fisher Scoring iterations: 4summary(glm(Receiver_Missing_Personal~Supporter_CareerStage_clean, data=Receiver_Missing, family="binomial")) #LR CS##
## Call:
## glm(formula = Receiver_Missing_Personal ~ Supporter_CareerStage_clean,
## family = "binomial", data = Receiver_Missing)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.8061 -0.6999 -0.6999 -0.6186 1.8697
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) -1.281891 0.104943 -12.215
## Supporter_CareerStage_cleanPostdocs -0.274554 0.174767 -1.571
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.324592 0.207840 1.562
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.255252 0.260010 0.982
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.007388 0.245412 0.030
## Pr(>|z|)
## (Intercept) <2e-16 ***
## Supporter_CareerStage_cleanPostdocs 0.116
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.118
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.326
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.976
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1309.5 on 1254 degrees of freedom
## Residual deviance: 1301.2 on 1250 degrees of freedom
## AIC: 1311.2
##
## Number of Fisher Scoring iterations: 4summary(glm(Receiver_Missing_Personal~Supporter_CareerStage_clean*All_Gender_clean, data=Receiver_Missing, family="binomial")) #LR gender-CS interactions##
## Call:
## glm(formula = Receiver_Missing_Personal ~ Supporter_CareerStage_clean *
## All_Gender_clean, family = "binomial", data = Receiver_Missing)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.8880 -0.7040 -0.6938 -0.5706 1.9471
##
## Coefficients:
## Estimate
## (Intercept) -1.23676
## Supporter_CareerStage_cleanPostdocs -0.03175
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.07542
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.13815
## Supporter_CareerStage_cleanGroup leaders (>10yr) -0.10152
## All_Gender_cleanWomen -0.06479
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -0.39962
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.64993
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.21432
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.23992
## Std. Error
## (Intercept) 0.18932
## Supporter_CareerStage_cleanPostdocs 0.28861
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.35567
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.38334
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.33873
## All_Gender_cleanWomen 0.22747
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.36444
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.43995
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.52768
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.51175
## z value
## (Intercept) -6.533
## Supporter_CareerStage_cleanPostdocs -0.110
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.212
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.360
## Supporter_CareerStage_cleanGroup leaders (>10yr) -0.300
## All_Gender_cleanWomen -0.285
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -1.097
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 1.477
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.406
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.469
## Pr(>|z|)
## (Intercept) 6.46e-11
## Supporter_CareerStage_cleanPostdocs 0.912
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.832
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.719
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.764
## All_Gender_cleanWomen 0.776
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.273
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.140
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.685
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.639
##
## (Intercept) ***
## Supporter_CareerStage_cleanPostdocs
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## All_Gender_cleanWomen
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1309.5 on 1254 degrees of freedom
## Residual deviance: 1295.8 on 1245 degrees of freedom
## AIC: 1315.8
##
## Number of Fisher Scoring iterations: 4Receiver_Missing_Managers xCS xGender xCSxGender
Missing: Support from supervisors and managers as above
#LR gender
summary(glm(Receiver_Missing_Managers~All_Gender_clean, data=Receiver_Missing, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Receiver_Missing_Managers ~ All_Gender_clean, family = "binomial",
## data = Receiver_Missing)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.3857 -1.1810 0.9825 0.9825 1.1738
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.008439 0.091864 0.092 0.927
## All_Gender_cleanWomen 0.469062 0.117721 3.985 6.76e-05 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1712.4 on 1254 degrees of freedom
## Residual deviance: 1696.5 on 1253 degrees of freedom
## AIC: 1700.5
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Receiver_Missing_Managers~All_Gender_clean, data=Receiver_Missing, family="binomial"))) #OR (men as baseline)## (Intercept) All_Gender_cleanWomen
## 1.008475 1.598494exp(confint(glm(Receiver_Missing_Managers~All_Gender_clean, data=Receiver_Missing, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.8422231 1.207599
## All_Gender_cleanWomen 1.2694232 2.014094#LR CS
summary(glm(Receiver_Missing_Managers~Supporter_CareerStage_clean, data=Receiver_Missing, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Receiver_Missing_Managers ~ Supporter_CareerStage_clean,
## family = "binomial", data = Receiver_Missing)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.6233 -0.7386 0.7895 0.8313 1.8317
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) 1.00586 0.09773 10.293
## Supporter_CareerStage_cleanPostdocs -0.12082 0.15210 -0.794
## Supporter_CareerStage_cleanGroup leaders (<5yr) -2.16563 0.21225 -10.203
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -2.47671 0.28616 -8.655
## Supporter_CareerStage_cleanGroup leaders (>10yr) -2.43472 0.25188 -9.666
## Pr(>|z|)
## (Intercept) <2e-16 ***
## Supporter_CareerStage_cleanPostdocs 0.427
## Supporter_CareerStage_cleanGroup leaders (<5yr) <2e-16 ***
## Supporter_CareerStage_cleanGroup leaders (5-10yr) <2e-16 ***
## Supporter_CareerStage_cleanGroup leaders (>10yr) <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1712.4 on 1254 degrees of freedom
## Residual deviance: 1425.5 on 1250 degrees of freedom
## AIC: 1435.5
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Receiver_Missing_Managers~Supporter_CareerStage_clean, data=Receiver_Missing, family="binomial"))) #OR (PhD as baseline)## (Intercept)
## 2.73426573
## Supporter_CareerStage_cleanPostdocs
## 0.88618926
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 0.11467771
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 0.08401880
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 0.08762255exp(confint(glm(Receiver_Missing_Managers~Supporter_CareerStage_clean, data=Receiver_Missing, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 2.26372323 3.3214728
## Supporter_CareerStage_cleanPostdocs 0.65822609 1.1953860
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.07482926 0.1722600
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.04656520 0.1438613
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.05239933 0.1411885#FLR gender x CS interactions
summary(glm(Receiver_Missing_Managers~Supporter_CareerStage_clean*All_Gender_clean, data=Receiver_Missing, family="binomial"))##
## Call:
## glm(formula = Receiver_Missing_Managers ~ Supporter_CareerStage_clean *
## All_Gender_clean, family = "binomial", data = Receiver_Missing)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.6429 -0.7585 0.7749 0.8161 2.0745
##
## Coefficients:
## Estimate
## (Intercept) 0.90756
## Supporter_CareerStage_cleanPostdocs -0.10249
## Supporter_CareerStage_cleanGroup leaders (<5yr) -2.21974
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -2.00617
## Supporter_CareerStage_cleanGroup leaders (>10yr) -2.50149
## All_Gender_cleanWomen 0.14175
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -0.01848
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.11628
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -1.07129
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.28903
## Std. Error
## (Intercept) 0.17467
## Supporter_CareerStage_cleanPostdocs 0.26190
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.34810
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.37633
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.35080
## All_Gender_cleanWomen 0.21079
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.32195
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.44014
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.61794
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.51791
## z value
## (Intercept) 5.196
## Supporter_CareerStage_cleanPostdocs -0.391
## Supporter_CareerStage_cleanGroup leaders (<5yr) -6.377
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -5.331
## Supporter_CareerStage_cleanGroup leaders (>10yr) -7.131
## All_Gender_cleanWomen 0.672
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -0.057
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.264
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -1.734
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.558
## Pr(>|z|)
## (Intercept) 2.04e-07
## Supporter_CareerStage_cleanPostdocs 0.696
## Supporter_CareerStage_cleanGroup leaders (<5yr) 1.81e-10
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 9.77e-08
## Supporter_CareerStage_cleanGroup leaders (>10yr) 9.97e-13
## All_Gender_cleanWomen 0.501
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.954
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.792
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.083
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.577
##
## (Intercept) ***
## Supporter_CareerStage_cleanPostdocs
## Supporter_CareerStage_cleanGroup leaders (<5yr) ***
## Supporter_CareerStage_cleanGroup leaders (5-10yr) ***
## Supporter_CareerStage_cleanGroup leaders (>10yr) ***
## All_Gender_cleanWomen
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen .
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1712.4 on 1254 degrees of freedom
## Residual deviance: 1420.8 on 1245 degrees of freedom
## AIC: 1440.8
##
## Number of Fisher Scoring iterations: 4Receiver_Missing_Peers xCS xGender xCSxGender
Missing: Support from peers and colleagues as above
summary(glm(Receiver_Missing_Peers~All_Gender_clean, data=Receiver_Missing, family="binomial")) #LR Gender##
## Call:
## glm(formula = Receiver_Missing_Peers ~ All_Gender_clean, family = "binomial",
## data = Receiver_Missing)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.8152 -0.8152 -0.8091 1.5896 1.5976
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.93111 0.10200 -9.129 <2e-16 ***
## All_Gender_cleanWomen -0.01768 0.12949 -0.137 0.891
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1489.4 on 1254 degrees of freedom
## Residual deviance: 1489.4 on 1253 degrees of freedom
## AIC: 1493.4
##
## Number of Fisher Scoring iterations: 4summary(glm(Receiver_Missing_Peers~Supporter_CareerStage_clean, data=Receiver_Missing, family="binomial")) #LR CS##
## Call:
## glm(formula = Receiver_Missing_Peers ~ Supporter_CareerStage_clean,
## family = "binomial", data = Receiver_Missing)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.8739 -0.8692 -0.7633 1.5150 1.8695
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) -0.77867 0.09319 -8.356
## Supporter_CareerStage_cleanPostdocs 0.01300 0.14714 0.088
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.77752 0.23137 -3.360
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -0.30534 0.25859 -1.181
## Supporter_CareerStage_cleanGroup leaders (>10yr) -0.76178 0.25789 -2.954
## Pr(>|z|)
## (Intercept) < 2e-16 ***
## Supporter_CareerStage_cleanPostdocs 0.929619
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.000778 ***
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.237680
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.003138 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1489.4 on 1254 degrees of freedom
## Residual deviance: 1467.2 on 1250 degrees of freedom
## AIC: 1477.2
##
## Number of Fisher Scoring iterations: 4summary(glm(Receiver_Missing_Peers~Supporter_CareerStage_clean*All_Gender_clean, data=Receiver_Missing, family="binomial")) #LR gender-CS interactions##
## Call:
## glm(formula = Receiver_Missing_Peers ~ Supporter_CareerStage_clean *
## All_Gender_clean, family = "binomial", data = Receiver_Missing)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.9489 -0.8381 -0.7687 1.4246 2.0544
##
## Coefficients:
## Estimate
## (Intercept) -0.56453
## Supporter_CareerStage_cleanPostdocs -0.02073
## Supporter_CareerStage_cleanGroup leaders (<5yr) -1.41647
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -0.53408
## Supporter_CareerStage_cleanGroup leaders (>10yr) -1.12495
## All_Gender_cleanWomen -0.31094
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.03044
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.98788
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.34171
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.70114
## Std. Error
## (Intercept) 0.16445
## Supporter_CareerStage_cleanPostdocs 0.24986
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.41144
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.37169
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.35463
## All_Gender_cleanWomen 0.19981
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.30968
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.49908
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.52268
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.52920
## z value
## (Intercept) -3.433
## Supporter_CareerStage_cleanPostdocs -0.083
## Supporter_CareerStage_cleanGroup leaders (<5yr) -3.443
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -1.437
## Supporter_CareerStage_cleanGroup leaders (>10yr) -3.172
## All_Gender_cleanWomen -1.556
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.098
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 1.979
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.654
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 1.325
## Pr(>|z|)
## (Intercept) 0.000598
## Supporter_CareerStage_cleanPostdocs 0.933884
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.000576
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.150750
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.001513
## All_Gender_cleanWomen 0.119667
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.921704
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.047770
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.513262
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.185207
##
## (Intercept) ***
## Supporter_CareerStage_cleanPostdocs
## Supporter_CareerStage_cleanGroup leaders (<5yr) ***
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## Supporter_CareerStage_cleanGroup leaders (>10yr) **
## All_Gender_cleanWomen
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen *
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1489.4 on 1254 degrees of freedom
## Residual deviance: 1460.4 on 1245 degrees of freedom
## AIC: 1480.4
##
## Number of Fisher Scoring iterations: 4Receiver_Missing_Institutions xCS xGender xCSxGender
Missing Institutional measures (e.g. extensions, intermission) as above
#LR gender
summary(glm(Receiver_Missing_Institutions~All_Gender_clean, data=Receiver_Missing, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Receiver_Missing_Institutions ~ All_Gender_clean,
## family = "binomial", data = Receiver_Missing)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.9656 -0.9656 -0.9005 1.4052 1.4823
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.69315 0.09744 -7.114 1.13e-12 ***
## All_Gender_cleanWomen 0.17207 0.12236 1.406 0.16
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1636.8 on 1254 degrees of freedom
## Residual deviance: 1634.8 on 1253 degrees of freedom
## AIC: 1638.8
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Receiver_Missing_Institutions~All_Gender_clean, data=Receiver_Missing, family="binomial"))) #OR (men as baseline)## (Intercept) All_Gender_cleanWomen
## 0.500000 1.187755exp(confint(glm(Receiver_Missing_Institutions~All_Gender_clean, data=Receiver_Missing, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.4121782 0.6040748
## All_Gender_cleanWomen 0.9352946 1.5112643#LR CS
summary(glm(Receiver_Missing_Institutions~Supporter_CareerStage_clean, data=Receiver_Missing, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Receiver_Missing_Institutions ~ Supporter_CareerStage_clean,
## family = "binomial", data = Receiver_Missing)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.0708 -0.9255 -0.9114 1.2879 1.9728
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) -0.25607 0.08726 -2.935
## Supporter_CareerStage_cleanPostdocs -0.40772 0.14189 -2.873
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.37006 0.18983 -1.950
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -0.77057 0.25339 -3.041
## Supporter_CareerStage_cleanGroup leaders (>10yr) -1.53569 0.27612 -5.562
## Pr(>|z|)
## (Intercept) 0.00334 **
## Supporter_CareerStage_cleanPostdocs 0.00406 **
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.05124 .
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.00236 **
## Supporter_CareerStage_cleanGroup leaders (>10yr) 2.67e-08 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1636.8 on 1254 degrees of freedom
## Residual deviance: 1591.0 on 1250 degrees of freedom
## AIC: 1601
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Receiver_Missing_Institutions~Supporter_CareerStage_clean, data=Receiver_Missing, family="binomial"))) #OR (PhD as baseline)## (Intercept)
## 0.7740864
## Supporter_CareerStage_cleanPostdocs
## 0.6651630
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 0.6906897
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 0.4627506
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 0.2153076exp(confint(glm(Receiver_Missing_Institutions~Supporter_CareerStage_clean, data=Receiver_Missing, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.6519265 0.9179939
## Supporter_CareerStage_cleanPostdocs 0.5029068 0.8773680
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.4738088 0.9983312
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.2768263 0.7504722
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.1214232 0.3606857#FLR gender-CS interactions
summary(glm(Receiver_Missing_Institutions~Supporter_CareerStage_clean*All_Gender_clean, data=Receiver_Missing, family="binomial"))##
## Call:
## glm(formula = Receiver_Missing_Institutions ~ Supporter_CareerStage_clean *
## All_Gender_clean, family = "binomial", data = Receiver_Missing)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.0831 -0.9508 -0.8958 1.2936 2.0720
##
## Coefficients:
## Estimate
## (Intercept) -0.22596
## Supporter_CareerStage_cleanPostdocs -0.35930
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.33366
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -1.10904
## Supporter_CareerStage_cleanGroup leaders (>10yr) -1.79633
## All_Gender_cleanWomen -0.04303
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -0.07776
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.07369
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.64980
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.61840
## Std. Error
## (Intercept) 0.15912
## Supporter_CareerStage_cleanPostdocs 0.24639
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.30132
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.38940
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.38876
## All_Gender_cleanWomen 0.19029
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.30165
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.38983
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.51811
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.56253
## z value
## (Intercept) -1.420
## Supporter_CareerStage_cleanPostdocs -1.458
## Supporter_CareerStage_cleanGroup leaders (<5yr) -1.107
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -2.848
## Supporter_CareerStage_cleanGroup leaders (>10yr) -4.621
## All_Gender_cleanWomen -0.226
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -0.258
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.189
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 1.254
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 1.099
## Pr(>|z|)
## (Intercept) 0.1556
## Supporter_CareerStage_cleanPostdocs 0.1448
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.2682
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.0044
## Supporter_CareerStage_cleanGroup leaders (>10yr) 3.83e-06
## All_Gender_cleanWomen 0.8211
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.7966
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.8501
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.2098
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.2716
##
## (Intercept)
## Supporter_CareerStage_cleanPostdocs
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## Supporter_CareerStage_cleanGroup leaders (5-10yr) **
## Supporter_CareerStage_cleanGroup leaders (>10yr) ***
## All_Gender_cleanWomen
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1636.8 on 1254 degrees of freedom
## Residual deviance: 1587.8 on 1245 degrees of freedom
## AIC: 1607.8
##
## Number of Fisher Scoring iterations: 4Receiver_Missing_Pro xCS xGender xCSxGender
Missing Support from mental health professionals within the institution as above
summary(glm(Receiver_Missing_Pro~All_Gender_clean, data=Receiver_Missing, family="binomial")) #LR Gender##
## Call:
## glm(formula = Receiver_Missing_Pro ~ All_Gender_clean, family = "binomial",
## data = Receiver_Missing)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.0458 -1.0458 -0.9774 1.3149 1.3916
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.49062 0.09464 -5.184 2.17e-07 ***
## All_Gender_cleanWomen 0.17300 0.11920 1.451 0.147
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1694.8 on 1254 degrees of freedom
## Residual deviance: 1692.7 on 1253 degrees of freedom
## AIC: 1696.7
##
## Number of Fisher Scoring iterations: 4summary(glm(Receiver_Missing_Pro~Supporter_CareerStage_clean, data=Receiver_Missing, family="binomial")) #LR CS##
## Call:
## glm(formula = Receiver_Missing_Pro ~ Supporter_CareerStage_clean,
## family = "binomial", data = Receiver_Missing)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.0761 -1.0677 -0.9719 1.2913 1.5285
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) -0.26369 0.08730 -3.020
## Supporter_CareerStage_cleanPostdocs -0.24115 0.13996 -1.723
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.03529 0.18442 -0.191
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.02074 0.22854 0.091
## Supporter_CareerStage_cleanGroup leaders (>10yr) -0.53211 0.21643 -2.459
## Pr(>|z|)
## (Intercept) 0.00252 **
## Supporter_CareerStage_cleanPostdocs 0.08489 .
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.84823
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.92768
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.01395 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1694.8 on 1254 degrees of freedom
## Residual deviance: 1686.4 on 1250 degrees of freedom
## AIC: 1696.4
##
## Number of Fisher Scoring iterations: 4summary(glm(Receiver_Missing_Pro~Supporter_CareerStage_clean*All_Gender_clean, data=Receiver_Missing, family="binomial")) #FLR gender-CS interactions##
## Call:
## glm(formula = Receiver_Missing_Pro ~ Supporter_CareerStage_clean *
## All_Gender_clean, family = "binomial", data = Receiver_Missing)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.1073 -1.0520 -0.8733 1.2840 1.5829
##
## Coefficients:
## Estimate
## (Intercept) -0.30228
## Supporter_CareerStage_cleanPostdocs -0.46497
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.05866
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.13523
## Supporter_CareerStage_cleanGroup leaders (>10yr) -0.61401
## All_Gender_cleanWomen 0.05504
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.33884
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.15174
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -0.21649
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.27346
## Std. Error
## (Intercept) 0.15992
## Supporter_CareerStage_cleanPostdocs 0.25124
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.29510
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.33090
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.29868
## All_Gender_cleanWomen 0.19088
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.30291
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.37974
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.46466
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.45141
## z value
## (Intercept) -1.890
## Supporter_CareerStage_cleanPostdocs -1.851
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.199
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.409
## Supporter_CareerStage_cleanGroup leaders (>10yr) -2.056
## All_Gender_cleanWomen 0.288
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 1.119
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.400
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -0.466
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.606
## Pr(>|z|)
## (Intercept) 0.0587
## Supporter_CareerStage_cleanPostdocs 0.0642
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.8424
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.6828
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.0398
## All_Gender_cleanWomen 0.7731
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.2633
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.6895
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.6413
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.5447
##
## (Intercept) .
## Supporter_CareerStage_cleanPostdocs .
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## Supporter_CareerStage_cleanGroup leaders (>10yr) *
## All_Gender_cleanWomen
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1694.8 on 1254 degrees of freedom
## Residual deviance: 1682.6 on 1245 degrees of freedom
## AIC: 1702.6
##
## Number of Fisher Scoring iterations: 4Receiver_Missing_OutPro xCS xGender xCSxGender
Missing Support from health professionals outside of the institution as above
summary(glm(Receiver_Missing_OutPro~All_Gender_clean, data=Receiver_Missing, family="binomial")) #LR gender##
## Call:
## glm(formula = Receiver_Missing_OutPro ~ All_Gender_clean, family = "binomial",
## data = Receiver_Missing)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.6795 -0.6795 -0.6080 -0.6080 1.8864
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.5943 0.1226 -13.001 <2e-16 ***
## All_Gender_cleanWomen 0.2460 0.1512 1.627 0.104
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1227.8 on 1254 degrees of freedom
## Residual deviance: 1225.1 on 1253 degrees of freedom
## AIC: 1229.1
##
## Number of Fisher Scoring iterations: 4summary(glm(Receiver_Missing_OutPro~Supporter_CareerStage_clean, data=Receiver_Missing, family="binomial")) #LR CS##
## Call:
## glm(formula = Receiver_Missing_OutPro ~ Supporter_CareerStage_clean,
## family = "binomial", data = Receiver_Missing)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.7825 -0.6404 -0.6404 -0.6131 1.8783
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) -1.48023 0.11135 -13.293
## Supporter_CareerStage_cleanPostdocs -0.09588 0.17940 -0.534
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.21080 0.22378 0.942
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.45359 0.26266 1.727
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.05137 0.25748 0.200
## Pr(>|z|)
## (Intercept) <2e-16 ***
## Supporter_CareerStage_cleanPostdocs 0.5930
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.3462
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.0842 .
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.8419
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1227.8 on 1254 degrees of freedom
## Residual deviance: 1223.1 on 1250 degrees of freedom
## AIC: 1233.1
##
## Number of Fisher Scoring iterations: 4summary(glm(Receiver_Missing_OutPro~Supporter_CareerStage_clean*All_Gender_clean, data=Receiver_Missing, family="binomial")) #FLR gender-CS interactions##
## Call:
## glm(formula = Receiver_Missing_OutPro ~ Supporter_CareerStage_clean *
## All_Gender_clean, family = "binomial", data = Receiver_Missing)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.8876 -0.6792 -0.6489 -0.4727 2.1200
##
## Coefficients:
## Estimate
## (Intercept) -1.55060
## Supporter_CareerStage_cleanPostdocs -0.58493
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.32682
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.21560
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.04652
## All_Gender_cleanWomen 0.09954
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.68683
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.17983
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.50722
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.10525
## Std. Error
## (Intercept) 0.20806
## Supporter_CareerStage_cleanPostdocs 0.35958
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.35995
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.41183
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.36137
## All_Gender_cleanWomen 0.24632
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.41584
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.46246
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.54120
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.53795
## z value
## (Intercept) -7.453
## Supporter_CareerStage_cleanPostdocs -1.627
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.908
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.524
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.129
## All_Gender_cleanWomen 0.404
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 1.652
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.389
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.937
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.196
## Pr(>|z|)
## (Intercept) 9.16e-14
## Supporter_CareerStage_cleanPostdocs 0.1038
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.3639
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.6006
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.8976
## All_Gender_cleanWomen 0.6861
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.0986
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.6974
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.3487
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.8449
##
## (Intercept) ***
## Supporter_CareerStage_cleanPostdocs
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## All_Gender_cleanWomen
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen .
