There is considerable uncertainty regarding the impact of tableware size on food consumption. Most existing studies have used small and unrepresentative samples and have not followed recommended procedures for randomised controlled trials, leading to increased risk of bias. In the first pre-registered study to date, we examined the impact on consumption of using larger versus smaller plates for self-served food. We also assessed impact on the underlying meal micro-structure, such as number of servings and eating rate, which has not previously been studied.
The setting was a purpose-built naturalistic eating behaviour laboratory. A general population sample of 134 adult participants (aged 18–61 years) was randomly allocated to one of two groups varying in the size of plate used for self-serving lunch: large or small. The primary outcome was amount of food energy (kcal) consumed during a meal. Additionally, we assessed impact on meal micro-structure, and examined potential modifying effects of executive function, socio-economic position, and sensitivity to perceptual cues.
There was no clear evidence of a difference in consumption between the two groups: Cohen’s
This study suggests that previous meta-analyses of a low-quality body of evidence may have considerably overestimated the effects of plate size on consumption. However, the possibility of a clinically significant effect – in either direction – cannot be excluded. Well-conducted trials of tableware size in real-world field settings are now needed to determine whether changing the size of tableware has potential to contribute to efforts to reduce consumption at population-level.
The study protocol (
The online version of this article (10.1186/s12966-019-0826-1) contains supplementary material, which is available to authorized users.
If people consumed less food and energy it would help to prevent weight gain and reduce the global burden of non-communicable diseases such as cardiovascular disease, cancer, and type 2 diabetes [
Recent systematic reviews and meta-analyses have reached slightly different conclusions about whether the size of the tableware – for example, plates or bowls – has an effect on the amount of food that is consumed. Robinson and colleagues [
Furthermore, a substantial proportion of the studies common to these reviews have been conducted by researchers at the Cornell University Food and Brand Lab, whose work has recently been subject to scrutiny for possible scientific misconduct [
In sum, there remains considerable uncertainty about the effects of plate size on food consumption, and high-quality pre-registered studies are needed to address this. Here we focused on consumption when food was self-served, as previous literature has suggested that this is where effects are most likely to be observed. The primary research question for this study was therefore: What is the impact on consumption of using larger (versus smaller) plates for self-serving food? In line with the weight of current review evidence, we hypothesised that using larger (versus smaller) tableware when self-serving food increases consumption of that food.
In addition, the potential mechanisms underlying any effect of plate size are not well understood. In particular, to our knowledge, no studies of the effect of plate size have examined possible impact on meal micro-structure – the pattern of behaviours that occur within an eating episode, such as number of helpings served or eating rate – that could explain any observed effects on amount consumed. In exploring potential mechanisms, the current study also examined possible modifiers of any observed intervention effect of plate size, with a secondary research question asking whether
Ethics approval was obtained from the Cambridge Psychology Research Ethics Committee (PRE.2018.036). The study protocol was pre-registered on the Open Science Framework (
In a two-group between-subjects design, participants were randomly allocated to self-serve lunch and eat from either a
A general population sample varying in socio-economic position (SEP) was recruited via a research agency during the study period (August to November 2018). The sample purposefully comprised similar proportions of men and women and similar proportions of participants with lower and higher education level (no Bachelor’s degree versus Bachelor’s degree or higher). The following inclusion criteria were applied: age 18–60; sufficient English and computer literacy skills to complete the study. The exclusion criteria comprised: disliking or restricting of the test foods [e.g. food allergies, vegan]; having a diagnosed eating disorder or taking prescribed medication that could considerably influence eating behaviour; performing vigorous exercise for more than 10 h a week; active smokers. Participants were reimbursed £50 for completing the study.
Sample size was determined based on the most recent meta-analysis of tableware size and consumption [
The study was conducted in the Wellcome-MRC Translational Research Facility (TRF) in Cambridge, UK, a purpose-built eating behaviour facility that includes rooms designed to replicate eating environments in people’s homes (see Procedure for further details).