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1227.8 on 1254 degrees of freedom
## Residual deviance: 1215.0 on 1245 degrees of freedom
## AIC: 1235
##
## Number of Fisher Scoring iterations: 4Q14 Length of support (Section 3, panel 2)
Dependent variable: Lenght_Support (note the typo) Recode
- 0- less than six months
- 1: Less than six months
- 4: I’m still supporting, <6 mo
- 1- more than six months
- 2: 6mo-1yr
- 3: >1yr
- 5: I’m still supporting, >6 mo (ignore 6- not sure and 7- PNTA)
Model: Factorial Logistical regression: Length_Support ~ Gender * CS
#subsetting data and recoding response to binary
Length_Support=genderCS[,c("Lenght_Support", "Supporter_CareerStage_clean", "All_Gender_clean")]
colnames(Length_Support)[1]="Length_Support"
Length_Support=Length_Support[Length_Support$Length_Support %in% seq(1,5), ] #remove NAs and some choices
dim(Length_Support)[1] #this gives N## [1] 1207#recode
Length_Support[Length_Support$Length_Support==1, "Length_Support"]=0
Length_Support[Length_Support$Length_Support==4, "Length_Support"]=0
Length_Support[Length_Support$Length_Support==2, "Length_Support"]=1
Length_Support[Length_Support$Length_Support==3, "Length_Support"]=1
Length_Support[Length_Support$Length_Support==5, "Length_Support"]=1
#LR gender
summary(glm(Length_Support~All_Gender_clean, data=Length_Support, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Length_Support ~ All_Gender_clean, family = "binomial",
## data = Length_Support)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.6548 -1.6443 0.7661 0.7739 0.7739
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 1.0524 0.1069 9.842 <2e-16 ***
## All_Gender_cleanWomen 0.0233 0.1358 0.172 0.864
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1373.2 on 1206 degrees of freedom
## Residual deviance: 1373.1 on 1205 degrees of freedom
## AIC: 1377.1
##
## Number of Fisher Scoring iterations: 4#LR CS
summary(glm(Length_Support~Supporter_CareerStage_clean, data=Length_Support, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Length_Support ~ Supporter_CareerStage_clean, family = "binomial",
## data = Length_Support)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.7574 -1.6042 0.7594 0.8011 0.8040
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) 1.0960 0.1029 10.650
## Supporter_CareerStage_cleanPostdocs -0.1324 0.1585 -0.835
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.1241 0.2083 -0.596
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.2081 0.2784 0.747
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.1191 0.2421 0.492
## Pr(>|z|)
## (Intercept) <2e-16 ***
## Supporter_CareerStage_cleanPostdocs 0.404
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.551
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.455
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.623
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1373.2 on 1206 degrees of freedom
## Residual deviance: 1370.8 on 1202 degrees of freedom
## AIC: 1380.8
##
## Number of Fisher Scoring iterations: 4#FLR gender-CS interactions
summary(glm(Length_Support~Supporter_CareerStage_clean*All_Gender_clean, data=Length_Support, family="binomial")) ##
## Call:
## glm(formula = Length_Support ~ Supporter_CareerStage_clean *
## All_Gender_clean, family = "binomial", data = Length_Support)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.8078 -1.5220 0.7671 0.7683 0.8683
##
## Coefficients:
## Estimate
## (Intercept) 1.15418
## Supporter_CareerStage_cleanPostdocs -0.37293
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.19441
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.05884
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.16757
## All_Gender_cleanWomen -0.08155
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.36923
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.10260
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.28559
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen -0.20412
## Std. Error
## (Intercept) 0.19386
## Supporter_CareerStage_cleanPostdocs 0.27558
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.33825
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.39436
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.34168
## All_Gender_cleanWomen 0.22875
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.33802
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.43180
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.57059
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.50464
## z value
## (Intercept) 5.954
## Supporter_CareerStage_cleanPostdocs -1.353
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.575
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.149
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.490
## All_Gender_cleanWomen -0.356
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 1.092
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.238
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.501
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen -0.404
## Pr(>|z|)
## (Intercept) 2.62e-09
## Supporter_CareerStage_cleanPostdocs 0.176
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.565
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.881
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.624
## All_Gender_cleanWomen 0.721
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.275
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.812
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.617
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.686
##
## (Intercept) ***
## Supporter_CareerStage_cleanPostdocs
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## All_Gender_cleanWomen
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1373.2 on 1206 degrees of freedom
## Residual deviance: 1368.7 on 1197 degrees of freedom
## AIC: 1388.7
##
## Number of Fisher Scoring iterations: 4#effect of gender on response
kruskal.test(Length_Support ~ All_Gender_clean, data=Length_Support)##
## Kruskal-Wallis rank sum test
##
## data: Length_Support by All_Gender_clean
## Kruskal-Wallis chi-squared = 0.029399, df = 1, p-value = 0.8639#effect of CS on response
kruskal.test(Length_Support ~ Supporter_CareerStage_clean, data=Length_Support)##
## Kruskal-Wallis rank sum test
##
## data: Length_Support by Supporter_CareerStage_clean
## Kruskal-Wallis chi-squared = 2.39, df = 4, p-value = 0.6644(This alternative is not reported) Alternative recode
- 0- less than six months
- 1: Less than six months
- 4: I’m still supporting, <6 mo
- 1- 6 mo to 1 yr
- 2: 6mo-1yr
- 2- >1 yr
- 3: >1yr
ignore 5
Length_Support=genderCS[,c("Lenght_Support", "Supporter_CareerStage_clean", "All_Gender_clean")]
colnames(Length_Support)[1]="Length_Support"
Length_Support=Length_Support[Length_Support$Length_Support %in% seq(1,4), ] #remove NAs and some choices
dim(Length_Support)[1] #this gives N## [1] 842#recode
Length_Support[Length_Support$Length_Support==1, "Length_Support"]=0
Length_Support[Length_Support$Length_Support==4, "Length_Support"]=0
Length_Support[Length_Support$Length_Support==2, "Length_Support"]=1
Length_Support[Length_Support$Length_Support==3, "Length_Support"]=2
summary(glm(Length_Support~All_Gender_clean, data=Length_Support, family="gaussian"))##
## Call:
## glm(formula = Length_Support ~ All_Gender_clean, family = "gaussian",
## data = Length_Support)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.99407 -0.95446 0.04554 1.00593 1.04554
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 0.99407 0.04573 21.739 <2e-16 ***
## All_Gender_cleanWomen -0.03961 0.05905 -0.671 0.503
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 0.7046912)
##
## Null deviance: 592.26 on 841 degrees of freedom
## Residual deviance: 591.94 on 840 degrees of freedom
## AIC: 2098.8
##
## Number of Fisher Scoring iterations: 2summary(glm(Length_Support~Supporter_CareerStage_clean, data=Length_Support, family="gaussian"))##
## Call:
## glm(formula = Length_Support ~ Supporter_CareerStage_clean, family = "gaussian",
## data = Length_Support)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.15385 -0.93137 0.03143 1.03143 1.10417
##
## Coefficients:
## Estimate Std. Error t value
## (Intercept) 0.96857 0.04477 21.632
## Supporter_CareerStage_cleanPostdocs -0.07274 0.07020 -1.036
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.03720 0.09425 -0.395
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.18527 0.11313 1.638
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.12555 0.10129 1.239
## Pr(>|t|)
## (Intercept) <2e-16 ***
## Supporter_CareerStage_cleanPostdocs 0.300
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.693
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.102
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.216
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 0.7016467)
##
## Null deviance: 592.26 on 841 degrees of freedom
## Residual deviance: 587.28 on 837 degrees of freedom
## AIC: 2098.1
##
## Number of Fisher Scoring iterations: 2summary(glm(Length_Support~Supporter_CareerStage_clean*All_Gender_clean, data=Length_Support, family="gaussian"))##
## Call:
## glm(formula = Length_Support ~ Supporter_CareerStage_clean *
## All_Gender_clean, family = "gaussian", data = Length_Support)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.17143 -0.93151 0.03659 1.01923 1.15957
##
## Coefficients:
## Estimate
## (Intercept) 0.98077
## Supporter_CareerStage_cleanPostdocs -0.14034
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.01923
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.19066
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.17713
## All_Gender_cleanWomen -0.01735
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.10844
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.10992
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -0.02074
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen -0.17625
## Std. Error
## (Intercept) 0.08227
## Supporter_CareerStage_cleanPostdocs 0.11940
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.14746
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.16395
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.13827
## All_Gender_cleanWomen 0.09813
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.14812
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.19340
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.23066
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.21707
## t value
## (Intercept) 11.921
## Supporter_CareerStage_cleanPostdocs -1.175
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.130
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 1.163
## Supporter_CareerStage_cleanGroup leaders (>10yr) 1.281
## All_Gender_cleanWomen -0.177
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.732
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.568
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -0.090
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen -0.812
## Pr(>|t|)
## (Intercept) <2e-16
## Supporter_CareerStage_cleanPostdocs 0.240
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.896
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.245
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.201
## All_Gender_cleanWomen 0.860
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.464
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.570
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.928
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.417
##
## (Intercept) ***
## Supporter_CareerStage_cleanPostdocs
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## All_Gender_cleanWomen
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for gaussian family taken to be 0.7038992)
##
## Null deviance: 592.26 on 841 degrees of freedom
## Residual deviance: 585.64 on 832 degrees of freedom
## AIC: 2105.8
##
## Number of Fisher Scoring iterations: 2Q17c Impact_draining (Section 4, Panel 3 and Section 4, Panel 4)
Response: Impact_draining (1-5), PNTA(6) is removed
Model: Ordinal logistical regression
#subset
ImpactDraining=genderCS[,c("Impact_Draining", "All_Gender_clean", "Supporter_CareerStage_clean")]
summary(ImpactDraining) #there are some NAs## Impact_Draining All_Gender_clean Supporter_CareerStage_clean
## Min. :1.000 Men :474 PhD students :534
## 1st Qu.:4.000 Women:781 Postdocs :356
## Median :4.000 Group leaders (<5yr) :155
## Mean :4.068 Group leaders (5-10yr): 91
## 3rd Qu.:5.000 Group leaders (>10yr) :119
## Max. :6.000
## NA's :5#clean
ImpactDraining=ImpactDraining[(ImpactDraining$Impact_Draining %in% seq(1:5)),]
dim(ImpactDraining)[1] #this gives N## [1] 1242#factorise
ImpactDraining$Impact_Draining=as.factor(ImpactDraining$Impact_Draining)Impact_draining x gender stats (Section 4, panel 3)
exp(coef(polr(Impact_Draining~All_Gender_clean, data=ImpactDraining, Hess=TRUE, method="logistic"))) #OR (men as baseline)## All_Gender_cleanWomen
## 1.439005exp(confint(polr(Impact_Draining~All_Gender_clean, data=ImpactDraining, Hess=TRUE, method="logistic"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## 1.164422 1.778559kruskal.test(Impact_Draining ~ All_Gender_clean, data=ImpactDraining) #effect of gender on response##
## Kruskal-Wallis rank sum test
##
## data: Impact_Draining by All_Gender_clean
## Kruskal-Wallis chi-squared = 11.334, df = 1, p-value = 0.000761Impact Draining x CS (Section 4, panel 4)
exp(coef(polr(Impact_Draining~Supporter_CareerStage_clean, data=ImpactDraining, Hess=TRUE, method="logistic"))) # OR (PhD as baseline)## Supporter_CareerStage_cleanPostdocs
## 0.8308404
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 1.6016706
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 0.9114593
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 0.7460091exp(confint(polr(Impact_Draining~Supporter_CareerStage_clean, data=ImpactDraining, Hess=TRUE, method="logistic"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## Supporter_CareerStage_cleanPostdocs 0.6486015 1.064394
## Supporter_CareerStage_cleanGroup leaders (<5yr) 1.1383083 2.267367
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.6031863 1.383784
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.5170316 1.078493kruskal.test(Impact_Draining ~ Supporter_CareerStage_clean, data=ImpactDraining) #effect of CS on response##
## Kruskal-Wallis rank sum test
##
## data: Impact_Draining by Supporter_CareerStage_clean
## Kruskal-Wallis chi-squared = 15.29, df = 4, p-value = 0.004136# multiway kruskal wallis
## CSpairs=combinations(n=5, r=2, v=levels(ImpactDraining$Supporter_CareerStage_clean), repeats.allowed = F)
# Compare between early PIs and other CSs only
## this requires the package "MHTdiscrete"
CSpairs=cbind(rep("Group leaders (<5yr)", 4), levels(ImpactDraining$Supporter_CareerStage_clean)[c(1,2,4,5)])
ImpactDrainingCSKW=data.frame(CS1=c(), CS2=c(), X2=c(), N=c(), df=c(), pVal=c(), adjpVal=c())
for (i in 1:dim(CSpairs)[1]) {
comparedData=ImpactDraining[(ImpactDraining$Supporter_CareerStage_clean==CSpairs[i,1]|ImpactDraining$Supporter_CareerStage_clean==CSpairs[i,2]),]
KW=kruskal.test(Impact_Draining ~ Supporter_CareerStage_clean, data=comparedData) #effect of CS on response
ImpactDrainingCSKW=rbind(ImpactDrainingCSKW, data.frame(CS1=CSpairs[i,1], CS2=CSpairs[i,2], X2=as.numeric(KW$statistic), N=dim(comparedData)[1], df=as.numeric(KW$parameter), pVal=as.numeric(KW$p.value)))
}
ImpactDrainingCSKW$adjpVal=Sidak.p.adjust(ImpactDrainingCSKW$pVal)
ImpactDrainingCSKW## CS1 CS2 X2 N df pVal
## 1 Group leaders (<5yr) PhD students 7.269973 680 1 0.0070116865
## 2 Group leaders (<5yr) Postdocs 12.494864 507 1 0.0004080724
## 3 Group leaders (<5yr) Group leaders (5-10yr) 4.776699 244 1 0.0288473645
## 4 Group leaders (<5yr) Group leaders (>10yr) 10.288929 273 1 0.0013383071
## adjpVal
## 1 0.027753140
## 2 0.001631291
## 3 0.110491767
## 4 0.005342492Impact Draining x early PI vs others (PhD/postdoc/midPI/latePI)
# code a new variable based on genderCS$Supporter_CareerStage_clean, 1= early PI, 0=PhD/postdoc/midPI/latePI
genderCS$earlyPI=rep("Group leaders (<5yr)", nrow(genderCS))
genderCS[(genderCS$Supporter_CareerStage_clean!="Group leaders (<5yr)"),"earlyPI"]="Other career stages"
genderCS$earlyPI=as.factor(genderCS$earlyPI)
#redo subsetting
ImpactDrainingPI=genderCS[,c("Impact_Draining", "All_Gender_clean", "earlyPI")]
summary(ImpactDrainingPI) #there are some NAs## Impact_Draining All_Gender_clean earlyPI
## Min. :1.000 Men :474 Group leaders (<5yr): 155
## 1st Qu.:4.000 Women:781 Other career stages :1100
## Median :4.000
## Mean :4.068
## 3rd Qu.:5.000
## Max. :6.000
## NA's :5#clean
ImpactDrainingPI=ImpactDrainingPI[(ImpactDrainingPI$Impact_Draining %in% seq(1:5)),]
dim(ImpactDrainingPI)[1] #this gives N## [1] 1242#factorise
ImpactDrainingPI$Impact_Draining=as.factor(ImpactDrainingPI$Impact_Draining)
#kruskal wallis
kruskal.test(Impact_Draining ~ earlyPI, data=ImpactDrainingPI)##
## Kruskal-Wallis rank sum test
##
## data: Impact_Draining by earlyPI
## Kruskal-Wallis chi-squared = 11.726, df = 1, p-value = 0.0006162Impact_draining x gender graph (graph 4.3.1)
levels(ImpactDraining$Impact_Draining)=c("Strongly disagree", "Somewhat disagree", "Neither agree or disagree", "Somewhat agree", "Strongly agree")
ImpactDraining$Impact_Draining=factor(ImpactDraining$Impact_Draining, levels=c("Strongly agree", "Somewhat agree", "Neither agree or disagree", "Somewhat disagree", "Strongly disagree"))
graphdata = ImpactDraining %>%
group_by(All_Gender_clean, Impact_Draining) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'All_Gender_clean' (override with `.groups` argument)#graphdata$perc=0
#graphdata[graphdata$All_Gender_clean=="Women",]=mutate(graphdata[graphdata$All_Gender_clean=="Women",], perc=n/sum(n))
#graphdata[graphdata$All_Gender_clean=="Men",]=mutate(graphdata[graphdata$All_Gender_clean=="Men",], perc=n/sum(n))
# This will save the image to your local code folder
#eps("images/ImpactDraining_gender.eps", width=1000, height=578)
ggplot(graphdata, aes(x=All_Gender_clean, y=perc, fill=Impact_Draining)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=12), axis.title.x = element_text(size = 8), axis.title.y = element_text(size = 16), legend.position="bottom", legend.text = element_text(size=7))+
geom_text(aes(label=round(perc, digit=1)), size=3, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="The experience was emotionally draining or stressful") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=ePalette)
ggsave(dpi=1000, "4-3-1.png", limitsize = FALSE)## Saving 7 x 5 in image#dev.off()Impact_draining x career graph (graph 4.4.1)
#levels(ImpactDraining$Impact_Draining)=c("Strongly disagree", "Somewhat disagree", "Neither agree or #disagree", "Somewhat agree", "Strongly agree")
#ImpactDraining$Impact_Draining=factor(ImpactDraining$Impact_Draining, levels=c("Strongly agree", "Somewhat agree", "Neither agree or disagree", "Somewhat disagree", "Strongly disagree"))
graphdata = ImpactDraining %>%
group_by(Supporter_CareerStage_clean, Impact_Draining) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'Supporter_CareerStage_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/ImpactDraining_CS.eps", width=1000, height=578)
ggplot(graphdata, aes(x=rev(Supporter_CareerStage_clean), y=rev(perc), fill=rev(Impact_Draining))) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=12), axis.title.x = element_text(size = 8), axis.title.y = element_text(size = 16), legend.position="bottom", legend.text = element_text(size=7))+
geom_text(aes(label=round(rev(perc), digit=1)), size=3.5, position=position_stack(vjust=0.5), color="white") +
scale_x_discrete(limits = rev(levels(ImpactDraining$Supporter_CareerStage_clean))) +
labs(x="", y="Percentage", title="The experience was emotionally draining or stressful") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=ePalette)
ggsave(dpi=1000, "4-4-1.png", limitsize = FALSE)## Saving 7 x 5 in image#dev.off()It looks like career stage 6 (early PIs) are the only group more likely to find it more draining than PhDs (career stage 3, also the intercept here).
We may be able to Chi-square here to show early career PIs are significantly different from other career stages? My concern is big differences in sample sizes
Chi squares for interactions between CS and gender [NOT REPORTED]
#split by career stages, gender effect
chisq.test(table(ImpactDraining[ImpactDraining$Supporter_CareerStage_clean==levels(ImpactDraining$Supporter_CareerStage_clean)[1],c("Impact_Draining","All_Gender_clean")]))## Warning in
## chisq.test(table(ImpactDraining[ImpactDraining$Supporter_CareerStage_clean == :
## Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: table(ImpactDraining[ImpactDraining$Supporter_CareerStage_clean == levels(ImpactDraining$Supporter_CareerStage_clean)[1], c("Impact_Draining", "All_Gender_clean")])
## X-squared = 1.2496, df = 4, p-value = 0.8699chisq.test(table(ImpactDraining[ImpactDraining$Supporter_CareerStage_clean==levels(ImpactDraining$Supporter_CareerStage_clean)[2],c("Impact_Draining","All_Gender_clean")]))##
## Pearson's Chi-squared test
##
## data: table(ImpactDraining[ImpactDraining$Supporter_CareerStage_clean == levels(ImpactDraining$Supporter_CareerStage_clean)[2], c("Impact_Draining", "All_Gender_clean")])
## X-squared = 3.1639, df = 4, p-value = 0.5308chisq.test(table(ImpactDraining[ImpactDraining$Supporter_CareerStage_clean==levels(ImpactDraining$Supporter_CareerStage_clean)[3],c("Impact_Draining","All_Gender_clean")]))## Warning in
## chisq.test(table(ImpactDraining[ImpactDraining$Supporter_CareerStage_clean == :
## Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: table(ImpactDraining[ImpactDraining$Supporter_CareerStage_clean == levels(ImpactDraining$Supporter_CareerStage_clean)[3], c("Impact_Draining", "All_Gender_clean")])
## X-squared = 6.2074, df = 4, p-value = 0.1842chisq.test(table(ImpactDraining[ImpactDraining$Supporter_CareerStage_clean==levels(ImpactDraining$Supporter_CareerStage_clean)[4],c("Impact_Draining","All_Gender_clean")]))## Warning in
## chisq.test(table(ImpactDraining[ImpactDraining$Supporter_CareerStage_clean == :
## Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: table(ImpactDraining[ImpactDraining$Supporter_CareerStage_clean == levels(ImpactDraining$Supporter_CareerStage_clean)[4], c("Impact_Draining", "All_Gender_clean")])
## X-squared = 6.2421, df = 4, p-value = 0.1818chisq.test(table(ImpactDraining[ImpactDraining$Supporter_CareerStage_clean==levels(ImpactDraining$Supporter_CareerStage_clean)[5],c("Impact_Draining","All_Gender_clean")]))## Warning in
## chisq.test(table(ImpactDraining[ImpactDraining$Supporter_CareerStage_clean == :
## Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: table(ImpactDraining[ImpactDraining$Supporter_CareerStage_clean == levels(ImpactDraining$Supporter_CareerStage_clean)[5], c("Impact_Draining", "All_Gender_clean")])
## X-squared = 14.345, df = 4, p-value = 0.006273#split by gender, career effect (same as KW above)
chisq.test(table(ImpactDraining[ImpactDraining$All_Gender_clean==levels(ImpactDraining$All_Gender_clean)[1],c("Impact_Draining","Supporter_CareerStage_clean")]))## Warning in chisq.test(table(ImpactDraining[ImpactDraining$All_Gender_clean == :
## Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: table(ImpactDraining[ImpactDraining$All_Gender_clean == levels(ImpactDraining$All_Gender_clean)[1], c("Impact_Draining", "Supporter_CareerStage_clean")])
## X-squared = 20.268, df = 16, p-value = 0.2084chisq.test(table(ImpactDraining[ImpactDraining$All_Gender_clean==levels(ImpactDraining$All_Gender_clean)[2],c("Impact_Draining","Supporter_CareerStage_clean")]))## Warning in chisq.test(table(ImpactDraining[ImpactDraining$All_Gender_clean == :
## Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: table(ImpactDraining[ImpactDraining$All_Gender_clean == levels(ImpactDraining$All_Gender_clean)[2], c("Impact_Draining", "Supporter_CareerStage_clean")])
## X-squared = 20.323, df = 16, p-value = 0.206Q17d Impact Time (Seciont 4, Panel 3)
#Subsetting
ImpactTime=genderCS[,c("Impact_Time", "All_Gender_clean", "Supporter_CareerStage_clean")]
#clean
ImpactTime=ImpactTime[(ImpactTime$Impact_Time %in% seq(1:5)),]
dim(ImpactTime)[1] #N## [1] 1241#factorise
ImpactTime$Impact_Time=as.factor(ImpactTime$Impact_Time)
table(ImpactTime$Impact_Time)##
## 1 2 3 4 5
## 80 189 218 406 348Impact Time x gender (Section 4, Panel 3)
exp(coef(polr(Impact_Time~All_Gender_clean, data=ImpactTime, Hess=TRUE, method="logistic"))) #OR (men as baseline)## All_Gender_cleanWomen
## 1.43038exp(confint(polr(Impact_Time~All_Gender_clean, data=ImpactTime, Hess=TRUE, method="logistic"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## 1.164197 1.758101kruskal.test(Impact_Time ~ All_Gender_clean, data=ImpactTime) #effect of gender on response##
## Kruskal-Wallis rank sum test
##
## data: Impact_Time by All_Gender_clean
## Kruskal-Wallis chi-squared = 11.577, df = 1, p-value = 0.0006679Impact Time x Career stage
exp(coef(polr(Impact_Time~Supporter_CareerStage_clean, data=ImpactTime, Hess=TRUE, method="logistic"))) #OR (PhD as baseline)## Supporter_CareerStage_cleanPostdocs
## 0.9808452
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 1.5752583
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 1.0570449
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 1.1509917kruskal.test(Impact_Time ~ Supporter_CareerStage_clean, data=ImpactTime) #effect of CS on response##
## Kruskal-Wallis rank sum test
##
## data: Impact_Time by Supporter_CareerStage_clean
## Kruskal-Wallis chi-squared = 8.9056, df = 4, p-value = 0.0635Impact Time x gender graph (graph 4.3.2)
levels(ImpactTime$Impact_Time)=c("Strongly disagree", "Somewhat disagree", "Neither agree or disagree", "Somewhat agree", "Strongly agree")
ImpactTime$Impact_Time=factor(ImpactTime$Impact_Time, levels=c("Strongly agree", "Somewhat agree", "Neither agree or disagree", "Somewhat disagree", "Strongly disagree"))
graphdata = ImpactTime %>%
group_by(All_Gender_clean, Impact_Time) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'All_Gender_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/ImpactTime_gender.eps", width=1000, height=578)
ggplot(graphdata, aes(x=All_Gender_clean, y=perc, fill=Impact_Time)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=12), axis.title.x = element_text(size = 8), axis.title.y = element_text(size = 16), legend.position="bottom", legend.text = element_text(size=7))+
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="The experience took a lot of my time") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=ePalette)
ggsave(dpi=1000, "4-3-2.png", limitsize = FALSE)## Saving 7 x 5 in imagedev.off()## null device
## 1Impact Time x Career stage graph
graphdata = ImpactTime %>%
group_by(Supporter_CareerStage_clean, Impact_Time) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'Supporter_CareerStage_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/ImpactTime_CS.eps", width=1000, height=578)
ggplot(graphdata, aes(x=Supporter_CareerStage_clean, y=perc, fill=Impact_Time)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=20), legend.position="bottom", legend.text = element_text(size=7))+
scale_x_discrete(limits = rev(levels(ImpactTime$Supporter_CareerStage_clean))) +
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="The experience took a lot of my time") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=ePalette)
dev.off()## null device
## 1Q17e Impact work (Section 4, Panel 3 and Section 4, Panel 4)
#Subset
ImpactWork=genderCS[,c("Impact_Work", "All_Gender_clean", "Supporter_CareerStage_clean")]
#clean
ImpactWork=ImpactWork[(ImpactWork$Impact_Work %in% seq(1:5)),]
dim(ImpactWork)[1]## [1] 1235#factorise
ImpactWork$Impact_Work=as.factor(ImpactWork$Impact_Work)
table(ImpactWork$Impact_Work)##
## 1 2 3 4 5
## 340 239 227 276 153Impact work x gender (Section 4, Panel 3)
exp(coef(polr(Impact_Work~All_Gender_clean, data=ImpactWork, Hess=TRUE, method="logistic"))) #OR (men as baseline)## All_Gender_cleanWomen
## 1.236169exp(confint(polr(Impact_Work~All_Gender_clean, data=ImpactWork, Hess=TRUE, method="logistic"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## 1.007312 1.517617kruskal.test(Impact_Work ~ All_Gender_clean, data=ImpactWork) #effect of gender on response##
## Kruskal-Wallis rank sum test
##
## data: Impact_Work by All_Gender_clean
## Kruskal-Wallis chi-squared = 4.1104, df = 1, p-value = 0.04262Impact work x Career stage (Section 4, Panel 4)
exp(coef(polr(Impact_Work~Supporter_CareerStage_clean, data=ImpactWork, Hess=TRUE, method="logistic"))) #OR (PhD as baseline)## Supporter_CareerStage_cleanPostdocs
## 1.052155
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 2.255425
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 1.477751
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 1.099569exp(confint(polr(Impact_Work~Supporter_CareerStage_clean, data=ImpactWork, Hess=TRUE, method="logistic"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## Supporter_CareerStage_cleanPostdocs 0.8274488 1.337784
## Supporter_CareerStage_cleanGroup leaders (<5yr) 1.6313683 3.122104
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.9903085 2.206576
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.7715605 1.565985kruskal.test(Impact_Work ~ Supporter_CareerStage_clean, data=ImpactWork) #effect of CS on response##
## Kruskal-Wallis rank sum test
##
## data: Impact_Work by Supporter_CareerStage_clean
## Kruskal-Wallis chi-squared = 26.305, df = 4, p-value = 2.747e-05## multiway kruskal wallis
# Compare between early PIs and other CSs onl
## this requires the package "MTHdiscrete", and having run the similar analysis in the Impact draining section
ImpactWorkCSKW=data.frame(CS1=c(), CS2=c(), X2=c(), N=c(), df=c(), pVal=c())
for (i in 1:dim(CSpairs)[1]) {
comparedData=ImpactWork[(ImpactWork$Supporter_CareerStage_clean==CSpairs[i,1]|ImpactWork$Supporter_CareerStage_clean==CSpairs[i,2]),]
KW=kruskal.test(Impact_Work ~ Supporter_CareerStage_clean, data=comparedData) #effect of CS on response
ImpactWorkCSKW=rbind(ImpactWorkCSKW, data.frame(CS1=CSpairs[i,1], CS2=CSpairs[i,2], X2=as.numeric(KW$statistic), N=dim(comparedData)[1], df=as.numeric(KW$parameter), pVal=as.numeric(KW$p.value)))
}
ImpactWorkCSKW$adjpVal=Sidak.p.adjust(ImpactWorkCSKW$pVal)
ImpactWorkCSKW## CS1 CS2 X2 N df pVal
## 1 Group leaders (<5yr) PhD students 23.078118 680 1 1.555510e-06
## 2 Group leaders (<5yr) Postdocs 18.083088 504 1 2.114714e-05
## 3 Group leaders (<5yr) Group leaders (5-10yr) 3.208835 238 1 7.324162e-02
## 4 Group leaders (<5yr) Group leaders (>10yr) 10.020563 272 1 1.548022e-03
## adjpVal
## 1 6.222024e-06
## 2 8.458587e-05
## 3 2.623233e-01
## 4 6.177724e-03Impact work x PI vs others
#redo subsetting
ImpactWorkPI=genderCS[,c("Impact_Work", "All_Gender_clean", "earlyPI")]
summary(ImpactWorkPI) #there are some NAs## Impact_Work All_Gender_clean earlyPI
## Min. :1.000 Men :474 Group leaders (<5yr): 155
## 1st Qu.:1.000 Women:781 Other career stages :1100
## Median :3.000
## Mean :2.764
## 3rd Qu.:4.000
## Max. :6.000
## NA's :6#clean
ImpactWorkPI=ImpactWorkPI[(ImpactWorkPI$Impact_Work %in% seq(1:5)),]
dim(ImpactWorkPI)[1] #this gives N## [1] 1235#factorise
ImpactWorkPI$Impact_Work=as.factor(ImpactWorkPI$Impact_Work)
#kruskal wallis
kruskal.test(Impact_Work ~ earlyPI, data=ImpactWorkPI)##
## Kruskal-Wallis rank sum test
##
## data: Impact_Work by earlyPI
## Kruskal-Wallis chi-squared = 22.576, df = 1, p-value = 2.02e-06Impact work x gender graph (Graph 4.3.4)
levels(ImpactWork$Impact_Work)=c("Strongly disagree", "Somewhat disagree", "Neither agree or disagree", "Somewhat agree", "Strongly agree")
ImpactWork$Impact_Work=factor(ImpactWork$Impact_Work, levels=c("Strongly agree", "Somewhat agree", "Neither agree or disagree", "Somewhat disagree", "Strongly disagree"))
graphdata = ImpactWork %>%
group_by(All_Gender_clean, Impact_Work) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'All_Gender_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/ImpactWork_gender.eps", width=1000, height=578)
ggplot(graphdata, aes(x=All_Gender_clean, y=perc, fill=Impact_Work)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=12), axis.title.x = element_text(size = 8), axis.title.y = element_text(size = 16), legend.position="bottom", legend.text = element_text(size=7))+
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="The experience negatively impacted my academic work") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=ePalette)
ggsave(dpi=1000, "4-3-4 GOOD.png", limitsize = FALSE)## Saving 7 x 5 in imagedev.off()## null device
## 1Impact work x Career stage graph (Graph 4.4.2)
graphdata = ImpactWork %>%
group_by(Supporter_CareerStage_clean, Impact_Work) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'Supporter_CareerStage_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/ImpactWork_CS.eps", width=1000, height=578)
ggplot(graphdata, aes(x=Supporter_CareerStage_clean, y=perc, fill=rev(Impact_Work))) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=12), axis.title.x = element_text(size = 8), axis.title.y = element_text(size = 16), legend.position="bottom", legend.text = element_text(size=7))+
scale_x_discrete(limits = rev(levels(ImpactWork$Supporter_CareerStage_clean))) +
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="The experience negatively impacted my academic work") +
guides(fill=guide_legend(reverse=FALSE)) +
coord_flip() +
scale_fill_manual(name="", values=ePalette)
ggsave(dpi=1000, "4-4-2.png", limitsize = FALSE)## Saving 7 x 5 in imagedev.off()## null device
## 1chisq for intersections [NOT REPORTED]
#split by career stages, gender effect
chisq.test(table(ImpactWork[ImpactWork$Supporter_CareerStage_clean==levels(ImpactWork$Supporter_CareerStage_clean)[1],c("Impact_Work","All_Gender_clean")]))##
## Pearson's Chi-squared test
##
## data: table(ImpactWork[ImpactWork$Supporter_CareerStage_clean == levels(ImpactWork$Supporter_CareerStage_clean)[1], c("Impact_Work", "All_Gender_clean")])
## X-squared = 6.4905, df = 4, p-value = 0.1654chisq.test(table(ImpactWork[ImpactWork$Supporter_CareerStage_clean==levels(ImpactWork$Supporter_CareerStage_clean)[2],c("Impact_Work","All_Gender_clean")]))##
## Pearson's Chi-squared test
##
## data: table(ImpactWork[ImpactWork$Supporter_CareerStage_clean == levels(ImpactWork$Supporter_CareerStage_clean)[2], c("Impact_Work", "All_Gender_clean")])
## X-squared = 6.6388, df = 4, p-value = 0.1562chisq.test(table(ImpactWork[ImpactWork$Supporter_CareerStage_clean==levels(ImpactWork$Supporter_CareerStage_clean)[3],c("Impact_Work","All_Gender_clean")]))##
## Pearson's Chi-squared test
##
## data: table(ImpactWork[ImpactWork$Supporter_CareerStage_clean == levels(ImpactWork$Supporter_CareerStage_clean)[3], c("Impact_Work", "All_Gender_clean")])
## X-squared = 1.8403, df = 4, p-value = 0.7651chisq.test(table(ImpactWork[ImpactWork$Supporter_CareerStage_clean==levels(ImpactWork$Supporter_CareerStage_clean)[4],c("Impact_Work","All_Gender_clean")]))##
## Pearson's Chi-squared test
##
## data: table(ImpactWork[ImpactWork$Supporter_CareerStage_clean == levels(ImpactWork$Supporter_CareerStage_clean)[4], c("Impact_Work", "All_Gender_clean")])
## X-squared = 6.5609, df = 4, p-value = 0.161chisq.test(table(ImpactWork[ImpactWork$Supporter_CareerStage_clean==levels(ImpactWork$Supporter_CareerStage_clean)[5],c("Impact_Work","All_Gender_clean")]))## Warning in chisq.test(table(ImpactWork[ImpactWork$Supporter_CareerStage_clean
## == : Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: table(ImpactWork[ImpactWork$Supporter_CareerStage_clean == levels(ImpactWork$Supporter_CareerStage_clean)[5], c("Impact_Work", "All_Gender_clean")])
## X-squared = 14.276, df = 4, p-value = 0.006464#split by gender, career effect (same as KW above)
chisq.test(table(ImpactWork[ImpactWork$All_Gender_clean==levels(ImpactWork$All_Gender_clean)[1],c("Impact_Work","Supporter_CareerStage_clean")]))## Warning in chisq.test(table(ImpactWork[ImpactWork$All_Gender_clean ==
## levels(ImpactWork$All_Gender_clean)[1], : Chi-squared approximation may be
## incorrect##
## Pearson's Chi-squared test
##
## data: table(ImpactWork[ImpactWork$All_Gender_clean == levels(ImpactWork$All_Gender_clean)[1], c("Impact_Work", "Supporter_CareerStage_clean")])
## X-squared = 30.103, df = 16, p-value = 0.01747chisq.test(table(ImpactWork[ImpactWork$All_Gender_clean==levels(ImpactWork$All_Gender_clean)[2],c("Impact_Work","Supporter_CareerStage_clean")]))##
## Pearson's Chi-squared test
##
## data: table(ImpactWork[ImpactWork$All_Gender_clean == levels(ImpactWork$All_Gender_clean)[2], c("Impact_Work", "Supporter_CareerStage_clean")])
## X-squared = 35.752, df = 16, p-value = 0.003132Q17f Impact personal (Section 4, Panel 3 and Section 4, Panel 4)
ImpactPersonal=genderCS[,c("Impact_Personal", "All_Gender_clean", "Supporter_CareerStage_clean")]
#clean
ImpactPersonal=ImpactPersonal[(ImpactPersonal$Impact_Personal %in% seq(1:5)),]
dim(ImpactPersonal)[1]## [1] 1235#factorise
ImpactPersonal$Impact_Personal=as.factor(ImpactPersonal$Impact_Personal)
table(ImpactPersonal$Impact_Personal)##
## 1 2 3 4 5
## 124 126 156 456 373Impact personal x gender stats (Section 4, Panel 3)
exp(coef(polr(Impact_Personal~All_Gender_clean, data=ImpactPersonal, Hess=TRUE, method="logistic"))) #OR (men as baseline)## All_Gender_cleanWomen
## 1.462125exp(confint(polr(Impact_Personal~All_Gender_clean, data=ImpactPersonal, Hess=TRUE, method="logistic"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## 1.186809 1.802054kruskal.test(Impact_Personal ~ All_Gender_clean, data=ImpactPersonal) #effect of gender on response##
## Kruskal-Wallis rank sum test
##
## data: Impact_Personal by All_Gender_clean
## Kruskal-Wallis chi-squared = 12.646, df = 1, p-value = 0.0003765Impact personal x Career stage (Section 4, Panel 4)
exp(coef(polr(Impact_Personal~Supporter_CareerStage_clean, data=ImpactPersonal, Hess=TRUE, method="logistic"))) #OR (PhD as baseline)## Supporter_CareerStage_cleanPostdocs
## 0.7817945
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 0.8789771
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 0.8183547
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 0.4916460exp(confint(polr(Impact_Personal~Supporter_CareerStage_clean, data=ImpactPersonal, Hess=TRUE, method="logistic"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## Supporter_CareerStage_cleanPostdocs 0.6109090 1.0002763
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.6330860 1.2218076
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.5489079 1.2234210
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.3439982 0.7027375kruskal.test(Impact_Personal ~ Supporter_CareerStage_clean, data=ImpactPersonal) #effect of CS on response##
## Kruskal-Wallis rank sum test
##
## data: Impact_Personal by Supporter_CareerStage_clean
## Kruskal-Wallis chi-squared = 16.143, df = 4, p-value = 0.002833Impact personal x gender graph (Graph 4.3.3)
levels(ImpactPersonal$Impact_Personal)=c("Strongly disagree", "Somewhat disagree", "Neither agree or disagree", "Somewhat agree", "Strongly agree")
ImpactPersonal$Impact_Personal=factor(ImpactPersonal$Impact_Personal, levels=c("Strongly agree", "Somewhat agree", "Neither agree or disagree", "Somewhat disagree", "Strongly disagree"))
# This will save the image to your local code folder
#eps("images/ImpactPersonal_gender.eps", width=1000, height=578)
graphdata = ImpactPersonal %>%
group_by(All_Gender_clean, Impact_Personal) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'All_Gender_clean' (override with `.groups` argument)ggplot(graphdata, aes(x=All_Gender_clean, y=perc, fill=Impact_Personal)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=12), axis.title.x = element_text(size = 8), axis.title.y = element_text(size = 16), legend.position="bottom", legend.text = element_text(size=7))+
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="The experience had an impact on my personal life") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=ePalette)
ggsave(dpi=1000, "4-3-3.png", limitsize = FALSE)## Saving 7 x 5 in imagedev.off()## null device
## 1Impact personal x Career stage (Graph 4.4.3)
#split by gender
kruskal.test(Impact_Personal ~ Supporter_CareerStage_clean, data=ImpactPersonal[ImpactPersonal$All_Gender_clean==levels(ImpactPersonal$All_Gender_clean)[1],])##
## Kruskal-Wallis rank sum test
##
## data: Impact_Personal by Supporter_CareerStage_clean
## Kruskal-Wallis chi-squared = 20.688, df = 4, p-value = 0.0003651kruskal.test(Impact_Personal ~ Supporter_CareerStage_clean, data=ImpactPersonal[ImpactPersonal$All_Gender_clean==levels(ImpactPersonal$All_Gender_clean)[2],])##
## Kruskal-Wallis rank sum test
##
## data: Impact_Personal by Supporter_CareerStage_clean
## Kruskal-Wallis chi-squared = 0.42513, df = 4, p-value = 0.9804#split by career stages, gender effect
chisq.test(table(ImpactPersonal[ImpactPersonal$Supporter_CareerStage_clean==levels(ImpactPersonal$Supporter_CareerStage_clean)[1],c("Impact_Personal","All_Gender_clean")]))##
## Pearson's Chi-squared test
##
## data: table(ImpactPersonal[ImpactPersonal$Supporter_CareerStage_clean == levels(ImpactPersonal$Supporter_CareerStage_clean)[1], c("Impact_Personal", "All_Gender_clean")])
## X-squared = 2.504, df = 4, p-value = 0.6439chisq.test(table(ImpactPersonal[ImpactPersonal$Supporter_CareerStage_clean==levels(ImpactPersonal$Supporter_CareerStage_clean)[2],c("Impact_Personal","All_Gender_clean")]))##
## Pearson's Chi-squared test
##
## data: table(ImpactPersonal[ImpactPersonal$Supporter_CareerStage_clean == levels(ImpactPersonal$Supporter_CareerStage_clean)[2], c("Impact_Personal", "All_Gender_clean")])
## X-squared = 11.282, df = 4, p-value = 0.02358chisq.test(table(ImpactPersonal[ImpactPersonal$Supporter_CareerStage_clean==levels(ImpactPersonal$Supporter_CareerStage_clean)[3],c("Impact_Personal","All_Gender_clean")]))##
## Pearson's Chi-squared test
##
## data: table(ImpactPersonal[ImpactPersonal$Supporter_CareerStage_clean == levels(ImpactPersonal$Supporter_CareerStage_clean)[3], c("Impact_Personal", "All_Gender_clean")])
## X-squared = 10.307, df = 4, p-value = 0.03556chisq.test(table(ImpactPersonal[ImpactPersonal$Supporter_CareerStage_clean==levels(ImpactPersonal$Supporter_CareerStage_clean)[4],c("Impact_Personal","All_Gender_clean")]))## Warning in
## chisq.test(table(ImpactPersonal[ImpactPersonal$Supporter_CareerStage_clean == :
## Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: table(ImpactPersonal[ImpactPersonal$Supporter_CareerStage_clean == levels(ImpactPersonal$Supporter_CareerStage_clean)[4], c("Impact_Personal", "All_Gender_clean")])
## X-squared = 4.2477, df = 4, p-value = 0.3735chisq.test(table(ImpactPersonal[ImpactPersonal$Supporter_CareerStage_clean==levels(ImpactPersonal$Supporter_CareerStage_clean)[5],c("Impact_Personal","All_Gender_clean")]))##
## Pearson's Chi-squared test
##
## data: table(ImpactPersonal[ImpactPersonal$Supporter_CareerStage_clean == levels(ImpactPersonal$Supporter_CareerStage_clean)[5], c("Impact_Personal", "All_Gender_clean")])
## X-squared = 8.6889, df = 4, p-value = 0.06936#split by gender, career effect (same as KW above)
chisq.test(table(ImpactPersonal[ImpactPersonal$All_Gender_clean==levels(ImpactPersonal$All_Gender_clean)[1],c("Impact_Personal","Supporter_CareerStage_clean")]))##
## Pearson's Chi-squared test
##
## data: table(ImpactPersonal[ImpactPersonal$All_Gender_clean == levels(ImpactPersonal$All_Gender_clean)[1], c("Impact_Personal", "Supporter_CareerStage_clean")])
## X-squared = 42.832, df = 16, p-value = 0.0002962chisq.test(table(ImpactPersonal[ImpactPersonal$All_Gender_clean==levels(ImpactPersonal$All_Gender_clean)[2],c("Impact_Personal","Supporter_CareerStage_clean")]))## Warning in chisq.test(table(ImpactPersonal[ImpactPersonal$All_Gender_clean == :
## Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: table(ImpactPersonal[ImpactPersonal$All_Gender_clean == levels(ImpactPersonal$All_Gender_clean)[2], c("Impact_Personal", "Supporter_CareerStage_clean")])
## X-squared = 16.755, df = 16, p-value = 0.4016graphdata = ImpactPersonal %>%
group_by(Supporter_CareerStage_clean, Impact_Personal) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'Supporter_CareerStage_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/ImpactPersonal_CS.eps", width=1000, height=578)
ggplot(graphdata, aes(x=Supporter_CareerStage_clean, y=perc, fill=Impact_Personal)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=12), axis.title.x = element_text(size = 8), axis.title.y = element_text(size = 16), legend.position="bottom", legend.text = element_text(size=7))+
scale_x_discrete(limits = rev(levels(ImpactPersonal$Supporter_CareerStage_clean))) +
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="The experience had an impact on my personal life") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=ePalette)
ggsave(dpi=1000, "4-4-3-TRY.png", limitsize = FALSE)## Saving 7 x 5 in image#dev.off()Q12 Professional support for the receiver (Section 3, Panel 2)
Dependent variable: Receiver_ProfessionalHelp
Original question: During the time I was supporting them, the person received at least some professional help * 1: No * 2: Yes, it started before I started supporting them * 3: Yes, it started after I started supporting them * 4: Yes, but I’m unsure when it started (remove 5- not sure, and 6 - PNTA)
We would like to check if respondents supported someone who doesn’t have professional support at the time the support was offered.
hence recode to: * 0: No(1), and (3) Yes it started after * 1: Yes, it started before (2) removing 2- unsure.
Model: First individual factors (Gender, CS), then Factorial logistical regression Supporter_NumberReceivers ~ Supporter_CareerStage * All_Gender_clean) to check for interactions
# subsetting data and recoding response to binary
Receiver_ProfessionalHelp=genderCS[,c("Receiver_ProfessionalHelp", "Supporter_CareerStage_clean", "All_Gender_clean")]
#recode andclean
Receiver_ProfessionalHelp[(Receiver_ProfessionalHelp$Receiver_ProfessionalHelp %in% c(1,3)),"Receiver_ProfessionalHelp"]=0
Receiver_ProfessionalHelp[(Receiver_ProfessionalHelp$Receiver_ProfessionalHelp %in% c(2)),"Receiver_ProfessionalHelp"]=1
Receiver_ProfessionalHelp=Receiver_ProfessionalHelp[Receiver_ProfessionalHelp$Receiver_ProfessionalHelp %in% seq(0,1), ] #remove NAs and some choices
dim(Receiver_ProfessionalHelp)[1]## [1] 873#LR gender
summary(glm(Receiver_ProfessionalHelp~All_Gender_clean, data=Receiver_ProfessionalHelp, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Receiver_ProfessionalHelp ~ All_Gender_clean, family = "binomial",
## data = Receiver_ProfessionalHelp)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.8531 -0.8531 -0.8216 1.5410 1.5812
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.82345 0.12180 -6.761 1.37e-11 ***
## All_Gender_cleanWomen -0.08906 0.15378 -0.579 0.563
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1056.4 on 872 degrees of freedom
## Residual deviance: 1056.0 on 871 degrees of freedom
## AIC: 1060
##
## Number of Fisher Scoring iterations: 4#LR CS
summary(glm(Receiver_ProfessionalHelp~Supporter_CareerStage_clean, data=Receiver_ProfessionalHelp, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Receiver_ProfessionalHelp ~ Supporter_CareerStage_clean,
## family = "binomial", data = Receiver_ProfessionalHelp)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.1362 -0.7644 -0.7608 1.2191 1.6620
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) -1.09179 0.11674 -9.352
## Supporter_CareerStage_cleanPostdocs 0.01087 0.19268 0.056
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.54656 0.23048 2.371
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.37617 0.28514 1.319
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.99415 0.25005 3.976
## Pr(>|z|)
## (Intercept) < 2e-16 ***
## Supporter_CareerStage_cleanPostdocs 0.9550
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.0177 *
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.1871
## Supporter_CareerStage_cleanGroup leaders (>10yr) 7.01e-05 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1056.4 on 872 degrees of freedom
## Residual deviance: 1036.0 on 868 degrees of freedom
## AIC: 1046
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Receiver_ProfessionalHelp~Supporter_CareerStage_clean, data=Receiver_ProfessionalHelp, family="binomial"))) #OR (PhD as baseline)## (Intercept)
## 0.3356164
## Supporter_CareerStage_cleanPostdocs
## 1.0109329
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 1.7272996
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 1.4566893
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 2.7024205exp(confint(glm(Receiver_ProfessionalHelp~Supporter_CareerStage_clean, data=Receiver_ProfessionalHelp, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.2657445 0.4201863
## Supporter_CareerStage_cleanPostdocs 0.6904333 1.4709849
## Supporter_CareerStage_cleanGroup leaders (<5yr) 1.0941086 2.7055760
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.8212745 2.5234654
## Supporter_CareerStage_cleanGroup leaders (>10yr) 1.6529346 4.4158696#FLR gender-CS interactions
summary(glm(Receiver_ProfessionalHelp~Supporter_CareerStage_clean*All_Gender_clean, data=Receiver_ProfessionalHelp, family="binomial"))##
## Call:
## glm(formula = Receiver_ProfessionalHelp ~ Supporter_CareerStage_clean *
## All_Gender_clean, family = "binomial", data = Receiver_ProfessionalHelp)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.2278 -0.7893 -0.7809 1.2858 1.7537
##
## Coefficients:
## Estimate
## (Intercept) -1.29578
## Supporter_CareerStage_cleanPostdocs 0.09808
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.98212
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 1.00810
## Supporter_CareerStage_cleanGroup leaders (>10yr) 1.04447
## All_Gender_cleanWomen 0.28924
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -0.12290
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.69226
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -1.27453
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.07985
## Std. Error
## (Intercept) 0.22133
## Supporter_CareerStage_cleanPostdocs 0.36105
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.37427
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.40701
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.36557
## All_Gender_cleanWomen 0.26068
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.42709
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.47950
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.60620
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.52025
## z value
## (Intercept) -5.854
## Supporter_CareerStage_cleanPostdocs 0.272
## Supporter_CareerStage_cleanGroup leaders (<5yr) 2.624
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 2.477
## Supporter_CareerStage_cleanGroup leaders (>10yr) 2.857
## All_Gender_cleanWomen 1.110
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -0.288
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -1.444
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -2.102
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.153
## Pr(>|z|)
## (Intercept) 4.79e-09
## Supporter_CareerStage_cleanPostdocs 0.78589
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.00869
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.01325
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.00428
## All_Gender_cleanWomen 0.26718
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.77353
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.14882
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.03551
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.87801
##
## (Intercept) ***
## Supporter_CareerStage_cleanPostdocs
## Supporter_CareerStage_cleanGroup leaders (<5yr) **
## Supporter_CareerStage_cleanGroup leaders (5-10yr) *
## Supporter_CareerStage_cleanGroup leaders (>10yr) **
## All_Gender_cleanWomen
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen *
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1056.4 on 872 degrees of freedom
## Residual deviance: 1029.4 on 863 degrees of freedom
## AIC: 1049.4
##
## Number of Fisher Scoring iterations: 4Professional support for the receiver x (PhD+ Postdoc) vs PIs
Code new variable PI: - 0: PhD+ postdoc - 1: PIs of all stages
genderCS$PI="Group leaders"
genderCS[(genderCS$Supporter_CareerStage_clean==levels(genderCS$Supporter_CareerStage_clean)[1] | genderCS$Supporter_CareerStage_clean==levels(genderCS$Supporter_CareerStage_clean)[2]),'PI']="PhDs and Postdocs"
genderCS$PI=as.factor(genderCS$PI)
genderCS$PI=relevel(genderCS$PI, "PhDs and Postdocs") #relevel so that "PhDs and postdocs"" is baseline
summary(genderCS$PI)## PhDs and Postdocs Group leaders
## 890 365# redo subsetting
# subsetting data and recoding response to binary
Receiver_ProfessionalHelp=genderCS[,c("Receiver_ProfessionalHelp", "Supporter_CareerStage_clean", "PI", "All_Gender_clean")]
#recode andclean
Receiver_ProfessionalHelp[(Receiver_ProfessionalHelp$Receiver_ProfessionalHelp %in% c(1,3)),"Receiver_ProfessionalHelp"]=0
Receiver_ProfessionalHelp[(Receiver_ProfessionalHelp$Receiver_ProfessionalHelp %in% c(2)),"Receiver_ProfessionalHelp"]=1
Receiver_ProfessionalHelp=Receiver_ProfessionalHelp[Receiver_ProfessionalHelp$Receiver_ProfessionalHelp %in% seq(0,1), ] #remove NAs and some choices
dim(Receiver_ProfessionalHelp)[1]## [1] 873#LR
summary(glm(Receiver_ProfessionalHelp~PI, data=Receiver_ProfessionalHelp, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Receiver_ProfessionalHelp ~ PI, family = "binomial",
## data = Receiver_ProfessionalHelp)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.9967 -0.7621 -0.7621 1.3696 1.6602
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.08780 0.09287 -11.713 < 2e-16 ***
## PIGroup leaders 0.64668 0.15778 4.098 4.16e-05 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1056.4 on 872 degrees of freedom
## Residual deviance: 1039.8 on 871 degrees of freedom
## AIC: 1043.8
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Receiver_ProfessionalHelp~PI, data=Receiver_ProfessionalHelp, family="binomial"))) #OR (PhD as baseline)## (Intercept) PIGroup leaders
## 0.3369565 1.9091843exp(confint(glm(Receiver_ProfessionalHelp~PI, data=Receiver_ProfessionalHelp, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.280071 0.4031764
## PIGroup leaders 1.400248 2.6003179### Professional support for the receiver x gender graph [NOT REPORTED]Receiver_ProfessionalHelp[Receiver_ProfessionalHelp$Receiver_ProfessionalHelp==0, "Receiver_ProfessionalHelp"]="No"
Receiver_ProfessionalHelp[Receiver_ProfessionalHelp$Receiver_ProfessionalHelp==1, "Receiver_ProfessionalHelp"]="Yes"
Receiver_ProfessionalHelp$Receiver_ProfessionalHelp=factor(Receiver_ProfessionalHelp$Receiver_ProfessionalHelp)
#ImpactPersonal$Impact_Personal=factor(ImpactPersonal$Impact_Personal, levels=c("No","Yes"))
graphdata =Receiver_ProfessionalHelp %>%
group_by(All_Gender_clean, Receiver_ProfessionalHelp) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'All_Gender_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/ReceiverProfessionalHelp_gender.eps", width=1000, height=578)
ggplot(graphdata, aes(x=All_Gender_clean, y=perc, fill=Receiver_ProfessionalHelp)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=20))+
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="During the time I was supporting them, the person received at least some professional help") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=c(eBlue, eGreen))
dev.off()## null device
## 1Professional support for the receiver career graph
graphdata = Receiver_ProfessionalHelp %>%
group_by(Supporter_CareerStage_clean, Receiver_ProfessionalHelp) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'Supporter_CareerStage_clean' (override with `.groups` argument)### Professional support for the receiver, PhDs& postdocs vs PIs graph (New graph 3.2.1?) graphdata = Receiver_ProfessionalHelp %>%
group_by(PI, Receiver_ProfessionalHelp) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'PI' (override with `.groups` argument)# This will save the image to your local code folder
ggplot(graphdata, aes(x=PI, y=perc, fill=Receiver_ProfessionalHelp)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=12), axis.title.x = element_text(size = 8), axis.title.y = element_text(size = 16), legend.text = element_text(size=7))+
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="During the time I was supporting them, the person received at least some professional help") +
guides(fill=guide_legend(reverse=TRUE)) +
scale_x_discrete(limits = rev(levels(graphdata$PI))) +
coord_flip() +
scale_fill_manual(name="", values=c(eBlue, eGreen))
ggsave(dpi=1000, "3-2-1.png", limitsize = FALSE)## Saving 7 x 5 in imagedev.off()## null device
## 1Q17a Comfortable_withMH (Seciont 6, Panel 1)
Model: ordinal logsitic regression
#Subset
ComfortablewithMH=genderCS[,c("Comfortable_withMH", "All_Gender_clean", "Supporter_CareerStage_clean")]
#clean
ComfortablewithMH=ComfortablewithMH[(ComfortablewithMH$Comfortable_withMH %in% seq(1:5)),]
dim(ComfortablewithMH)[1]## [1] 1246#factorise
ComfortablewithMH$Comfortable_withMH=as.factor(ComfortablewithMH$Comfortable_withMH)
table(ComfortablewithMH$Comfortable_withMH)##
## 1 2 3 4 5
## 52 173 113 391 517Comfortable_withMH x gender
exp(coef(polr(Comfortable_withMH~All_Gender_clean, data=ComfortablewithMH, Hess=TRUE, method="logistic"))) #OR (men as baseline)## All_Gender_cleanWomen
## 1.276734exp(confint(polr(Comfortable_withMH~All_Gender_clean, data=ComfortablewithMH, Hess=TRUE, method="logistic"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## 1.034486 1.575691kruskal.test(Comfortable_withMH ~ All_Gender_clean, data=ComfortablewithMH) #effect of gender on response##
## Kruskal-Wallis rank sum test
##
## data: Comfortable_withMH by All_Gender_clean
## Kruskal-Wallis chi-squared = 5.1423, df = 1, p-value = 0.02335Comfortable_withMH x CS
exp(coef(polr(Comfortable_withMH~Supporter_CareerStage_clean, data=ComfortablewithMH, Hess=TRUE, method="logistic"))) #OR (PhD as baseline)## Supporter_CareerStage_cleanPostdocs
## 0.9371691
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 0.6524247
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 0.6759996
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 0.8628607exp(confint(polr(Comfortable_withMH~Supporter_CareerStage_clean, data=ComfortablewithMH, Hess=TRUE, method="logistic"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## Supporter_CareerStage_cleanPostdocs 0.7318633 1.2006074
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.4719229 0.9027507
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.4459509 1.0276223
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.5986961 1.2469950kruskal.test(Comfortable_withMH ~ Supporter_CareerStage_clean, data=ComfortablewithMH) #effect of CS on baseline##
## Kruskal-Wallis rank sum test
##
## data: Comfortable_withMH by Supporter_CareerStage_clean
## Kruskal-Wallis chi-squared = 8.8764, df = 4, p-value = 0.06426Comfortable_withMH x gender graph (Graph 6.1.3)
levels(ComfortablewithMH$Comfortable_withMH)=c("Strongly disagree", "Somewhat disagree", "Neither agree or disagree", "Somewhat agree", "Strongly agree")
ComfortablewithMH$Comfortable_withMH=factor(ComfortablewithMH$Comfortable_withMH, levels=c("Strongly agree", "Somewhat agree", "Neither agree or disagree", "Somewhat disagree", "Strongly disagree"))
graphdata = ComfortablewithMH %>%
group_by(All_Gender_clean, Comfortable_withMH) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'All_Gender_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/ComfortablewithMH_gender.eps", width=1000, height=578)
ggplot(graphdata, aes(x=rev(All_Gender_clean), y=rev(perc), fill=rev(Comfortable_withMH))) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=12), axis.title.x = element_text(size = 8), axis.title.y = element_text(size = 16), legend.position="bottom", legend.text = element_text(size=7))+
geom_text(aes(label=round(rev(perc), digit=1)), size=3.5, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="I was comfortable discussing mental health problems") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=ePalette)
ggsave(dpi=1000, "6-1-3.png", limitsize = FALSE)## Saving 7 x 5 in imagedev.off()## null device
## 1Comfortable_withMH x CS graph
graphdata = ComfortablewithMH %>%
group_by(Supporter_CareerStage_clean, Comfortable_withMH) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'Supporter_CareerStage_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/ComfortablewithMH_CS.eps", width=1000, height=578)
ggplot(graphdata, aes(x=Supporter_CareerStage_clean, y=perc, fill=Comfortable_withMH)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=20))+
scale_x_discrete(limits = rev(levels(ComfortablewithMH$Supporter_CareerStage_clean))) +
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="I was comfortable discussing mental health problems") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=ePalette)
dev.off()## null device
## 1Q17b Confident_withMH (Section 6, Panel 1)
Model: ordinal logsitic regression
#Subset
ConfidentwithMH=genderCS[,c("Confident_withMH", "All_Gender_clean", "Supporter_CareerStage_clean")]
#clean
ConfidentwithMH=ConfidentwithMH[(ConfidentwithMH$Confident_withMH %in% seq(1:5)),]
dim(ConfidentwithMH)[1]## [1] 1251#factorise
ConfidentwithMH$Confident_withMH=as.factor(ConfidentwithMH$Confident_withMH)
table(ConfidentwithMH$Confident_withMH)##
## 1 2 3 4 5
## 100 292 219 478 162Confident_withMH x gender
exp(coef(polr(Confident_withMH~All_Gender_clean, data=ConfidentwithMH, Hess=TRUE, method="logistic"))) #OR (men as baseline)## All_Gender_cleanWomen
## 0.9797103exp(confint(polr(Confident_withMH~All_Gender_clean, data=ConfidentwithMH, Hess=TRUE, method="logistic"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## 0.7966081 1.2046537kruskal.test(Confident_withMH ~ All_Gender_clean, data=ConfidentwithMH) #effect of gender on response##
## Kruskal-Wallis rank sum test
##
## data: Confident_withMH by All_Gender_clean
## Kruskal-Wallis chi-squared = 0.037494, df = 1, p-value = 0.8465Confident_withMH x CS
exp(coef(polr(Confident_withMH~Supporter_CareerStage_clean, data=ConfidentwithMH, Hess=TRUE, method="logistic"))) #OR (PhD as baseline)## Supporter_CareerStage_cleanPostdocs
## 0.9544268
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 0.6795118
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 0.8960816
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 1.3411865exp(confint(polr(Confident_withMH~Supporter_CareerStage_clean, data=ConfidentwithMH, Hess=TRUE, method="logistic"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## Supporter_CareerStage_cleanPostdocs 0.7495507 1.2154293
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.4916361 0.9389662
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.5987433 1.3439043
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.9348656 1.9273882kruskal.test(Confident_withMH ~ Supporter_CareerStage_clean, data=ConfidentwithMH) #effect of CS on response##
## Kruskal-Wallis rank sum test
##
## data: Confident_withMH by Supporter_CareerStage_clean
## Kruskal-Wallis chi-squared = 9.9089, df = 4, p-value = 0.04199## multiway kruskal wallis
## this requries the package 'gtools'
## CSpairs=combinations(n=5, r=2, v=levels(ConfidentwithMH$Supporter_CareerStage_clean), repeats.allowed = F)
# Compare between early PIs and other CSs only
## this requires the package "MTHdiscrete"
CSpairs=cbind(rep("Group leaders (<5yr)", 4), levels(ImpactDraining$Supporter_CareerStage_clean)[c(1,2,4,5)])
ConfidentwithMHCSKW=data.frame(CS1=c(), CS2=c(), X2=c(), N=c(), df=c(), pVal=c())
for (i in 1:dim(CSpairs)[1]) {
comparedData=ConfidentwithMH[(ConfidentwithMH$Supporter_CareerStage_clean==CSpairs[i,1]|ConfidentwithMH$Supporter_CareerStage_clean==CSpairs[i,2]),]
KW=kruskal.test(Confident_withMH ~ Supporter_CareerStage_clean, data=comparedData) #effect of CS on response
ConfidentwithMHCSKW=rbind(ConfidentwithMHCSKW, data.frame(CS1=CSpairs[i,1], CS2=CSpairs[i,2], X2=as.numeric(KW$statistic), N=dim(comparedData)[1], df=as.numeric(KW$parameter), pVal=as.numeric(KW$p.value)))
}
ConfidentwithMHCSKW$adjpVal=Sidak.p.adjust(ConfidentwithMHCSKW$pVal)
ConfidentwithMHCSKW## CS1 CS2 X2 N df pVal
## 1 Group leaders (<5yr) PhD students 5.384519 686 1 0.020316188
## 2 Group leaders (<5yr) Postdocs 3.810071 511 1 0.050945296
## 3 Group leaders (<5yr) Group leaders (5-10yr) 1.312923 245 1 0.251866056
## 4 Group leaders (<5yr) Group leaders (>10yr) 8.892148 274 1 0.002863996
## adjpVal
## 1 0.07882164
## 2 0.18873081
## 3 0.68673099
## 4 0.01140686Confident_with MH x early PI vs others
Make sure you run the corresponding “impact draining” chunk (under heading “### Impact Draining x early PI vs others (PhD/postdoc/midPI/latePI)”) before running this (need to have early PI variable in genderCS)
#redo subsetting
ConfidentwithMHPI=genderCS[,c("Confident_withMH", "All_Gender_clean", "earlyPI")]
summary(ConfidentwithMHPI) #there are some NAs## Confident_withMH All_Gender_clean earlyPI
## Min. :1.000 Men :474 Group leaders (<5yr): 155
## 1st Qu.:2.000 Women:781 Other career stages :1100
## Median :4.000
## Mean :3.252
## 3rd Qu.:4.000
## Max. :6.000
## NA's :2#clean
ConfidentwithMHPI=ConfidentwithMHPI[(ConfidentwithMHPI$Confident_withMH %in% seq(1:5)),]
dim(ConfidentwithMHPI)[1] #this gives N## [1] 1251#factorise
ConfidentwithMHPI$Confident_withMH=as.factor(ConfidentwithMHPI$Confident_withMH)
#kruskal wallis
kruskal.test(Confident_withMH ~ earlyPI, data=ConfidentwithMHPI)##
## Kruskal-Wallis rank sum test
##
## data: Confident_withMH by earlyPI
## Kruskal-Wallis chi-squared = 6.3366, df = 1, p-value = 0.01183Confident_withMH x gender graph
levels(ConfidentwithMH$Confident_withMH)=c("Strongly disagree", "Somewhat disagree", "Neither agree or disagree", "Somewhat agree", "Strongly agree")
ConfidentwithMH$Confident_withMH=factor(ConfidentwithMH$Confident_withMH, levels=c("Strongly agree", "Somewhat agree", "Neither agree or disagree", "Somewhat disagree", "Strongly disagree"))
graphdata = ConfidentwithMH %>%
group_by(All_Gender_clean, Confident_withMH) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'All_Gender_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/ConfidentwithMH_gender.eps", width=1000, height=578)
ggplot(graphdata, aes(x=All_Gender_clean, y=perc, fill=Confident_withMH)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=20))+
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="I was confident in my ability to do or say the right thing") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=ePalette)
dev.off()## null device
## 1Confident_withMH x CS graph (Graph 6.1.4)
graphdata = ConfidentwithMH %>%
group_by(Supporter_CareerStage_clean, Confident_withMH) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'Supporter_CareerStage_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/ConfidentwithMH_CS.eps", width=1000, height=578)
ggplot(graphdata, aes(x=Supporter_CareerStage_clean, y=perc, fill=Confident_withMH)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=12), axis.title.x = element_text(size = 8), axis.title.y = element_text(size = 16), legend.position="bottom", legend.text = element_text(size=7))+
scale_x_discrete(limits = rev(levels(ConfidentwithMH$Supporter_CareerStage_clean))) +
geom_text(aes(label=round(perc, digit=1)), size=3, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="I was confident in my ability to do or say the right thing") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=ePalette)
ggsave(dpi=1000, "6-1-4 GOOD.png", limitsize = FALSE)## Saving 7 x 5 in imagedev.off()## null device
## 1Q18a Impact Rewarding (Seciont 4, Panel 1)
Model: ordinal logistic regression
#subsetting
ImpactRewarding=genderCS[,c("Impact_Rewarding", "All_Gender_clean", "Supporter_CareerStage_clean")]
#clean
ImpactRewarding=ImpactRewarding[(ImpactRewarding$Impact_Rewarding %in% seq(1:5)),]
dim(ImpactRewarding)[1]## [1] 1243#factorise
ImpactRewarding$Impact_Rewarding=as.factor(ImpactRewarding$Impact_Rewarding)
table(ImpactRewarding$Impact_Rewarding)##
## 1 2 3 4 5
## 30 98 270 445 400gender
exp(coef(polr(Impact_Rewarding~All_Gender_clean, data=ImpactRewarding, Hess=TRUE, method="logistic"))) #OR (men as baseline)## All_Gender_cleanWomen
## 0.952375kruskal.test(Impact_Rewarding ~ All_Gender_clean, data=ImpactRewarding) #effect of gender on response##
## Kruskal-Wallis rank sum test
##
## data: Impact_Rewarding by All_Gender_clean
## Kruskal-Wallis chi-squared = 0.2094, df = 1, p-value = 0.6472CS
exp(coef(polr(Impact_Rewarding~Supporter_CareerStage_clean, data=ImpactRewarding, Hess=TRUE, method="logistic"))) #OR (men as baseline)## Supporter_CareerStage_cleanPostdocs
## 1.0672600
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 0.6867169
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 0.7174809
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 0.9056449kruskal.test(Impact_Rewarding ~ Supporter_CareerStage_clean, data=ImpactRewarding) #effect of CS on response##
## Kruskal-Wallis rank sum test
##
## data: Impact_Rewarding by Supporter_CareerStage_clean
## Kruskal-Wallis chi-squared = 8.9339, df = 4, p-value = 0.06277Q18b Impact Positive Difference
Model: ordinal logistic regression
#Subsetting
ImpactPositiveDifference=genderCS[,c("Impact_PositiveDifference", "All_Gender_clean", "Supporter_CareerStage_clean")]
#clean
ImpactPositiveDifference=ImpactPositiveDifference[(ImpactPositiveDifference$Impact_PositiveDifference %in% seq(1:5)),]
dim(ImpactPositiveDifference)[1]## [1] 1241#factorise
ImpactPositiveDifference$Impact_PositiveDifference=as.factor(ImpactPositiveDifference$Impact_PositiveDifference)
table(ImpactPositiveDifference$Impact_PositiveDifference)##
## 1 2 3 4 5
## 14 33 234 548 412Impact Positive Difference x gender
exp(coef(polr(Impact_PositiveDifference~All_Gender_clean, data=ImpactPositiveDifference, Hess=TRUE, method="logistic"))) #OR (men as baseline)## All_Gender_cleanWomen
## 0.9596224kruskal.test(Impact_PositiveDifference ~ All_Gender_clean, data=ImpactPositiveDifference) #effect of gender on response##
## Kruskal-Wallis rank sum test
##
## data: Impact_PositiveDifference by All_Gender_clean
## Kruskal-Wallis chi-squared = 0.14425, df = 1, p-value = 0.7041Impact Positive Difference x CS
exp(coef(polr(Impact_PositiveDifference~Supporter_CareerStage_clean, data=ImpactPositiveDifference, Hess=TRUE, method="logistic"))) #OR PhD as baseline## Supporter_CareerStage_cleanPostdocs
## 1.1737603
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 0.9169238
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 1.2814621
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 1.1194034kruskal.test(Impact_PositiveDifference ~ Supporter_CareerStage_clean, data=ImpactPositiveDifference) #effect of CS on response##
## Kruskal-Wallis rank sum test
##
## data: Impact_PositiveDifference by Supporter_CareerStage_clean
## Kruskal-Wallis chi-squared = 3.582, df = 4, p-value = 0.4655Q20d Support Receiver (Section 4, Panel 1)
Model: ordinal logistic regression
#subset
SupportReceiver=genderCS[,c("Support_Receiver", "All_Gender_clean", "Supporter_CareerStage_clean")]
#clean
SupportReceiver=SupportReceiver[(SupportReceiver$Support_Receiver %in% seq(1:5)),]
dim(SupportReceiver)[1]## [1] 1227#factorise
SupportReceiver$Support_Receiver=as.factor(SupportReceiver$Support_Receiver)
table(SupportReceiver$Support_Receiver)##
## 1 2 3 4 5
## 27 58 87 362 693gender
exp(coef(polr(Support_Receiver~All_Gender_clean, data=SupportReceiver, Hess=TRUE, method="logistic"))) #OR (men as baseline)## All_Gender_cleanWomen
## 1.016317kruskal.test(Support_Receiver ~ All_Gender_clean, data=SupportReceiver) # effect of gender on response##
## Kruskal-Wallis rank sum test
##
## data: Support_Receiver by All_Gender_clean
## Kruskal-Wallis chi-squared = 0.020246, df = 1, p-value = 0.8869CS
exp(coef(polr(Support_Receiver~Supporter_CareerStage_clean, data=SupportReceiver, Hess=TRUE, method="logistic"))) #OR (PhD as baseline)## Supporter_CareerStage_cleanPostdocs
## 1.2207761
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 0.6228073
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 0.8558125
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 0.8469562kruskal.test(Support_Receiver ~ Supporter_CareerStage_clean, data=SupportReceiver) #effect of CS on response##
## Kruskal-Wallis rank sum test
##
## data: Support_Receiver by Supporter_CareerStage_clean
## Kruskal-Wallis chi-squared = 14.34, df = 4, p-value = 0.006285Q18c Impact Beacon (Section 4, Panel 1)
Model: ordinal logistic regression
#subset
ImpactBeacon=genderCS[,c("Impact_Beacon", "All_Gender_clean", "Supporter_CareerStage_clean")]
#clean
ImpactBeacon=ImpactBeacon[(ImpactBeacon$Impact_Beacon %in% seq(1:5)),]
dim(ImpactBeacon)[1]## [1] 1161#factorise
ImpactBeacon$Impact_Beacon=as.factor(ImpactBeacon$Impact_Beacon)
table(ImpactBeacon$Impact_Beacon)##
## 1 2 3 4 5
## 118 137 334 350 222Impact Beacon x gender
exp(coef(polr(Impact_Beacon~All_Gender_clean, data=ImpactBeacon, Hess=TRUE, method="logistic"))) #OR (men as baseline)## All_Gender_cleanWomen
## 1.426692exp(confint(polr(Impact_Beacon~All_Gender_clean, data=ImpactBeacon, Hess=TRUE, method="logistic"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## 1.153487 1.765366kruskal.test(Impact_Beacon ~ All_Gender_clean, data=ImpactBeacon) #effect of gender on response##
## Kruskal-Wallis rank sum test
##
## data: Impact_Beacon by All_Gender_clean
## Kruskal-Wallis chi-squared = 10.694, df = 1, p-value = 0.001075Impact Beacon x CS
exp(coef(polr(Impact_Beacon~Supporter_CareerStage_clean, data=ImpactBeacon, Hess=TRUE, method="logistic"))) #OR (PhD as baseline)## Supporter_CareerStage_cleanPostdocs
## 1.124024
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 1.403487
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 1.222366
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 1.545442kruskal.test(Impact_Beacon ~ Supporter_CareerStage_clean, data=ImpactBeacon) #effect of CS on response##
## Kruskal-Wallis rank sum test
##
## data: Impact_Beacon by Supporter_CareerStage_clean
## Kruskal-Wallis chi-squared = 7.7745, df = 4, p-value = 0.1002Q20a Support Institutions (Section 5, Panel 2)
Model: ordinal logistic regression
SupportInstitution=genderCS[,c("Support_Institution", "All_Gender_clean", "Supporter_CareerStage_clean")]
#clean
SupportInstitution=SupportInstitution[(SupportInstitution$Support_Institution %in% seq(1:5)),]
dim(SupportInstitution)[1]## [1] 1104#factorise
SupportInstitution$Support_Institution=as.factor(SupportInstitution$Support_Institution)
table(SupportInstitution$Support_Institution)##
## 1 2 3 4 5
## 436 245 255 124 44Support Institutions x gender
exp(coef(polr(Support_Institution~All_Gender_clean, data=SupportInstitution, Hess=TRUE, method="logistic"))) #OR (men as baseline)## All_Gender_cleanWomen
## 0.7659521exp(confint(polr(Support_Institution~All_Gender_clean, data=SupportInstitution, Hess=TRUE, method="logistic"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## 0.6145262 0.9546073kruskal.test(Support_Institution ~ All_Gender_clean, data=SupportInstitution) #effect of gender on response##
## Kruskal-Wallis rank sum test
##
## data: Support_Institution by All_Gender_clean
## Kruskal-Wallis chi-squared = 5.6368, df = 1, p-value = 0.01759Support Institutions x CS
exp(coef(polr(Support_Institution~Supporter_CareerStage_clean, data=SupportInstitution, Hess=TRUE, method="logistic"))) # OR (PhD as baseline)## Supporter_CareerStage_cleanPostdocs
## 1.019736
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 2.624482
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 2.016627
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 5.248899exp(confint(polr(Support_Institution~Supporter_CareerStage_clean, data=SupportInstitution, Hess=TRUE, method="logistic"))) #CI ## Waiting for profiling to be done...## 2.5 % 97.5 %
## Supporter_CareerStage_cleanPostdocs 0.7818768 1.329219
## Supporter_CareerStage_cleanGroup leaders (<5yr) 1.8744130 3.680966
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 1.3235401 3.072055
## Supporter_CareerStage_cleanGroup leaders (>10yr) 3.5424203 7.802391kruskal.test(Support_Institution ~ Supporter_CareerStage_clean, data=SupportInstitution) #effect on CS on response##
## Kruskal-Wallis rank sum test
##
## data: Support_Institution by Supporter_CareerStage_clean
## Kruskal-Wallis chi-squared = 92.751, df = 4, p-value < 2.2e-16Support Institution x gender graph (Graph 5.2.4)
levels(SupportInstitution$Support_Institution)=c("Strongly disagree", "Somewhat disagree", "Neither agree or disagree", "Somewhat agree", "Strongly agree")
SupportInstitution$Support_Institution=factor(SupportInstitution$Support_Institution, levels=c("Strongly agree", "Somewhat agree", "Neither agree or disagree", "Somewhat disagree", "Strongly disagree"))
graphdata = SupportInstitution %>%
group_by(All_Gender_clean, Support_Institution) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'All_Gender_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/SupportInstitution_gender.eps", width=1000, height=578)
ggplot(graphdata, aes(x=All_Gender_clean, y=perc, fill=Support_Institution)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=12), axis.title.x = element_text(size = 8), axis.title.y = element_text(size = 16), legend.position="bottom", legend.text = element_text(size=7))+
geom_text(aes(label=round(perc, digit=1)), size=3, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="I feel supported or valued by my institution \n(e.g. managers, department) for the help I was providing") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=ePalette)
ggsave(dpi=1000, "5-2-4.png", limitsize = FALSE)## Saving 7 x 5 in imagedev.off()## null device
## 1Support instituions x CS graph (Graph 5.2.5)
graphdata = SupportInstitution %>%
group_by(Supporter_CareerStage_clean, Support_Institution) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'Supporter_CareerStage_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/SupportInstitution_CS.eps", width=1000, height=578)
ggplot(graphdata, aes(x=Supporter_CareerStage_clean, y=perc, fill=Support_Institution)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=12), axis.title.x = element_text(size = 8), axis.title.y = element_text(size = 16), legend.position="bottom", legend.text = element_text(size=7))+
scale_x_discrete(limits = rev(levels(SupportInstitution$Supporter_CareerStage_clean))) +
geom_text(aes(label=round(perc, digit=1)), size=3, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="I feel supported or valued by my institution (e.g. managers, department) for the help I was providing") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=ePalette)
ggsave(dpi=1000, "5-2-5.png", limitsize = FALSE)## Saving 7 x 5 in imagedev.off()## null device
## 1Q20b Support colleagues (Section 5, Panel 2)
Model: ordinal logistic regression
#subset
SupportColleagues=genderCS[,c("Support_Colleagues", "All_Gender_clean", "Supporter_CareerStage_clean")]
#clean
SupportColleagues=SupportColleagues[(SupportColleagues$Support_Colleagues %in% seq(1:5)),]
dim(SupportColleagues)[1]## [1] 1111#factorise
SupportColleagues$Support_Colleagues=as.factor(SupportColleagues$Support_Colleagues)
table(SupportColleagues$Support_Colleagues)##
## 1 2 3 4 5
## 151 181 293 339 147Support colleagues x gender
exp(coef(polr(Support_Colleagues~All_Gender_clean, data=SupportColleagues, Hess=TRUE, method="logistic"))) #OR (men as baseline)## All_Gender_cleanWomen
## 0.8551606exp(confint(polr(Support_Colleagues~All_Gender_clean, data=SupportColleagues, Hess=TRUE, method="logistic"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## 0.6887148 1.0614565kruskal.test(Support_Colleagues ~ All_Gender_clean, data=SupportColleagues) #effect of gender on response##
## Kruskal-Wallis rank sum test
##
## data: Support_Colleagues by All_Gender_clean
## Kruskal-Wallis chi-squared = 2.0173, df = 1, p-value = 0.1555Support colleagues x CS
exp(coef(polr(Support_Colleagues~Supporter_CareerStage_clean, data=SupportColleagues, Hess=TRUE, method="logistic"))) #OR (PhD as baseline)## Supporter_CareerStage_cleanPostdocs
## 1.040821
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 1.150612
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 1.071028
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 2.401214exp(confint(polr(Support_Colleagues~Supporter_CareerStage_clean, data=SupportColleagues, Hess=TRUE, method="logistic"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## Supporter_CareerStage_cleanPostdocs 0.8063473 1.343653
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.8295246 1.597604
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.6971757 1.647557
## Supporter_CareerStage_cleanGroup leaders (>10yr) 1.6387705 3.530297kruskal.test(Support_Colleagues ~ Supporter_CareerStage_clean, data=SupportColleagues) #effect of CS on response##
## Kruskal-Wallis rank sum test
##
## data: Support_Colleagues by Supporter_CareerStage_clean
## Kruskal-Wallis chi-squared = 20.333, df = 4, p-value = 0.0004293Support colleagues x gender graph
levels(SupportColleagues$Support_Colleagues)=c("Strongly disagree", "Somewhat disagree", "Neither agree or disagree", "Somewhat agree", "Strongly agree")
SupportColleagues$Support_Colleagues=factor(SupportColleagues$Support_Colleagues, levels=c("Strongly agree", "Somewhat agree", "Neither agree or disagree", "Somewhat disagree", "Strongly disagree"))
graphdata = SupportColleagues %>%
group_by(All_Gender_clean, Support_Colleagues) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'All_Gender_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/SupportColleagues_gender.eps", width=1000, height=578)
ggplot(graphdata, aes(x=All_Gender_clean, y=perc, fill=Support_Colleagues)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=20))+
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="I felt supported or valued by my colleagues (e.g. other members of the lab) for the help I was providing") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=ePalette)
dev.off()## null device
## 1Support colleagues x CS graph
graphdata = SupportColleagues %>%
group_by(Supporter_CareerStage_clean, Support_Colleagues) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'Supporter_CareerStage_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/SupportColleagues_CS.eps", width=1000, height=578)
ggplot(graphdata, aes(x=Supporter_CareerStage_clean, y=perc, fill=Support_Colleagues)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=20))+
scale_x_discrete(limits = rev(levels(SupportColleagues$Supporter_CareerStage_clean))) +
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="I felt supported or valued by my colleagues (e.g. other members of the lab) for the help I was providing") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=ePalette)
dev.off()## null device
## 1Q20c Support Personal (Section 5, Panel 2)
Model: ordinal logistic regression
#subset
SupportPersonal=genderCS[,c("Support_Personal", "All_Gender_clean", "Supporter_CareerStage_clean")]
#clean
SupportPersonal=SupportPersonal[(SupportPersonal$Support_Personal %in% seq(1:5)),]
dim(SupportPersonal)[1]## [1] 1125#factorise
SupportPersonal$Support_Personal=as.factor(SupportPersonal$Support_Personal)
table(SupportPersonal$Support_Personal)##
## 1 2 3 4 5
## 51 94 280 409 291Support Personal x gender
exp(coef(polr(Support_Personal~All_Gender_clean, data=SupportPersonal, Hess=TRUE, method="logistic"))) #OR (men as baseline)## All_Gender_cleanWomen
## 0.9921087kruskal.test(Support_Personal ~ All_Gender_clean, data=SupportPersonal) #effect of gender on response##
## Kruskal-Wallis rank sum test
##
## data: Support_Personal by All_Gender_clean
## Kruskal-Wallis chi-squared = 0.0053151, df = 1, p-value = 0.9419Support Personal x CS
exp(coef(polr(Support_Personal~Supporter_CareerStage_clean, data=SupportPersonal, Hess=TRUE, method="logistic"))) #OR (PhD as baseline)## Supporter_CareerStage_cleanPostdocs
## 1.0396756
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 0.8833687
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 0.6973069
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 1.7224356exp(confint(polr(Support_Personal~Supporter_CareerStage_clean, data=SupportPersonal, Hess=TRUE, method="logistic"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## Supporter_CareerStage_cleanPostdocs 0.8047695 1.343481
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.6317658 1.235267
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.4618079 1.053534
## Supporter_CareerStage_cleanGroup leaders (>10yr) 1.1663603 2.552038kruskal.test(Support_Personal ~ Supporter_CareerStage_clean, data=SupportPersonal) #effect of CS on response##
## Kruskal-Wallis rank sum test
##
## data: Support_Personal by Supporter_CareerStage_clean
## Kruskal-Wallis chi-squared = 13.295, df = 4, p-value = 0.00992Support Personal x gender graph
levels(SupportPersonal$Support_Personal)=c("Strongly disagree", "Somewhat disagree", "Neither agree or disagree", "Somewhat agree", "Strongly agree")
SupportPersonal$Support_Personal=factor(SupportPersonal$Support_Personal, levels=c("Strongly agree", "Somewhat agree", "Neither agree or disagree", "Somewhat disagree", "Strongly disagree"))
# This will save the image to your local code folder
#eps("images/SupportPersonal_gender.eps", width=1000, height=578)
graphdata = SupportPersonal %>%
group_by(All_Gender_clean, Support_Personal) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'All_Gender_clean' (override with `.groups` argument)ggplot(graphdata, aes(x=All_Gender_clean, y=perc, fill=Support_Personal)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=20))+
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="I felt supported or valued by my friends, partners and family outside of academia for the help I was providing") +
coord_flip() +
scale_fill_manual(name="", values=ePalette)
dev.off()## null device
## 1Support Personal x CS graph
graphdata = SupportPersonal %>%
group_by(Supporter_CareerStage_clean, Support_Personal) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'Supporter_CareerStage_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/SupportPersonal_CS.eps", width=1000, height=578)
ggplot(graphdata, aes(x=Supporter_CareerStage_clean, y=perc, fill=Support_Personal)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=20))+
scale_x_discrete(limits = rev(levels(SupportPersonal$Supporter_CareerStage_clean))) +
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="I felt supported or valued by my friends, partners and family outside of academia for the help I was providing") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=ePalette)
dev.off()## null device
## 1Q21 Need and access to emotional support (Section 5, Panel 1)
Filter down to people who picked at least 1 of the 6 meaningful options (i.e. not PNTA or unsure):
- Support_Emotional_No
- Support_Emotional_NoFind
- Support_Emotional_Institution
- Support_Emotional_Colleagues
- Support_Emotional_Personal
- Support_Emotional_Other_Dummy
Support_Emotional=genderCS[!is.na(genderCS$Support_Emotional_No==1 | genderCS$Support_Emotional_NoFind==1 | genderCS$Support_Emotional_Institution==1 | genderCS$Support_Emotional_Colleagues==1 | genderCS$Support_Emotional_Personal==1 | genderCS$Support_Emotional_Other_Dummy==1),]
Support_Emotional=Support_Emotional[,c("Support_Emotional_No", "Support_Emotional_NoFind", "Support_Emotional_Institution","Support_Emotional_Colleagues" , "Support_Emotional_Personal", "Supporter_CareerStage_clean","All_Gender_clean")]
dim(Support_Emotional)[1]## [1] 1184#fill all na with 0
Support_Emotional[is.na(Support_Emotional)]=0Support_Emotional_No (Yes/No) x Gender // xCS
Dependent variable -
- 0 - didn’t select this option
- 1 - did select this option - i.e. DON’T NEED EMOTIONAL SUPPORT
Model: Factorial logistical regression
#LR gender
summary(glm(Support_Emotional_No~All_Gender_clean, data=Support_Emotional, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Support_Emotional_No ~ All_Gender_clean, family = "binomial",
## data = Support_Emotional)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.9565 -0.6881 -0.6881 1.4157 1.7645
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.54460 0.09846 -5.531 3.18e-08 ***
## All_Gender_cleanWomen -0.77544 0.13348 -5.810 6.27e-09 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1379.8 on 1183 degrees of freedom
## Residual deviance: 1346.0 on 1182 degrees of freedom
## AIC: 1350
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Support_Emotional_No~All_Gender_clean, data=Support_Emotional, family="binomial"))) #OR (men as baseline)## (Intercept) All_Gender_cleanWomen
## 0.5800712 0.4605009exp(confint(glm(Support_Emotional_No~All_Gender_clean, data=Support_Emotional, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.4774083 0.7024838
## All_Gender_cleanWomen 0.3542797 0.5980024#LR CS
summary(glm(Support_Emotional_No~Supporter_CareerStage_clean, data=Support_Emotional, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Support_Emotional_No ~ Supporter_CareerStage_clean,
## family = "binomial", data = Support_Emotional)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.1338 -0.7976 -0.7197 1.2217 1.7246
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) -1.21865 0.10614 -11.481
## Supporter_CareerStage_cleanPostdocs 0.23645 0.16239 1.456
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.01223 0.22454 -0.054
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.35999 0.26118 1.378
## Supporter_CareerStage_cleanGroup leaders (>10yr) 1.11511 0.21411 5.208
## Pr(>|z|)
## (Intercept) < 2e-16 ***
## Supporter_CareerStage_cleanPostdocs 0.145
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.957
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.168
## Supporter_CareerStage_cleanGroup leaders (>10yr) 1.91e-07 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1379.8 on 1183 degrees of freedom
## Residual deviance: 1351.4 on 1179 degrees of freedom
## AIC: 1361.4
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Support_Emotional_No~Supporter_CareerStage_clean, data=Support_Emotional, family="binomial"))) #OR (PhDs as baseline)## (Intercept)
## 0.2956298
## Supporter_CareerStage_cleanPostdocs
## 1.2667382
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 0.9878415
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 1.4333088
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 3.0498931exp(confint(glm(Support_Emotional_No~Supporter_CareerStage_clean, data=Support_Emotional, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.2391124 0.3626434
## Supporter_CareerStage_cleanPostdocs 0.9203645 1.7405027
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.6291522 1.5207163
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.8473504 2.3679512
## Supporter_CareerStage_cleanGroup leaders (>10yr) 2.0033880 4.6438774#FLR gender-CS interactions
summary(glm(Support_Emotional_No~Supporter_CareerStage_clean*All_Gender_clean, data=Support_Emotional, family="binomial"))##
## Call:
## glm(formula = Support_Emotional_No ~ Supporter_CareerStage_clean *
## All_Gender_clean, family = "binomial", data = Support_Emotional)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.2917 -0.8446 -0.6478 1.0673 1.9728
##
## Coefficients:
## Estimate
## (Intercept) -0.753772
## Supporter_CareerStage_cleanPostdocs 0.047202
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.084722
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -0.008368
## Supporter_CareerStage_cleanGroup leaders (>10yr) 1.018464
## All_Gender_cleanWomen -0.701018
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.272608
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.421692
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.493757
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen -0.410973
## Std. Error
## (Intercept) 0.175035
## Supporter_CareerStage_cleanPostdocs 0.266454
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.320385
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.367966
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.290166
## All_Gender_cleanWomen 0.221465
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.337458
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.467203
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.528394
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.470801
## z value
## (Intercept) -4.306
## Supporter_CareerStage_cleanPostdocs 0.177
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.264
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -0.023
## Supporter_CareerStage_cleanGroup leaders (>10yr) 3.510
## All_Gender_cleanWomen -3.165
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.808
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.903
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.934
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen -0.873
## Pr(>|z|)
## (Intercept) 1.66e-05
## Supporter_CareerStage_cleanPostdocs 0.859393
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.791442
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.981856
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.000448
## All_Gender_cleanWomen 0.001549
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.419190
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.366744
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.350072
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.382705
##
## (Intercept) ***
## Supporter_CareerStage_cleanPostdocs
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## Supporter_CareerStage_cleanGroup leaders (>10yr) ***
## All_Gender_cleanWomen **
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1379.8 on 1183 degrees of freedom
## Residual deviance: 1323.3 on 1174 degrees of freedom
## AIC: 1343.3
##
## Number of Fisher Scoring iterations: 4Support_Emotional_NoFind x Gender // xCS
As above
#LR gender
summary(glm(Support_Emotional_NoFind~All_Gender_clean, data=Support_Emotional, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Support_Emotional_NoFind ~ All_Gender_clean, family = "binomial",
## data = Support_Emotional)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.7021 -0.7021 -0.6425 -0.6425 1.8327
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.2746 0.1149 -11.098 <2e-16 ***
## All_Gender_cleanWomen -0.1984 0.1487 -1.335 0.182
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1179.9 on 1183 degrees of freedom
## Residual deviance: 1178.2 on 1182 degrees of freedom
## AIC: 1182.2
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Support_Emotional_NoFind~All_Gender_clean, data=Support_Emotional, family="binomial"))) #OR (men as baseline)## (Intercept) All_Gender_cleanWomen
## 0.2795389 0.8200500exp(confint(glm(Support_Emotional_NoFind~All_Gender_clean, data=Support_Emotional, family="binomial")))#CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.2220724 0.3485369
## All_Gender_cleanWomen 0.6134713 1.0992745#LR CS
summary(glm(Support_Emotional_NoFind~Supporter_CareerStage_clean, data=Support_Emotional, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Support_Emotional_NoFind ~ Supporter_CareerStage_clean,
## family = "binomial", data = Support_Emotional)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.7410 -0.7410 -0.5807 -0.5570 1.9697
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) -1.1522 0.1043 -11.049
## Supporter_CareerStage_cleanPostdocs -0.6325 0.1876 -3.371
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.2001 0.2298 -0.871
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -0.1470 0.2856 -0.515
## Supporter_CareerStage_cleanGroup leaders (>10yr) -0.5424 0.2768 -1.959
## Pr(>|z|)
## (Intercept) < 2e-16 ***
## Supporter_CareerStage_cleanPostdocs 0.000748 ***
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.383868
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.606697
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.050096 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1179.9 on 1183 degrees of freedom
## Residual deviance: 1166.3 on 1179 degrees of freedom
## AIC: 1176.3
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Support_Emotional_NoFind~Supporter_CareerStage_clean, data=Support_Emotional, family="binomial"))) #OR (PhD as baseline)## (Intercept)
## 0.3159269
## Supporter_CareerStage_cleanPostdocs
## 0.5312374
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 0.8186093
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 0.8632607
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 0.5813797exp(confint(glm(Support_Emotional_NoFind~Supporter_CareerStage_clean, data=Support_Emotional, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.2565420 0.3862451
## Supporter_CareerStage_cleanPostdocs 0.3651036 0.7628461
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.5148326 1.2709867
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.4809342 1.4823633
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.3287440 0.9787698#FLR gender-CS interactions
summary(glm(Support_Emotional_NoFind~Supporter_CareerStage_clean*All_Gender_clean, data=Support_Emotional, family="binomial"))##
## Call:
## glm(formula = Support_Emotional_NoFind ~ Supporter_CareerStage_clean *
## All_Gender_clean, family = "binomial", data = Support_Emotional)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.7981 -0.7209 -0.6501 -0.4991 2.0710
##
## Coefficients:
## Estimate
## (Intercept) -1.01160
## Supporter_CareerStage_cleanPostdocs -0.39717
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.22054
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.03077
## Supporter_CareerStage_cleanGroup leaders (>10yr) -0.99587
## All_Gender_cleanWomen -0.20342
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -0.40783
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.01135
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -0.55035
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.97413
## Std. Error
## (Intercept) 0.18464
## Supporter_CareerStage_cleanPostdocs 0.30109
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.35546
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.38558
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.40016
## All_Gender_cleanWomen 0.22383
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.38759
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.46858
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.60063
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.56525
## z value
## (Intercept) -5.479
## Supporter_CareerStage_cleanPostdocs -1.319
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.620
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.080
## Supporter_CareerStage_cleanGroup leaders (>10yr) -2.489
## All_Gender_cleanWomen -0.909
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -1.052
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.024
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -0.916
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 1.723
## Pr(>|z|)
## (Intercept) 4.28e-08
## Supporter_CareerStage_cleanPostdocs 0.1871
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.5350
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.9364
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.0128
## All_Gender_cleanWomen 0.3634
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.2927
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.9807
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.3595
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.0848
##
## (Intercept) ***
## Supporter_CareerStage_cleanPostdocs
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## Supporter_CareerStage_cleanGroup leaders (>10yr) *
## All_Gender_cleanWomen
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1179.9 on 1183 degrees of freedom
## Residual deviance: 1157.4 on 1174 degrees of freedom
## AIC: 1177.4
##
## Number of Fisher Scoring iterations: 4Support_Emotional_Institution x Gender // xCS
As above
#LR gender
summary(glm(Support_Emotional_Institution~All_Gender_clean, data=Support_Emotional, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Support_Emotional_Institution ~ All_Gender_clean,
## family = "binomial", data = Support_Emotional)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.3021 -0.3021 -0.3021 -0.2622 2.6030
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -3.3534 0.2627 -12.77 <2e-16 ***
## All_Gender_cleanWomen 0.2889 0.3174 0.91 0.363
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 401.75 on 1183 degrees of freedom
## Residual deviance: 400.90 on 1182 degrees of freedom
## AIC: 404.9
##
## Number of Fisher Scoring iterations: 6#LR CS
summary(glm(Support_Emotional_Institution~Supporter_CareerStage_clean, data=Support_Emotional, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Support_Emotional_Institution ~ Supporter_CareerStage_clean,
## family = "binomial", data = Support_Emotional)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.3381 -0.3381 -0.2649 -0.2202 2.7319
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) -2.8332 0.1945 -14.570
## Supporter_CareerStage_cleanPostdocs -0.8742 0.4073 -2.147
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.7363 0.5430 -1.356
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -0.1625 0.5480 -0.297
## Supporter_CareerStage_cleanGroup leaders (>10yr) -0.4990 0.5447 -0.916
## Pr(>|z|)
## (Intercept) <2e-16 ***
## Supporter_CareerStage_cleanPostdocs 0.0318 *
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.1751
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.7668
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.3597
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 401.75 on 1183 degrees of freedom
## Residual deviance: 395.42 on 1179 degrees of freedom
## AIC: 405.42
##
## Number of Fisher Scoring iterations: 6exp(coef(glm(Support_Emotional_Institution~Supporter_CareerStage_clean, data=Support_Emotional, family="binomial"))) #OR (PhD as baseline)## (Intercept)
## 0.05882353
## Supporter_CareerStage_cleanPostdocs
## 0.41717791
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 0.47887324
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 0.85000000
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 0.60714286exp(confint(glm(Support_Emotional_Institution~Supporter_CareerStage_clean, data=Support_Emotional, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.03926516 0.08439725
## Supporter_CareerStage_cleanPostdocs 0.17555276 0.88573845
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.14006152 1.24581479
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.24683438 2.23869209
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.17713690 1.58624590#FLR gender-CS interactions
summary(glm(Support_Emotional_Institution~Supporter_CareerStage_clean*All_Gender_clean, data=Support_Emotional, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Support_Emotional_Institution ~ Supporter_CareerStage_clean *
## All_Gender_clean, family = "binomial", data = Support_Emotional)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.3949 -0.3582 -0.2561 -0.2144 2.7544
##
## Coefficients:
## Estimate
## (Intercept) -3.17805
## Supporter_CareerStage_cleanPostdocs -0.41468
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.22314
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -0.58315
## Supporter_CareerStage_cleanGroup leaders (>10yr) -0.01379
## All_Gender_cleanWomen 0.46396
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -0.64169
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.77634
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.78493
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen -0.93568
## Std. Error
## (Intercept) 0.41667
## Supporter_CareerStage_cleanPostdocs 0.71841
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.83083
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 1.09399
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.72156
## All_Gender_cleanWomen 0.47125
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.87703
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 1.11845
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 1.26714
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 1.26283
## z value
## (Intercept) -7.627
## Supporter_CareerStage_cleanPostdocs -0.577
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.269
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -0.533
## Supporter_CareerStage_cleanGroup leaders (>10yr) -0.019
## All_Gender_cleanWomen 0.985
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -0.732
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.694
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.619
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen -0.741
## Pr(>|z|)
## (Intercept) 2.4e-14
## Supporter_CareerStage_cleanPostdocs 0.564
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.788
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.594
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.985
## All_Gender_cleanWomen 0.325
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.464
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.488
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.536
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.459
##
## (Intercept) ***
## Supporter_CareerStage_cleanPostdocs
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## All_Gender_cleanWomen
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 401.75 on 1183 degrees of freedom
## Residual deviance: 392.74 on 1174 degrees of freedom
## AIC: 412.74
##
## Number of Fisher Scoring iterations: 6Support_Emotional_Colleagues x Gender // xCS
As above
#LR gender
summary(glm(Support_Emotional_Colleagues~All_Gender_clean, data=Support_Emotional, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Support_Emotional_Colleagues ~ All_Gender_clean,
## family = "binomial", data = Support_Emotional)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.7442 -0.7442 -0.6083 -0.6083 1.8859
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.5933 0.1267 -12.579 < 2e-16 ***
## All_Gender_cleanWomen 0.4510 0.1530 2.947 0.00321 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1231.1 on 1183 degrees of freedom
## Residual deviance: 1222.1 on 1182 degrees of freedom
## AIC: 1226.1
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Support_Emotional_Colleagues~All_Gender_clean, data=Support_Emotional, family="binomial"))) #OR (men as baseline)## (Intercept) All_Gender_cleanWomen
## 0.203252 1.569840exp(confint(glm(Support_Emotional_Colleagues~All_Gender_clean, data=Support_Emotional, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.1574342 0.2588495
## All_Gender_cleanWomen 1.1672347 2.1278937#LR CS
summary(glm(Support_Emotional_Colleagues~Supporter_CareerStage_clean, data=Support_Emotional, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Support_Emotional_Colleagues ~ Supporter_CareerStage_clean,
## family = "binomial", data = Support_Emotional)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.7645 -0.7090 -0.6692 -0.5807 1.9304
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) -1.25276 0.10714 -11.692
## Supporter_CareerStage_cleanPostdocs -0.12979 0.17364 -0.747
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.17233 0.21836 0.789
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.08961 0.27768 0.323
## Supporter_CareerStage_cleanGroup leaders (>10yr) -0.44183 0.27792 -1.590
## Pr(>|z|)
## (Intercept) <2e-16 ***
## Supporter_CareerStage_cleanPostdocs 0.455
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.430
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.747
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.112
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1231.1 on 1183 degrees of freedom
## Residual deviance: 1226.4 on 1179 degrees of freedom
## AIC: 1236.4
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Support_Emotional_Colleagues~Supporter_CareerStage_clean, data=Support_Emotional, family="binomial"))) #OR (PhD as baseline)## (Intercept)
## 0.2857143
## Supporter_CareerStage_cleanPostdocs
## 0.8782772
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 1.1880734
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 1.0937500
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 0.6428571exp(confint(glm(Support_Emotional_Colleagues~Supporter_CareerStage_clean, data=Support_Emotional, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.2306011 0.3511131
## Supporter_CareerStage_cleanPostdocs 0.6228445 1.2313162
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.7676586 1.8104257
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.6215545 1.8554734
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.3628349 1.0848971#FLR gender-CS interactions
summary(glm(Support_Emotional_Colleagues~Supporter_CareerStage_clean*All_Gender_clean, data=Support_Emotional, family="binomial")) #glm and p-values##
## Call:
## glm(formula = Support_Emotional_Colleagues ~ Supporter_CareerStage_clean *
## All_Gender_clean, family = "binomial", data = Support_Emotional)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.8203 -0.7511 -0.6860 -0.4717 2.1219
##
## Coefficients:
## Estimate
## (Intercept) -1.609438
## Supporter_CareerStage_cleanPostdocs -0.111226
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.282567
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.628609
## Supporter_CareerStage_cleanGroup leaders (>10yr) -0.530628
## All_Gender_cleanWomen 0.488097
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -0.002905
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.077517
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -0.893562
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.553356
## Std. Error
## (Intercept) 0.219089
## Supporter_CareerStage_cleanPostdocs 0.342564
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.381223
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.403216
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.433235
## All_Gender_cleanWomen 0.251475
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.397999
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.467872
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.577990
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.579884
## z value
## (Intercept) -7.346
## Supporter_CareerStage_cleanPostdocs -0.325
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.741
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 1.559
## Supporter_CareerStage_cleanGroup leaders (>10yr) -1.225
## All_Gender_cleanWomen 1.941
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -0.007
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.166
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -1.546
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.954
## Pr(>|z|)
## (Intercept) 2.04e-13
## Supporter_CareerStage_cleanPostdocs 0.7454
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.4586
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.1190
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.2206
## All_Gender_cleanWomen 0.0523
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.9942
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.8684
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.1221
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.3400
##
## (Intercept) ***
## Supporter_CareerStage_cleanPostdocs
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## All_Gender_cleanWomen .
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1231.1 on 1183 degrees of freedom
## Residual deviance: 1214.1 on 1174 degrees of freedom
## AIC: 1234.1
##
## Number of Fisher Scoring iterations: 4Support_Emotional_Personal x Gender // xCS
As above
#LR gender
summary(glm(Support_Emotional_Personal~All_Gender_clean, data=Support_Emotional, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Support_Emotional_Personal ~ All_Gender_clean,
## family = "binomial", data = Support_Emotional)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.2284 -1.2284 -0.9491 1.1273 1.4244
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.56404 0.09872 -5.714 1.10e-08 ***
## All_Gender_cleanWomen 0.68310 0.12316 5.546 2.92e-08 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1636.2 on 1183 degrees of freedom
## Residual deviance: 1604.8 on 1182 degrees of freedom
## AIC: 1608.8
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Support_Emotional_Personal~All_Gender_clean, data=Support_Emotional, family="binomial"))) #OR (men as baseline)## (Intercept) All_Gender_cleanWomen
## 0.5689046 1.9800100exp(confint(glm(Support_Emotional_Personal~All_Gender_clean, data=Support_Emotional, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.4679493 0.6892667
## All_Gender_cleanWomen 1.5570387 2.5238231#LR CS
summary(glm(Support_Emotional_Personal~Supporter_CareerStage_clean, data=Support_Emotional, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Support_Emotional_Personal ~ Supporter_CareerStage_clean,
## family = "binomial", data = Support_Emotional)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.2029 -1.1272 -0.8909 1.2284 1.4940
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) -0.11919 0.08925 -1.336
## Supporter_CareerStage_cleanPostdocs 0.17909 0.14125 1.268
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.17400 0.18810 0.925
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -0.41712 0.24309 -1.716
## Supporter_CareerStage_cleanGroup leaders (>10yr) -0.59993 0.21703 -2.764
## Pr(>|z|)
## (Intercept) 0.1817
## Supporter_CareerStage_cleanPostdocs 0.2048
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.3550
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.0862 .
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.0057 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1636.2 on 1183 degrees of freedom
## Residual deviance: 1619.3 on 1179 degrees of freedom
## AIC: 1629.3
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Support_Emotional_Personal~Supporter_CareerStage_clean, data=Support_Emotional, family="binomial"))) #OR (PhD as baseline)## (Intercept)
## 0.8876404
## Supporter_CareerStage_cleanPostdocs
## 1.1961244
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 1.1900517
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 0.6589444
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 0.5488478exp(confint(glm(Support_Emotional_Personal~Supporter_CareerStage_clean, data=Support_Emotional, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.7448872 1.0571082
## Supporter_CareerStage_cleanPostdocs 0.9069908 1.5782485
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.8230520 1.7222213
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.4055372 1.0548842
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.3557587 0.8346663#FLR gender-CS interactions
summary(glm(Support_Emotional_Personal~Supporter_CareerStage_clean*All_Gender_clean, data=Support_Emotional, family="binomial")) #glm and p-values##
## Call:
## glm(formula = Support_Emotional_Personal ~ Supporter_CareerStage_clean *
## All_Gender_clean, family = "binomial", data = Support_Emotional)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.345 -1.197 -0.849 1.158 1.665
##
## Coefficients:
## Estimate
## (Intercept) -0.5179
## Supporter_CareerStage_cleanPostdocs 0.1569
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.1259
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -0.5807
## Supporter_CareerStage_cleanGroup leaders (>10yr) -0.3169
## All_Gender_cleanWomen 0.5632
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.0698
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.2146
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.5355
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen -0.2392
## Std. Error
## (Intercept) 0.1688
## Supporter_CareerStage_cleanPostdocs 0.2557
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.3091
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.3869
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.3014
## All_Gender_cleanWomen 0.1995
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.3083
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.3953
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.5109
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.4570
## z value
## (Intercept) -3.068
## Supporter_CareerStage_cleanPostdocs 0.614
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.407
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -1.501
## Supporter_CareerStage_cleanGroup leaders (>10yr) -1.051
## All_Gender_cleanWomen 2.823
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.226
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.543
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 1.048
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen -0.523
## Pr(>|z|)
## (Intercept) 0.00215
## Supporter_CareerStage_cleanPostdocs 0.53941
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.68374
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.13342
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.29313
## All_Gender_cleanWomen 0.00476
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.82089
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.58726
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.29460
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.60069
##
## (Intercept) **
## Supporter_CareerStage_cleanPostdocs
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## All_Gender_cleanWomen **
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1636.2 on 1183 degrees of freedom
## Residual deviance: 1592.2 on 1174 degrees of freedom
## AIC: 1612.2
##
## Number of Fisher Scoring iterations: 4Q15 Supporter action advocacy (Section 3, Panel 3)
Answered to Q15: When supporting the person (select all that applies), I advocated for the person within my institution
Exclude everyone that only answered PNTA and/or unsure, or did not check any options at all
#subset
Supporter_Actions_Advocacy=genderCS[!is.na(genderCS$Supporter_Actions_OpenUp==1|genderCS$Supporter_Actions_Resources==1|genderCS$Supporter_Actions_HelpResources==1|genderCS$Supporter_Actions_Emotional==1|genderCS$Supporter_Actions_Advice==1|genderCS$Supporter_Actions_Work==1|genderCS$Supporter_Actions_Personal==1|genderCS$Supporter_Actions_Crisis==1|genderCS$Supporter_Actions_Adjustments==1|genderCS$Supporter_Actions_Advocacy==1|genderCS$Supporter_Actions_Other_Dummy==1),c("Supporter_Actions_Advocacy", "Supporter_CareerStage_clean", "All_Gender_clean")]
dim(Supporter_Actions_Advocacy)[1] #N## [1] 1247#fill all na with 0
Supporter_Actions_Advocacy[is.na(Supporter_Actions_Advocacy)]=0
#factorise
Supporter_Actions_Advocacy$Supporter_Actions_Advocacy=as.factor(Supporter_Actions_Advocacy$Supporter_Actions_Advocacy)
table(Supporter_Actions_Advocacy$Supporter_Actions_Advocacy)##
## 0 1
## 943 304Supporter_Action_Advocacy x gender // x CS
#LR gender
summary(glm(Supporter_Actions_Advocacy~All_Gender_clean, data=Supporter_Actions_Advocacy, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Supporter_Actions_Advocacy ~ All_Gender_clean,
## family = "binomial", data = Supporter_Actions_Advocacy)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.7585 -0.7585 -0.7409 -0.7409 1.6894
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.09861 0.10630 -10.335 <2e-16 ***
## All_Gender_cleanWomen -0.05407 0.13555 -0.399 0.69
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1385.2 on 1246 degrees of freedom
## Residual deviance: 1385.0 on 1245 degrees of freedom
## AIC: 1389
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Supporter_Actions_Advocacy~All_Gender_clean, data=Supporter_Actions_Advocacy, family="binomial"))) #OR (men as baseline)## (Intercept) All_Gender_cleanWomen
## 0.3333333 0.9473684exp(confint(glm(Supporter_Actions_Advocacy~All_Gender_clean, data=Supporter_Actions_Advocacy, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.2696083 0.4091423
## All_Gender_cleanWomen 0.7271645 1.2375270#LR CS
summary(glm(Supporter_Actions_Advocacy~Supporter_CareerStage_clean, data=Supporter_Actions_Advocacy, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Supporter_Actions_Advocacy ~ Supporter_CareerStage_clean,
## family = "binomial", data = Supporter_Actions_Advocacy)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.9741 -0.7585 -0.6305 -0.6305 1.8511
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) -1.5145 0.1133 -13.365
## Supporter_CareerStage_cleanPostdocs 0.4159 0.1668 2.494
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.5572 0.2122 2.626
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 1.0156 0.2452 4.142
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.9450 0.2219 4.258
## Pr(>|z|)
## (Intercept) < 2e-16 ***
## Supporter_CareerStage_cleanPostdocs 0.01264 *
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.00864 **
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 3.44e-05 ***
## Supporter_CareerStage_cleanGroup leaders (>10yr) 2.06e-05 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1385.2 on 1246 degrees of freedom
## Residual deviance: 1355.7 on 1242 degrees of freedom
## AIC: 1365.7
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Supporter_Actions_Advocacy~Supporter_CareerStage_clean, data=Supporter_Actions_Advocacy, family="binomial"))) #OR (PhDs as baseline)## (Intercept)
## 0.2199074
## Supporter_CareerStage_cleanPostdocs
## 1.5157895
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 1.7458647
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 2.7609023
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 2.5728532exp(confint(glm(Supporter_Actions_Advocacy~Supporter_CareerStage_clean, data=Supporter_Actions_Advocacy, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.1751382 0.2732252
## Supporter_CareerStage_cleanPostdocs 1.0925284 2.1023615
## Supporter_CareerStage_cleanGroup leaders (<5yr) 1.1450254 2.6349638
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 1.6971979 4.4487705
## Supporter_CareerStage_cleanGroup leaders (>10yr) 1.6581630 3.9646690#FLR gender-CS interactions
summary(glm(Supporter_Actions_Advocacy~Supporter_CareerStage_clean*All_Gender_clean, data=Supporter_Actions_Advocacy, family="binomial"))##
## Call:
## glm(formula = Supporter_Actions_Advocacy ~ Supporter_CareerStage_clean *
## All_Gender_clean, family = "binomial", data = Supporter_Actions_Advocacy)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.0974 -0.7642 -0.6434 -0.5997 1.8997
##
## Coefficients:
## Estimate
## (Intercept) -1.6247
## Supporter_CareerStage_cleanPostdocs 0.4933
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.6439
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.8362
## Supporter_CareerStage_cleanGroup leaders (>10yr) 1.1208
## All_Gender_cleanWomen 0.1550
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -0.1051
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.1142
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.4424
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen -0.3443
## Std. Error
## (Intercept) 0.2146
## Supporter_CareerStage_cleanPostdocs 0.3002
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.3499
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.3782
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.3184
## All_Gender_cleanWomen 0.2527
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.3614
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.4426
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.5069
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.4757
## z value
## (Intercept) -7.572
## Supporter_CareerStage_cleanPostdocs 1.643
## Supporter_CareerStage_cleanGroup leaders (<5yr) 1.840
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 2.211
## Supporter_CareerStage_cleanGroup leaders (>10yr) 3.521
## All_Gender_cleanWomen 0.613
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -0.291
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.258
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.873
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen -0.724
## Pr(>|z|)
## (Intercept) 3.67e-14
## Supporter_CareerStage_cleanPostdocs 0.100338
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.065739
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.027012
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.000431
## All_Gender_cleanWomen 0.539570
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.771129
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.796369
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.382821
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.469272
##
## (Intercept) ***
## Supporter_CareerStage_cleanPostdocs
## Supporter_CareerStage_cleanGroup leaders (<5yr) .
## Supporter_CareerStage_cleanGroup leaders (5-10yr) *
## Supporter_CareerStage_cleanGroup leaders (>10yr) ***
## All_Gender_cleanWomen
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1385.2 on 1246 degrees of freedom
## Residual deviance: 1353.2 on 1237 degrees of freedom
## AIC: 1373.2
##
## Number of Fisher Scoring iterations: 4Q22 Needing practical information (Section 6, Panel 2)
Dependent variable: Support_Practical
- 1: No, did not need info
- 2: No, already had info
- 3: Yes, need info
(remove 4- not sure, 5 - PNTA and 6- other)
3 approaches:
- chi square (but cannot check for intersect)
- turn the 3 options into 3 x [yes/no]s
- multinomial logistic regression
FINAL APPROACH (immediately following here) recode dependent variable: - No - 1 and 2 - Yes - 3 Run as binomial logistic regression ### Needing practical information x Gender x CS xGenderxCS (Section 6, Panel 2)
#Subset
Support_Practical=genderCS[,c("Support_Practical", "Supporter_CareerStage_clean", "All_Gender_clean")]
#Recode
Support_Practical=Support_Practical[Support_Practical$Support_Practical %in% seq(1,3),] #remove options 4&5
Support_Practical[Support_Practical$Support_Practical %in% c(1,2), "Support_Practical"]=0
Support_Practical[Support_Practical$Support_Practical==3, "Support_Practical"]=1
# LR gender
summary(glm(Support_Practical~All_Gender_clean, data=Support_Practical, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Support_Practical ~ All_Gender_clean, family = "binomial",
## data = Support_Practical)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.5579 -1.4592 0.8397 0.8397 0.9196
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.6419 0.1084 5.924 3.15e-09 ***
## All_Gender_cleanWomen 0.2192 0.1392 1.574 0.115
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1249.9 on 1002 degrees of freedom
## Residual deviance: 1247.4 on 1001 degrees of freedom
## AIC: 1251.4
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Support_Practical~All_Gender_clean, data=Support_Practical, family="binomial"))) #OR (men as baseline)## (Intercept) All_Gender_cleanWomen
## 1.900000 1.245048exp(confint(glm(Support_Practical~All_Gender_clean, data=Support_Practical, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 1.5397399 2.355546
## All_Gender_cleanWomen 0.9469236 1.634964# LR CS
summary(glm(Support_Practical~Supporter_CareerStage_clean, data=Support_Practical, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Support_Practical ~ Supporter_CareerStage_clean,
## family = "binomial", data = Support_Practical)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.7941 -1.4736 0.8586 0.9073 0.9772
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) 0.674998 0.104198 6.478
## Supporter_CareerStage_cleanPostdocs 0.133219 0.168781 0.789
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.711296 0.235585 3.019
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -0.001269 0.262539 -0.005
## Supporter_CareerStage_cleanGroup leaders (>10yr) -0.183877 0.223993 -0.821
## Pr(>|z|)
## (Intercept) 9.29e-11 ***
## Supporter_CareerStage_cleanPostdocs 0.42994
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.00253 **
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.99614
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.41170
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1249.9 on 1002 degrees of freedom
## Residual deviance: 1237.4 on 998 degrees of freedom
## AIC: 1247.4
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Support_Practical~Supporter_CareerStage_clean, data=Support_Practical, family="binomial")))## (Intercept)
## 1.9640288
## Supporter_CareerStage_cleanPostdocs
## 1.1424998
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 2.0366300
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 0.9987320
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 0.8320379exp(confint(glm(Support_Practical~Supporter_CareerStage_clean, data=Support_Practical, family="binomial")))## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 1.6045696 2.414938
## Supporter_CareerStage_cleanPostdocs 0.8219678 1.593839
## Supporter_CareerStage_cleanGroup leaders (<5yr) 1.2989290 3.280392
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.6018136 1.690628
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.5381918 1.297426#FLR gender-CS interactions
summary(glm(Support_Practical~Supporter_CareerStage_clean*All_Gender_clean, data=Support_Practical, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Support_Practical ~ Supporter_CareerStage_clean *
## All_Gender_clean, family = "binomial", data = Support_Practical)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.919 -1.366 0.840 0.885 1.048
##
## Coefficients:
## Estimate
## (Intercept) 0.4329
## Supporter_CareerStage_cleanPostdocs 0.5322
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.6203
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.2603
## Supporter_CareerStage_cleanGroup leaders (>10yr) -0.1205
## All_Gender_cleanWomen 0.3496
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -0.5793
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.2664
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -0.3888
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.1982
## Std. Error
## (Intercept) 0.1853
## Supporter_CareerStage_cleanPostdocs 0.3031
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.3525
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.3870
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.3034
## All_Gender_cleanWomen 0.2244
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.3652
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.4814
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.5317
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.4873
## z value
## (Intercept) 2.336
## Supporter_CareerStage_cleanPostdocs 1.756
## Supporter_CareerStage_cleanGroup leaders (<5yr) 1.760
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.673
## Supporter_CareerStage_cleanGroup leaders (>10yr) -0.397
## All_Gender_cleanWomen 1.558
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -1.586
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.553
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -0.731
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.407
## Pr(>|z|)
## (Intercept) 0.0195
## Supporter_CareerStage_cleanPostdocs 0.0791
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.0785
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.5012
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.6913
## All_Gender_cleanWomen 0.1193
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.1127
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.5799
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.4646
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.6841
##
## (Intercept) *
## Supporter_CareerStage_cleanPostdocs .
## Supporter_CareerStage_cleanGroup leaders (<5yr) .
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## All_Gender_cleanWomen
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1249.9 on 1002 degrees of freedom
## Residual deviance: 1230.6 on 993 degrees of freedom
## AIC: 1250.6
##
## Number of Fisher Scoring iterations: 4### Needing practical information x CS pairwise# Compare between early PIs and other CSs only
## this requires the package "MHTdiscrete"
CSpairs=cbind(rep("Group leaders (<5yr)", 4), levels(ImpactDraining$Supporter_CareerStage_clean)[c(1,2,4,5)])
Support_PracticalCSKW=data.frame(CS1=c(), CS2=c(), X2=c(), N=c(), df=c(), pVal=c())
for (i in 1:dim(CSpairs)[1]) {
comparedData=Support_Practical[(Support_Practical$Supporter_CareerStage_clean==CSpairs[i,1]|Support_Practical$Supporter_CareerStage_clean==CSpairs[i,2]),]
comparedData$Support_Practical=as.factor(comparedData$Support_Practical)
comparedData$Supporter_CareerStage_clean=droplevels(comparedData$Supporter_CareerStage_clean) #drop levels to avoid incorrect chisq approx
# KW=kruskal.test(Support_Practical ~ Supporter_CareerStage_clean, data=comparedData) #effect of CS on response
chisq=chisq.test(table(comparedData[,c("Support_Practical", "Supporter_CareerStage_clean")]))
# Support_PracticalCSKW=rbind(Support_PracticalCSKW, data.frame(CS1=CSpairs[i,1], CS2=CSpairs[i,2], X2=as.numeric(KW$statistic), df=as.numeric(KW$parameter), pVal=as.numeric(KW$p.value)))
Support_PracticalCSKW=rbind(Support_PracticalCSKW, data.frame(CS1=CSpairs[i,1], CS2=CSpairs[i,2], X2=as.numeric(chisq$statistic), N=dim(comparedData)[1], df=as.numeric(chisq$parameter), pVal=as.numeric(chisq$p.value)))
}
Support_PracticalCSKW$adjpVal=Sidak.p.adjust(Support_PracticalCSKW$pVal)
Support_PracticalCSKW## CS1 CS2 X2 N df pVal
## 1 Group leaders (<5yr) PhD students 8.706275 552 1 0.003171165
## 2 Group leaders (<5yr) Postdocs 4.909109 406 1 0.026715415
## 3 Group leaders (<5yr) Group leaders (5-10yr) 4.326939 217 1 0.037513645
## 4 Group leaders (<5yr) Group leaders (>10yr) 8.922114 248 1 0.002817382
## adjpVal
## 1 0.01262445
## 2 0.10265514
## 3 0.14182013
## 4 0.01122199Support_Practical=genderCS[,c("Support_Practical", "Supporter_CareerStage_clean", "All_Gender_clean")]
chisq.test(table(Support_Practical[,c("Support_Practical", "All_Gender_clean")]))## Warning in chisq.test(table(Support_Practical[, c("Support_Practical",
## "All_Gender_clean")])): Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: table(Support_Practical[, c("Support_Practical", "All_Gender_clean")])
## X-squared = 12.896, df = 5, p-value = 0.02437chisq.test(table(Support_Practical[,c("Support_Practical", "Supporter_CareerStage_clean")]))## Warning in chisq.test(table(Support_Practical[, c("Support_Practical",
## "Supporter_CareerStage_clean")])): Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: table(Support_Practical[, c("Support_Practical", "Supporter_CareerStage_clean")])
## X-squared = 58.601, df = 20, p-value = 1.17e-05Approach 2: turn the 3 options into 3 x [yes/no]s [not reported]
#Subset
Support_Practical=genderCS[,c("Support_Practical", "Supporter_CareerStage_clean", "All_Gender_clean")]
#Factorise
Support_Practical$Support_Practical=as.factor(Support_Practical$Support_Practical)
#recode and clean
Support_Practical=cbind(sapply(levels(Support_Practical$Support_Practical), function(x) as.integer(x==Support_Practical$Support_Practical)), Support_Practical[2:3])
Support_Practical=Support_Practical[,c(2,3,4,7,8)] #remove some options
#rename columns
colnames(Support_Practical)=c("NoNeed", "Had", "Need", "Supporter_CareerStage_clean", "All_Gender_clean")Choosing option 1 - No, I did not need the information
# LR gender
summary(glm(NoNeed~All_Gender_clean, data=Support_Practical, family="binomial")) #glm and p-value##
## Call:
## glm(formula = NoNeed ~ All_Gender_clean, family = "binomial",
## data = Support_Practical)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.6335 -0.6335 -0.5232 -0.5232 2.0280
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.5041 0.1192 -12.617 < 2e-16 ***
## All_Gender_cleanWomen -0.4155 0.1606 -2.587 0.00968 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1047.1 on 1246 degrees of freedom
## Residual deviance: 1040.5 on 1245 degrees of freedom
## (8 observations deleted due to missingness)
## AIC: 1044.5
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(NoNeed~All_Gender_clean, data=Support_Practical, family="binomial"))) #OR (men as baseline)## (Intercept) All_Gender_cleanWomen
## 0.2222222 0.6600000exp(confint(glm(NoNeed~All_Gender_clean, data=Support_Practical, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.1748478 0.2791723
## All_Gender_cleanWomen 0.4819197 0.9051816# LR CS
summary(glm(NoNeed~Supporter_CareerStage_clean, data=Support_Practical, family="binomial")) #glm and p-value##
## Call:
## glm(formula = NoNeed ~ Supporter_CareerStage_clean, family = "binomial",
## data = Support_Practical)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.6712 -0.5804 -0.5467 -0.5256 2.1016
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) -1.6958 0.1201 -14.116
## Supporter_CareerStage_cleanPostdocs -0.1294 0.1953 -0.663
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.2137 0.2680 -0.797
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -0.3960 0.3561 -1.112
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.3200 0.2581 1.240
## Pr(>|z|)
## (Intercept) <2e-16 ***
## Supporter_CareerStage_cleanPostdocs 0.508
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.425
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.266
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.215
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1047.1 on 1246 degrees of freedom
## Residual deviance: 1042.6 on 1242 degrees of freedom
## (8 observations deleted due to missingness)
## AIC: 1052.6
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(NoNeed~Supporter_CareerStage_clean, data=Support_Practical, family="binomial")))## (Intercept)
## 0.1834452
## Supporter_CareerStage_cleanPostdocs
## 0.8786505
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 0.8075881
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 0.6729901
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 1.3771502exp(confint(glm(NoNeed~Supporter_CareerStage_clean, data=Support_Practical, family="binomial")))## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.1439909 0.2307417
## Supporter_CareerStage_cleanPostdocs 0.5961332 1.2836426
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.4666066 1.3410920
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.3164114 1.2959563
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.8169176 2.2557908#FLR gender-CS interactions
summary(glm(NoNeed~Supporter_CareerStage_clean*All_Gender_clean, data=Support_Practical, family="binomial")) #glm and p-value##
## Call:
## glm(formula = NoNeed ~ Supporter_CareerStage_clean * All_Gender_clean,
## family = "binomial", data = Support_Practical)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.7528 -0.5478 -0.5338 -0.4972 2.1951
##
## Coefficients:
## Estimate
## (Intercept) -1.3477
## Supporter_CareerStage_cleanPostdocs -0.4731
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.1564
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -0.8041
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.2317
## All_Gender_cleanWomen -0.5288
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.5220
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.2821
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.6524
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen -0.3567
## Std. Error
## (Intercept) 0.1954
## Supporter_CareerStage_cleanPostdocs 0.3264
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.3742
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.5113
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.3287
## All_Gender_cleanWomen 0.2484
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.4078
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.5485
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.7150
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.5988
## z value
## (Intercept) -6.897
## Supporter_CareerStage_cleanPostdocs -1.449
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.418
## Supporter_CareerStage_cleanGroup leaders (5-10yr) -1.573
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.705
## All_Gender_cleanWomen -2.129
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 1.280
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen -0.514
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.912
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen -0.596
## Pr(>|z|)
## (Intercept) 5.3e-12
## Supporter_CareerStage_cleanPostdocs 0.1472
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.6760
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.1158
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.4809
## All_Gender_cleanWomen 0.0333
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.2006
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.6070
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.3615
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.5515
##
## (Intercept) ***
## Supporter_CareerStage_cleanPostdocs
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## All_Gender_cleanWomen *
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1047.1 on 1246 degrees of freedom
## Residual deviance: 1032.4 on 1237 degrees of freedom
## (8 observations deleted due to missingness)
## AIC: 1052.4
##
## Number of Fisher Scoring iterations: 4Choosing option 2 - No, I have already received information before
# LR gender
summary(glm(Had~All_Gender_clean, data=Support_Practical, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Had ~ All_Gender_clean, family = "binomial", data = Support_Practical)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.4883 -0.4883 -0.4883 -0.4419 2.1794
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -2.2773 0.1583 -14.388 <2e-16 ***
## All_Gender_cleanWomen 0.2108 0.1949 1.082 0.279
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 838.09 on 1246 degrees of freedom
## Residual deviance: 836.90 on 1245 degrees of freedom
## (8 observations deleted due to missingness)
## AIC: 840.9
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Had~All_Gender_clean, data=Support_Practical, family="binomial"))) #OR (men as baseline)## (Intercept) All_Gender_cleanWomen
## 0.1025641 1.2347161exp(confint(glm(Had~All_Gender_clean, data=Support_Practical, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.07417135 0.1381601
## All_Gender_cleanWomen 0.84728128 1.8227017#LR CS
summary(glm(Had~Supporter_CareerStage_clean, data=Support_Practical, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Had ~ Supporter_CareerStage_clean, family = "binomial",
## data = Support_Practical)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.6219 -0.4775 -0.4775 -0.4430 2.4347
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) -2.1139 0.1402 -15.075
## Supporter_CareerStage_cleanPostdocs -0.1579 0.2304 -0.685
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.7971 0.3892 -2.048
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.5690 0.3090 1.841
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.3222 0.2971 1.084
## Pr(>|z|)
## (Intercept) <2e-16 ***
## Supporter_CareerStage_cleanPostdocs 0.4932
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.0406 *
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.0656 .
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.2783
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 838.09 on 1246 degrees of freedom
## Residual deviance: 826.08 on 1242 degrees of freedom
## (8 observations deleted due to missingness)
## AIC: 836.08
##
## Number of Fisher Scoring iterations: 5exp(coef(glm(Had~Supporter_CareerStage_clean, data=Support_Practical, family="binomial"))) #OR (PhD as baseline)## (Intercept)
## 0.1207627
## Supporter_CareerStage_cleanPostdocs
## 0.8539474
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 0.4506504
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 1.7665497
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 1.3801170exp(confint(glm(Had~Supporter_CareerStage_clean, data=Support_Practical, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.09079219 0.1574771
## Supporter_CareerStage_cleanPostdocs 0.53884742 1.3335462
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.19486697 0.9140110
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.93845737 3.1742942
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.75127402 2.4240553#FLR gender-CS interactions
summary(glm(Had~Supporter_CareerStage_clean*All_Gender_clean, data=Support_Practical, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Had ~ Supporter_CareerStage_clean * All_Gender_clean,
## family = "binomial", data = Support_Practical)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.6416 -0.4916 -0.4887 -0.3438 2.4864
##
## Coefficients:
## Estimate
## (Intercept) -2.268684
## Supporter_CareerStage_cleanPostdocs -0.530338
## Supporter_CareerStage_cleanGroup leaders (<5yr) -0.775839
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.659246
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.476924
## All_Gender_cleanWomen 0.216393
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.517716
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.006751
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -0.082862
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen -0.216393
## Std. Error
## (Intercept) 0.271225
## Supporter_CareerStage_cleanPostdocs 0.474459
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.650149
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.472825
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.423821
## All_Gender_cleanWomen 0.316908
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.543432
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.813301
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.635612
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.633195
## z value
## (Intercept) -8.365
## Supporter_CareerStage_cleanPostdocs -1.118
## Supporter_CareerStage_cleanGroup leaders (<5yr) -1.193
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 1.394
## Supporter_CareerStage_cleanGroup leaders (>10yr) 1.125
## All_Gender_cleanWomen 0.683
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.953
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.008
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -0.130
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen -0.342
## Pr(>|z|)
## (Intercept) <2e-16
## Supporter_CareerStage_cleanPostdocs 0.264
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.233
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.163
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.260
## All_Gender_cleanWomen 0.495
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.341
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.993
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.896
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.733
##
## (Intercept) ***
## Supporter_CareerStage_cleanPostdocs
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## All_Gender_cleanWomen
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 838.09 on 1246 degrees of freedom
## Residual deviance: 822.37 on 1237 degrees of freedom
## (8 observations deleted due to missingness)
## AIC: 842.37
##
## Number of Fisher Scoring iterations: 5Choosing option 3 - Yes I needed information
#LR Gender
summary(glm(Need~All_Gender_clean, data=Support_Practical, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Need ~ All_Gender_clean, family = "binomial", data = Support_Practical)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.296 -1.296 1.063 1.063 1.140
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.08885 0.09205 0.965 0.334
## All_Gender_cleanWomen 0.18678 0.11722 1.593 0.111
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1715.8 on 1246 degrees of freedom
## Residual deviance: 1713.2 on 1245 degrees of freedom
## (8 observations deleted due to missingness)
## AIC: 1717.2
##
## Number of Fisher Scoring iterations: 3exp(coef(glm(Need~All_Gender_clean, data=Support_Practical, family="binomial"))) #OR (men as baseline)## (Intercept) All_Gender_cleanWomen
## 1.092920 1.205363exp(confint(glm(Need~All_Gender_clean, data=Support_Practical, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.9126118 1.309485
## All_Gender_cleanWomen 0.9579020 1.516837#LR CS
summary(glm(Need~Supporter_CareerStage_clean, data=Support_Practical, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Need ~ Supporter_CareerStage_clean, family = "binomial",
## data = Support_Practical)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.6014 -1.2048 0.8061 1.1415 1.1502
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) 0.06429 0.08700 0.739
## Supporter_CareerStage_cleanPostdocs 0.02074 0.13755 0.151
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.89300 0.19938 4.479
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.17865 0.22842 0.782
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.18915 0.20427 0.926
## Pr(>|z|)
## (Intercept) 0.460
## Supporter_CareerStage_cleanPostdocs 0.880
## Supporter_CareerStage_cleanGroup leaders (<5yr) 7.51e-06 ***
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.434
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.354
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1715.8 on 1246 degrees of freedom
## Residual deviance: 1692.5 on 1242 degrees of freedom
## (8 observations deleted due to missingness)
## AIC: 1702.5
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Need~Supporter_CareerStage_clean, data=Support_Practical, family="binomial"))) #OR (PhD as baseline)## (Intercept)
## 1.066406
## Supporter_CareerStage_cleanPostdocs
## 1.020959
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## 2.442457
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## 1.195604
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## 1.208228exp(confint(glm(Need~Supporter_CareerStage_clean, data=Support_Practical, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 0.8992690 1.265002
## Supporter_CareerStage_cleanPostdocs 0.7796964 1.337207
## Supporter_CareerStage_cleanGroup leaders (<5yr) 1.6634233 3.640113
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.7654491 1.878551
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.8108139 1.808527#FLR gender-CS interactions
summary(glm(Need~Supporter_CareerStage_clean*All_Gender_clean, data=Support_Practical, family="binomial")) #glm and p-value##
## Call:
## glm(formula = Need ~ Supporter_CareerStage_clean * All_Gender_clean,
## family = "binomial", data = Support_Practical)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.7280 -1.2181 0.7135 1.1113 1.2419
##
## Coefficients:
## Estimate
## (Intercept) -0.15028
## Supporter_CareerStage_cleanPostdocs 0.21588
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.77599
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.31734
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.28034
## All_Gender_cleanWomen 0.30779
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -0.27808
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.30488
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -0.14634
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.04767
## Std. Error
## (Intercept) 0.15856
## Supporter_CareerStage_cleanPostdocs 0.24076
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.30311
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.33024
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.27805
## All_Gender_cleanWomen 0.18987
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.29364
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.40899
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.46424
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.43495
## z value
## (Intercept) -0.948
## Supporter_CareerStage_cleanPostdocs 0.897
## Supporter_CareerStage_cleanGroup leaders (<5yr) 2.560
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.961
## Supporter_CareerStage_cleanGroup leaders (>10yr) 1.008
## All_Gender_cleanWomen 1.621
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen -0.947
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.745
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen -0.315
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.110
## Pr(>|z|)
## (Intercept) 0.3432
## Supporter_CareerStage_cleanPostdocs 0.3699
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.0105
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.3366
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.3133
## All_Gender_cleanWomen 0.1050
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen 0.3436
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen 0.4560
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen 0.7526
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen 0.9127
##
## (Intercept)
## Supporter_CareerStage_cleanPostdocs
## Supporter_CareerStage_cleanGroup leaders (<5yr) *
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## All_Gender_cleanWomen
## Supporter_CareerStage_cleanPostdocs:All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (<5yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (5-10yr):All_Gender_cleanWomen
## Supporter_CareerStage_cleanGroup leaders (>10yr):All_Gender_cleanWomen
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1715.8 on 1246 degrees of freedom
## Residual deviance: 1686.0 on 1237 degrees of freedom
## (8 observations deleted due to missingness)
## AIC: 1706
##
## Number of Fisher Scoring iterations: 4Needing practical information x gender graph
Support_Practical=genderCS[,c("Support_Practical", "Supporter_CareerStage_clean", "All_Gender_clean")]
#Subset
Support_Practical=genderCS[,c("Support_Practical", "Supporter_CareerStage_clean", "All_Gender_clean")]
#Recode
Support_Practical=Support_Practical[Support_Practical$Support_Practical %in% seq(1,3),] #remove options 4&5
Support_Practical[Support_Practical$Support_Practical %in% c(1,2), "Support_Practical"]="No"
Support_Practical[Support_Practical$Support_Practical==3, "Support_Practical"]="Yes"
#ImpactPersonal$Impact_Personal=factor(ImpactPersonal$Impact_Personal, levels=c("No","Yes"))
graphdata =Support_Practical %>%
group_by(All_Gender_clean, Support_Practical) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'All_Gender_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/SupportPractical_gender.eps", width=1000, height=578)
ggplot(graphdata, aes(x=All_Gender_clean, y=perc, fill=Support_Practical)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=12), axis.title.x = element_text(size = 8), axis.title.y = element_text(size = 16), legend.position="bottom", legend.text = element_text(size=7))+
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="
I needed practical information \n(e.g. advice, guidelines, training...) \nto help me in my supporting role.") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=c(eBlue, eGreen))
ggsave(dpi=1000, "6-2-2 (not reported).png", limitsize = FALSE)## Saving 7 x 5 in imagedev.off()## null device
## 1Needing practical information x CS graph (Graph 6.2.2)
graphdata = Support_Practical %>%
group_by(Supporter_CareerStage_clean, Support_Practical) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'Supporter_CareerStage_clean' (override with `.groups` argument)# This will save the image to your local code folder
#eps("images/SupportPractical_CS.eps", width=1000, height=578)
ggplot(graphdata, aes(x=Supporter_CareerStage_clean, y=perc, fill=Support_Practical)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=12), axis.title.x = element_text(size = 8), axis.title.y = element_text(size = 16), legend.text = element_text(size=7))+
geom_text(aes(label=round(perc, digit=1)), size=3, position=position_stack(vjust=0.5), color="white") +
scale_x_discrete(limits = rev(levels(graphdata$Supporter_CareerStage_clean))) +
labs(x="", y="Percentage", title="I needed practical information \n(e.g. advice, guidelines, training...) to help me \nin my supporting role.") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=c(eBlue, eGreen))
ggsave(dpi=1000, "6-2-2.png", limitsize = FALSE)## Saving 7 x 5 in imagedev.off()## null device
## 1Model: Multinomial logistic regression (not reported)
Makes the most sense to use “Yes, I needed information” (option 3) as the baseline. This tests whether gender/CS has an effect of a respondent choosing the other two responses (“No, I did not need information” and “No, I had already received information before”) over the baseline.
#subset
Support_Practical=genderCS[,c("Support_Practical", "Supporter_CareerStage_clean", "All_Gender_clean")]
Support_Practical=Support_Practical[Support_Practical$Support_Practical %in% seq(1,3), ] #remove NAs and some choices
dim(Support_Practical)[1] #N## [1] 1003Support_Practical$Support_Practical=as.factor(Support_Practical$Support_Practical) #factorise
#reorder factor to make "Yes, I needed information" the baseline
Support_Practical$Support_Practical=relevel(Support_Practical$Support_Practical, "3")
levels(Support_Practical$Support_Practical)= c("Yes, I needed information", "No, I did not need information", "No, I had already received information before")
#multinomial model -> p-values (2-tailed z test)
# Effect of gender
SP_gender=multinom(Support_Practical~All_Gender_clean, data=Support_Practical)## # weights: 9 (4 variable)
## initial value 1101.908126
## iter 10 value 835.460887
## iter 10 value 835.460886
## final value 835.460886
## convergedsummary(SP_gender)## Call:
## multinom(formula = Support_Practical ~ All_Gender_clean, data = Support_Practical)
##
## Coefficients:
## (Intercept) All_Gender_cleanWomen
## No, I did not need information -1.055119 -0.4365019
## No, I had already received information before -1.725203 0.1043311
##
## Std. Errors:
## (Intercept) All_Gender_cleanWomen
## No, I did not need information 0.1252083 0.1674830
## No, I had already received information before 0.1636321 0.2013519
##
## Residual Deviance: 1670.922
## AIC: 1678.922# Z-statistics and p-value using Wald tests
SP_gender_z=summary(SP_gender)$coefficients/summary(SP_gender)$standard.errors #Z statistics
(1-pnorm(abs(SP_gender_z), 0, 1))*2 #p-values## (Intercept) All_Gender_cleanWomen
## No, I did not need information 0 0.009154068
## No, I had already received information before 0 0.604351393exp(coef(SP_gender)) #effect size (baseline IV is men, and baseline response is "yes")## (Intercept) All_Gender_cleanWomen
## No, I did not need information 0.3481512 0.6462933
## No, I had already received information before 0.1781369 1.1099679exp(confint(SP_gender)) ## , , No, I did not need information
##
## 2.5 % 97.5 %
## (Intercept) 0.2723895 0.4449849
## All_Gender_cleanWomen 0.4654444 0.8974111
##
## , , No, I had already received information before
##
## 2.5 % 97.5 %
## (Intercept) 0.1292617 0.2454922
## All_Gender_cleanWomen 0.7480307 1.6470298# Effect of career stage
SP_CS=multinom(Support_Practical~Supporter_CareerStage_clean, data=Support_Practical)## # weights: 18 (10 variable)
## initial value 1101.908126
## iter 10 value 830.236732
## final value 829.937190
## convergedsummary(SP_CS)## Call:
## multinom(formula = Support_Practical ~ Supporter_CareerStage_clean,
## data = Support_Practical)
##
## Coefficients:
## (Intercept)
## No, I did not need information -1.202753
## No, I had already received information before -1.566422
## Supporter_CareerStage_cleanPostdocs
## No, I did not need information -0.1203630
## No, I had already received information before -0.1520059
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## No, I did not need information -0.5200135
## No, I had already received information before -1.0726390
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## No, I did not need information -0.4264911
## No, I had already received information before 0.4071852
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## No, I did not need information 0.1761153
## No, I had already received information before 0.1949468
##
## Std. Errors:
## (Intercept)
## No, I did not need information 0.1259291
## No, I had already received information before 0.1456258
## Supporter_CareerStage_cleanPostdocs
## No, I did not need information 0.2042085
## No, I had already received information before 0.2386308
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## No, I did not need information 0.2734716
## No, I had already received information before 0.3938729
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## No, I did not need information 0.3680575
## No, I had already received information before 0.3214260
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## No, I did not need information 0.2691657
## No, I had already received information before 0.3081486
##
## Residual Deviance: 1659.874
## AIC: 1679.874# Z-statistics and p-value using Wald tests
SP_CS_z=summary(SP_CS)$coefficients/summary(SP_CS)$standard.errors #Z statistics
(1-pnorm(abs(SP_CS_z), 0, 1))*2 #p-values## (Intercept)
## No, I did not need information 0
## No, I had already received information before 0
## Supporter_CareerStage_cleanPostdocs
## No, I did not need information 0.5555847
## No, I had already received information before 0.5241303
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## No, I did not need information 0.057233076
## No, I had already received information before 0.006463086
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## No, I did not need information 0.2465531
## No, I had already received information before 0.2052237
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## No, I did not need information 0.5129182
## No, I had already received information before 0.5269696exp(coef(SP_CS)) #effect size## (Intercept)
## No, I did not need information 0.3003661
## No, I had already received information before 0.2087910
## Supporter_CareerStage_cleanPostdocs
## No, I did not need information 0.8865986
## No, I had already received information before 0.8589833
## Supporter_CareerStage_cleanGroup leaders (<5yr)
## No, I did not need information 0.5945125
## No, I had already received information before 0.3421045
## Supporter_CareerStage_cleanGroup leaders (5-10yr)
## No, I did not need information 0.6527957
## No, I had already received information before 1.5025824
## Supporter_CareerStage_cleanGroup leaders (>10yr)
## No, I did not need information 1.192576
## No, I had already received information before 1.215246exp(confint(SP_CS))## , , No, I did not need information
##
## 2.5 % 97.5 %
## (Intercept) 0.2346712 0.3844518
## Supporter_CareerStage_cleanPostdocs 0.5941615 1.3229687
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.3478417 1.0161094
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.3173118 1.3429764
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.7036742 2.0211576
##
## , , No, I had already received information before
##
## 2.5 % 97.5 %
## (Intercept) 0.1569476 0.2777593
## Supporter_CareerStage_cleanPostdocs 0.5380985 1.3712215
## Supporter_CareerStage_cleanGroup leaders (<5yr) 0.1580861 0.7403277
## Supporter_CareerStage_cleanGroup leaders (5-10yr) 0.8002764 2.8212176
## Supporter_CareerStage_cleanGroup leaders (>10yr) 0.6643053 2.2231098multinomial LR - Career stages pairwise analyses
using multinomial to compare between all career stages, pairwise
#reverse CSpairs column order to not have early PIs on the baseline
CSpairs=CSpairs[,c(2,1)]
#new dataframe
SupportPracticalCSMLR=data.frame(CS1=c(), CS2=c(), response=c(), z=c(), edf=c(), pVal=c(), OR=c(), CIlower=c(), CIupper=c())
for (i in 1:dim(CSpairs)[1]) {
comparedData=Support_Practical[(Support_Practical$Supporter_CareerStage_clean==CSpairs[i,1]|Support_Practical$Supporter_CareerStage_clean==CSpairs[i,2]),]
comparedData$Supporter_CareerStage_clean=droplevels(comparedData$Supporter_CareerStage_clean) #drop levels to avoid levels with 0 values producing NaNs
comparedData$Supporter_CareerStage_clean=relevel(comparedData$Supporter_CareerStage_clean, CSpairs[i,1]) #put the baseline first in factor level!
model=multinom(Support_Practical~Supporter_CareerStage_clean, data=comparedData)
# Z-statistics and p-value using Wald tests
z=summary(model)$coefficients/summary(model)$standard.errors #Z statistics- from this line onwards, code produces 4 values, one for each non-baseline response, and intercept and CS2 over CS1. We ignore the intercept values (recording column 2)
pVal=(1-pnorm(abs(z), 0, 1))*2 #p-values
or=exp(coef(model)) #effect size
CI=exp(confint(model)) #CI - [3,4] for first repsonse, [7,8] for second
SupportPracticalCSMLR=rbind(SupportPracticalCSMLR,
data.frame(CS1=CSpairs[i,1], CS2=CSpairs[i,2], response=rownames(z)[1], z=z[1,2], df=model$edf, pVal=pVal[1,2], OR=or[1,2], CIlower=CI[3], CIupper=CI[4]),
data.frame(CS1=CSpairs[i,1], CS2=CSpairs[i,2], response=rownames(z)[2], z=z[2,2], df=model$edf, pVal=pVal[2,2], OR=or[2,2], CIlower=CI[7], CIupper=CI[8]))
}## # weights: 9 (4 variable)
## initial value 606.433983
## iter 10 value 444.278831
## final value 444.278808
## converged
## # weights: 9 (4 variable)
## initial value 446.036589
## iter 10 value 306.419345
## final value 306.385782
## converged
## # weights: 9 (4 variable)
## initial value 238.398867
## iter 10 value 153.370530
## final value 153.370528
## converged
## # weights: 9 (4 variable)
## initial value 272.455848
## iter 10 value 186.325810
## final value 186.325736
## convergedSupportPracticalCSMLR## CS1 CS2
## 1 PhD students Group leaders (<5yr)
## 2 PhD students Group leaders (<5yr)
## 3 Postdocs Group leaders (<5yr)
## 4 Postdocs Group leaders (<5yr)
## 5 Group leaders (5-10yr) Group leaders (<5yr)
## 6 Group leaders (5-10yr) Group leaders (<5yr)
## 7 Group leaders (>10yr) Group leaders (<5yr)
## 8 Group leaders (>10yr) Group leaders (<5yr)
## response z df pVal
## 1 No, I did not need information -1.9006683 4 0.057345477
## 2 No, I had already received information before -2.7228910 4 0.006471340
## 3 No, I did not need information -1.3726343 4 0.169866053
## 4 No, I had already received information before -2.2350517 4 0.025413953
## 5 No, I did not need information -0.2215956 4 0.824628732
## 6 No, I had already received information before -3.1832592 4 0.001456272
## 7 No, I did not need information -2.0479815 4 0.040561817
## 8 No, I had already received information before -2.7814155 4 0.005412242
## OR CIlower CIupper
## 1 0.5946760 0.3844602 1.0163466
## 2 0.3421955 0.2777661 0.7404714
## 3 0.6705548 0.3649348 1.1864957
## 4 0.3982676 0.2597834 0.8928736
## 5 0.9106214 0.3862392 2.0844225
## 6 0.2277518 0.5500803 0.5663421
## 7 0.4985416 0.5709868 0.9705276
## 8 0.2815239 0.4320394 0.6877366Q32c All_MH
Q32c All_MH x gender x Supporter Status (no career stage here)
Looking at 2 independent variables:
Gender (label: “All_Gender_clean”): Man -1, Woman -3
Supporter/ Not supporter (label “Provided_Support”):
recode to new variable “Provided_Support_clean”
- 1- No - recode to “No”
- 2 - Yes, to one individual - recode to “Yes”
- 3 - Yes, to between one and five individuals - recode to “Yes”
- 4 - Yes, to more than five individuals - recode to “Yes”
- 5 - I am not sure - remove
- 6 - Prefer not to answer - remove
#Filtering by gender and supporter status
genderSupport=data[data$All_Gender_clean %in% c(1,3) & data$Provided_Support %in% seq(1,4),]
#recode Supporter/not Supporter - Provided_Support_clean
genderSupport$Provided_Support_clean=genderSupport$Provided_Support
genderSupport[genderSupport$Provided_Support==1, "Provided_Support_clean"]="No"
genderSupport[genderSupport$Provided_Support %in% seq(2,4), "Provided_Support_clean"]="Yes"
genderSupport$All_Gender_clean=as.factor(genderSupport$All_Gender_clean)
genderSupport$Provided_Support_clean=as.factor(genderSupport$Provided_Support_clean)
levels(genderSupport$All_Gender_clean)=c("Men", "Women")
dim(genderSupport)## [1] 1697 197Q01xQ28 Supporter vs gender (Section 2, Panel 1)
Chi-square & Logistical regression
gender=genderSupport[,c("Provided_Support_clean","All_Gender_clean")]
dim(gender)[1] #this gives N## [1] 1697table(gender)## All_Gender_clean
## Provided_Support_clean Men Women
## No 100 110
## Yes 530 957chisq.test(table(gender)) #not reported##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: table(gender)
## X-squared = 10.801, df = 1, p-value = 0.001015summary(glm(Provided_Support_clean~All_Gender_clean, data=genderSupport, family="binomial")) ##
## Call:
## glm(formula = Provided_Support_clean ~ All_Gender_clean, family = "binomial",
## data = genderSupport)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -2.1317 0.4665 0.4665 0.5879 0.5879
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 1.6677 0.1090 15.30 < 2e-16 ***
## All_Gender_cleanWomen 0.4956 0.1484 3.34 0.000838 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1270.5 on 1696 degrees of freedom
## Residual deviance: 1259.4 on 1695 degrees of freedom
## AIC: 1263.4
##
## Number of Fisher Scoring iterations: 4exp(coef(glm(Provided_Support_clean~All_Gender_clean, data=genderSupport, family="binomial"))) #OR (men as baseline)## (Intercept) All_Gender_cleanWomen
## 5.300000 1.641509exp(confint(glm(Provided_Support_clean~All_Gender_clean, data=genderSupport, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 4.301614 6.598465
## All_Gender_cleanWomen 1.226278 2.195431filtering by response to All_MH
ordinal logistical regression: 1-5 strongly disagree - strongly agree
AllMH=genderSupport[,c("All_MH", "All_Gender_clean", "Provided_Support_clean")]
#clean
AllMH=AllMH[(AllMH$All_MH %in% seq(1:5)),]
dim(AllMH)[1]## [1] 1676#factorise
AllMH$All_MH=as.factor(AllMH$All_MH)
table(AllMH$All_MH)##
## 1 2 3 4 5
## 88 79 87 357 1065All_MH x gender (Section 1, last paragraph)
exp(coef(polr(All_MH~All_Gender_clean, data=AllMH, Hess=TRUE, method="logistic"))) #OR (men as baseline)## All_Gender_cleanWomen
## 1.748233exp(confint(polr(All_MH~All_Gender_clean, data=AllMH, Hess=TRUE, method="logistic"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## 1.433968 2.131327kruskal.test(All_MH ~ All_Gender_clean, data=AllMH) #effect of gender on response##
## Kruskal-Wallis rank sum test
##
## data: All_MH by All_Gender_clean
## Kruskal-Wallis chi-squared = 30.623, df = 1, p-value = 3.133e-08All_MH x Supporter
exp(coef(polr(All_MH~Provided_Support_clean, data=AllMH, Hess=TRUE, method="logistic"))) #OR (non-supporters are baseline)## Provided_Support_cleanYes
## 1.263584kruskal.test(All_MH ~Provided_Support_clean, data=AllMH) #effect of supporter status on response##
## Kruskal-Wallis rank sum test
##
## data: All_MH by Provided_Support_clean
## Kruskal-Wallis chi-squared = 2.5083, df = 1, p-value = 0.1133All_MH x gender graph
##reordering factors order for the graph
levels(AllMH$All_MH)=c("Strongly disagree", "Somewhat disagree", "Neither agree or disagree", "Somewhat agree", "Strongly agree")
AllMH$All_MH=factor(AllMH$All_MH, levels=c("Strongly agree", "Somewhat agree", "Neither agree or disagree", "Somewhat disagree", "Strongly disagree"))
#graph split by gender
graphdata = AllMH %>%
group_by(All_Gender_clean, All_MH) %>%
summarize(n=n()) %>%
mutate(perc=n*100/sum(n))## `summarise()` regrouping output by 'All_Gender_clean' (override with `.groups` argument)#tiff("AllMH_gender.tiff", width=1000, height=578)
ggplot(graphdata, aes(x=All_Gender_clean, y=perc, fill=All_MH)) +
geom_bar(stat="identity") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5), text=element_text(size=20))+
geom_text(aes(label=round(perc, digit=1)), size=4, position=position_stack(vjust=0.5), color="white") +
labs(x="", y="Percentage", title="I have gone through times in my life when my mental health was poor") +
guides(fill=guide_legend(reverse=TRUE)) +
coord_flip() +
scale_fill_manual(name="", values=ePalette)
#dev.off()All_MH x Supporter
exp(coef(polr(All_MH~Provided_Support_clean, data=AllMH, Hess=TRUE, method="logistic"))) #OR (non-supporters are baseline)## Provided_Support_cleanYes
## 0.7913638kruskal.test(All_MH ~Provided_Support_clean, data=AllMH) #effect of supporter status on response##
## Kruskal-Wallis rank sum test
##
## data: All_MH by Provided_Support_clean
## Kruskal-Wallis chi-squared = 2.5083, df = 1, p-value = 0.1133Chisq for intersections [NOT REPORTED]
# #split by supporter status, gender effect
# summary(AllMH$Provided_Support)
# chisq.test(table(AllMH[AllMH$Provided_Support_clean==levels(AllMH$Provided_Support_clean)[1],c("All_MH","All_Gender_clean")]))
# chisq.test(table(AllMH[AllMH$Provided_Support_clean==levels(AllMH$Provided_Support_clean)[2],c("All_MH","All_Gender_clean")]))
# #split by gender, supporter status (same as KW above)
# chisq.test(table(AllMH[AllMH$All_Gender_clean==levels(AllMH$All_Gender_clean)[1],c("All_MH","Provided_Support_clean")]))
# chisq.test(table(AllMH[AllMH$All_Gender_clean==levels(AllMH$All_Gender_clean)[2],c("All_MH","Provided_Support_clean")]))Q32d All_UnderstandingMH (Section 1, last paragraph)
ordinal logistical regression: 1-5 strongly disagree - strongly agree
AllUnderstandingMH=genderSupport[,c("All_UnderstandingMH", "All_Gender_clean", "Provided_Support_clean")]
#clean
AllUnderstandingMH=AllUnderstandingMH[(AllUnderstandingMH$All_UnderstandingMH %in% seq(1:5)),]
dim(AllUnderstandingMH)[1]## [1] 1685#factorise
AllUnderstandingMH$All_UnderstandingMH=as.factor(AllUnderstandingMH$All_UnderstandingMH)
table(AllUnderstandingMH$All_UnderstandingMH)##
## 1 2 3 4 5
## 45 142 248 745 505All_UnderstandingMH x gender
exp(coef(polr(All_UnderstandingMH~All_Gender_clean, data=AllUnderstandingMH, Hess=TRUE, method="logistic"))) #OR (men as baseline)## All_Gender_cleanWomen
## 1.645369exp(confint(polr(All_UnderstandingMH~All_Gender_clean, data=AllUnderstandingMH, Hess=TRUE, method="logistic"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## 1.369448 1.977934kruskal.test(All_UnderstandingMH ~ All_Gender_clean, data=AllUnderstandingMH) #effect of gender on response##
## Kruskal-Wallis rank sum test
##
## data: All_UnderstandingMH by All_Gender_clean
## Kruskal-Wallis chi-squared = 28.132, df = 1, p-value = 1.133e-07All_UnderstandingMH x Supporter
exp(coef(polr(All_UnderstandingMH~Provided_Support_clean, data=AllUnderstandingMH, Hess=TRUE, method="logistic"))) #OR (non-supporters as baseline)## Provided_Support_cleanYes
## 1.143644kruskal.test(All_UnderstandingMH ~Provided_Support_clean, data=AllUnderstandingMH) #effect of supporter status on response##
## Kruskal-Wallis rank sum test
##
## data: All_UnderstandingMH by Provided_Support_clean
## Kruskal-Wallis chi-squared = 0.94881, df = 1, p-value = 0.33Q32c All_MH x career stage
Looking at the single indepedent variable - present career stage. This is because this section focuses on understanding the population that filled in the survey
- CS (label: “All_CurrentRole_clean”): PhD - 3, postdoc -4, earlyPI - 6, midPI- 7, late PI -8 drop all other stages
AllCS=data[data$All_CurrentRole_clean %in% c(3,4,6,7,8), ] #subset
AllCS$All_CurrentRole_clean=as.factor(AllCS$All_CurrentRole_clean) #factorise
dim(AllCS) ## [1] 1501 196levels(AllCS$All_CurrentRole_clean)=c("PhD students", "Postdocs", "Group leaders (<5yr)", "Group leaders (5-10yr)", "Group leaders (>10yr)")Filter All_MH response as before
#subset
AllMH_CS=AllCS[,c("All_MH", "All_CurrentRole_clean")]
#clean
AllMH_CS=AllMH_CS[(AllMH_CS$All_MH %in% seq(1:5)),]
dim(AllMH_CS)[1] #N## [1] 1488#factorise
AllMH_CS$All_MH=as.factor(AllMH_CS$All_MH)
table(AllMH_CS$All_MH)##
## 1 2 3 4 5
## 85 66 78 331 928All_MH x CS (Section 1, last paragraph)
ordinal logistic regression, single Indepedent variable
exp(coef(polr(All_MH~All_CurrentRole_clean, data=AllMH_CS, Hess=TRUE, method="logistic"))) #OR (men as baseline)## All_CurrentRole_cleanPostdocs
## 0.8131627
## All_CurrentRole_cleanGroup leaders (<5yr)
## 0.4493026
## All_CurrentRole_cleanGroup leaders (5-10yr)
## 0.3492036
## All_CurrentRole_cleanGroup leaders (>10yr)
## 0.1881167exp(confint(polr(All_MH~All_CurrentRole_clean, data=AllMH_CS, Hess=TRUE, method="logistic"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## All_CurrentRole_cleanPostdocs 0.6168290 1.0699657
## All_CurrentRole_cleanGroup leaders (<5yr) 0.3253815 0.6203544
## All_CurrentRole_cleanGroup leaders (5-10yr) 0.2359750 0.5180630
## All_CurrentRole_cleanGroup leaders (>10yr) 0.1326973 0.2659600kruskal.test(All_MH ~ All_CurrentRole_clean, data=AllMH_CS) #effect of CS on response##
## Kruskal-Wallis rank sum test
##
## data: All_MH by All_CurrentRole_clean
## Kruskal-Wallis chi-squared = 114.25, df = 4, p-value < 2.2e-16Q1x Q32 Supporter vs CS
chi-square (not reported) and logistic regression
#subset
CSSupport=AllCS[,c("Provided_Support","All_CurrentRole_clean")]
CSSupport$Provided_Support_clean=CSSupport$Provided_Support
CSSupport=CSSupport[!is.na(CSSupport$Provided_Support),] #remove NA
CSSupport[CSSupport$Provided_Support==1, 'Provided_Support_clean']=0
CSSupport[CSSupport$Provided_Support %in% seq(2,4), 'Provided_Support_clean']=1
CSSupport=CSSupport[CSSupport$Provided_Support_clean %in% c(0,1), c("Provided_Support_clean", "All_CurrentRole_clean")]
dim(CSSupport)[1] #this gives N## [1] 1391table(CSSupport)## All_CurrentRole_clean
## Provided_Support_clean PhD students Postdocs Group leaders (<5yr)
## 0 64 52 25
## 1 345 423 188
## All_CurrentRole_clean
## Provided_Support_clean Group leaders (5-10yr) Group leaders (>10yr)
## 0 11 7
## 1 109 167chisq.test(table(CSSupport)) #not reported##
## Pearson's Chi-squared test
##
## data: table(CSSupport)
## X-squared = 17.353, df = 4, p-value = 0.00165summary(glm(Provided_Support_clean~All_CurrentRole_clean, data=CSSupport, family="binomial")) ##
## Call:
## glm(formula = Provided_Support_clean ~ All_CurrentRole_clean,
## family = "binomial", data = CSSupport)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -2.5350 0.4385 0.4815 0.4997 0.5834
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) 1.6847 0.1361 12.378
## All_CurrentRole_cleanPostdocs 0.4115 0.2003 2.054
## All_CurrentRole_cleanGroup leaders (<5yr) 0.3329 0.2527 1.318
## All_CurrentRole_cleanGroup leaders (5-10yr) 0.6088 0.3444 1.768
## All_CurrentRole_cleanGroup leaders (>10yr) 1.4874 0.4090 3.637
## Pr(>|z|)
## (Intercept) < 2e-16 ***
## All_CurrentRole_cleanPostdocs 0.039947 *
## All_CurrentRole_cleanGroup leaders (<5yr) 0.187657
## All_CurrentRole_cleanGroup leaders (5-10yr) 0.077108 .
## All_CurrentRole_cleanGroup leaders (>10yr) 0.000276 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 988.79 on 1390 degrees of freedom
## Residual deviance: 969.27 on 1386 degrees of freedom
## AIC: 979.27
##
## Number of Fisher Scoring iterations: 5exp(coef(glm(Provided_Support_clean~All_CurrentRole_clean, data=CSSupport, family="binomial"))) #OR (PhD as baseline)## (Intercept)
## 5.390625
## All_CurrentRole_cleanPostdocs
## 1.509030
## All_CurrentRole_cleanGroup leaders (<5yr)
## 1.395014
## All_CurrentRole_cleanGroup leaders (5-10yr)
## 1.838208
## All_CurrentRole_cleanGroup leaders (>10yr)
## 4.425672exp(confint(glm(Provided_Support_clean~All_CurrentRole_clean, data=CSSupport, family="binomial"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## (Intercept) 4.1604363 7.099635
## All_CurrentRole_cleanPostdocs 1.0203209 2.241046
## All_CurrentRole_cleanGroup leaders (<5yr) 0.8600750 2.324661
## All_CurrentRole_cleanGroup leaders (5-10yr) 0.9714635 3.794156
## All_CurrentRole_cleanGroup leaders (>10yr) 2.1211088 10.800584Q32d All_UnderstandingMH x career stage (Section 1, last paragraph)
ordinal logistical regression: 1-5 strongly disagree - strongly agree
AllUnderstandingMH_CS=AllCS[,c("All_UnderstandingMH", "All_CurrentRole_clean")]
#clean
AllUnderstandingMH_CS=AllUnderstandingMH_CS[(AllUnderstandingMH_CS$All_UnderstandingMH %in% seq(1:5)),]
dim(AllUnderstandingMH_CS)[1]## [1] 1496#factorise
AllUnderstandingMH_CS$All_UnderstandingMH=as.factor(AllUnderstandingMH_CS$All_UnderstandingMH)
table(AllUnderstandingMH_CS$All_UnderstandingMH)##
## 1 2 3 4 5
## 42 135 220 667 432exp(coef(polr(All_UnderstandingMH~All_CurrentRole_clean, data=AllUnderstandingMH_CS, Hess=TRUE, method="logistic"))) #OR (PhD as baseline)## All_CurrentRole_cleanPostdocs
## 0.8415217
## All_CurrentRole_cleanGroup leaders (<5yr)
## 0.7849314
## All_CurrentRole_cleanGroup leaders (5-10yr)
## 0.5921041
## All_CurrentRole_cleanGroup leaders (>10yr)
## 0.7440227exp(confint(polr(All_UnderstandingMH~All_CurrentRole_clean, data=AllUnderstandingMH_CS, Hess=TRUE, method="logistic"))) #CI## Waiting for profiling to be done...## 2.5 % 97.5 %
## All_CurrentRole_cleanPostdocs 0.6666041 1.0619965
## All_CurrentRole_cleanGroup leaders (<5yr) 0.5843670 1.0543121
## All_CurrentRole_cleanGroup leaders (5-10yr) 0.4125412 0.8500038
## All_CurrentRole_cleanGroup leaders (>10yr) 0.5376563 1.0298044kruskal.test(All_UnderstandingMH ~ All_CurrentRole_clean, data=AllUnderstandingMH_CS) #effect of CS on response##
## Kruskal-Wallis rank sum test
##
## data: All_UnderstandingMH by All_CurrentRole_clean
## Kruskal-Wallis chi-squared = 9.7721, df = 4, p-value = 0.04445