White, circular, unmarked plates with a 46% difference in surface area were used for self-serving food: Large plate: China by Denby Dinner Plate (29 cm diameter, surface area = 660.5 cm2); Small plate: China by Denby Dessert/Salad Plate (23 cm diameter, surface area = 415.5 cm2). The plates were purposefully selected from a product range with plain, featureless designs, and that offered both a standard dinner plate and a smaller meal plate that was otherwise identical. These plate types were chosen due to public and research interest in the potential impact on consumption of reducing the size of dinner plates.
A vegetarian cheese and tomato pasta bake (150.33 kcal per 100 g) was the default food provided. If a participant was unwilling to consume it for any reason, chicken korma curried rice (129.6 kcal per 100 g), matched in nutritional characteristics to the pasta dish, was offered as an alternative.
An A&D GX6100 6100 g × 0.01 g laboratory balance was used to weigh the study food.
Two cameras were used to record the eating session (Silverlabel Focus Action Cam Ultra HD; Go Pro Hero 3). Participants consented to being filmed within the facility although the cameras were concealed throughout.
The amount of food consumed was measured in grams, weighing the food taken from the serving dish minus the amount of food left on the participant’s plate. This was translated into energy consumed (kcal), which is the measure used in the analysis.
While eating, participants were filmed using two concealed cameras, one of which was directly pointed towards the front of the participant’s chair. Two researchers (DK and LK) independently verified the number of helpings and number of bites from the videos, and interrater agreement was calculated (
We assessed hedonic experience with a single item question “How did you like the food?” with an answer on a 100 mm VAS [
To conceal the true purpose of the study, at the time of recruitment, participants were informed that the study was examining the impact of time of the day on a range of mental processes, and that they had been allocated to a lunchtime session. All data were collected in individual experimental sessions that took place between approximately 12:00 to 14:45 to ensure similar hunger levels. Participants were instructed to consume one of three suggested breakfasts and fast for at least three hours before the study session. After providing written informed consent, participants initially completed a series of baseline measures in a testing room, including the computer-based cognitive tasks. Participants were then guided to a lounge room with a small dining table, a chair, a sofa, bookshelves, a television showing a standardised nature programme, and a heated food trolley with food (see Additional file
Analysis was conducted by an analyst not involved in the collection of the data and who was blinded to allocation. Descriptive statistics were calculated for baseline characteristics of participants in the two plate size conditions. SPSS 25 software was used for data analysis.
The primary outcome (energy consumed) was analysed using an independent samples t-test. Sensitivity analysis was performed after outliers for the primary outcome were removed (i.e. participants who exceeded a distance of 1.5 times the interquartile range (IQR) below the first quartile (applying to no participants) or 1.5 times the IQR above the third quartile (equating to 1250 kcal and applying to five participants). Mean differences and effect size (Cohen’s
A two-way analysis of variance (ANOVA) was applied to estimate the impact of the effect modifiers (executive function, SEP and sensitivity to perceptual cues), and the interaction between plate size and each dichotomised covariate in turn. Executive function variables and sensitivity to perceptual cues were dichotomised at the median. For SEP measures, income and Index of Multiple Deprivation (IMD) were split at the median, while highest qualification was dichotomised as ‘Bachelor’s degree-level and higher’ versus lower. Interrater agreement for meal micro-structure measures was determined using intraclass correlation coefficients (ICC) in R (version 3.5.1) with package irr (version 0.84) [
A total of 134 participants consented to participate and were randomised. Mean age was 35.9 ( Participant characteristics Measure Small Plate ( Large Plate (n = 67) Gender, Men, n (%) 33 (49%) 32 (48%) Women, n (%) 33 (49%) 35 (52%) Other, n (%) 1 (1%) 0 Age (Mean (SD)), n = 134 35.9 (12.9) 35.9 Ethnicity, White, n (%) 57 (86%) 57 (86%) Non-white, n (%) 9 (14%) 9 (14%) BMI (Mean (SD)), n = 134 27.3 (4.9) 26.5 (4.7) Highest qualificationa, n = 132 No qualifications 3 (5%) 3 (5%) Up to 4 GCSE’s 7 (11%) 7 (10%) 5 or more GCSE’s or 1 A-level 10 (15%) 11 (17%) 2 or more A-levels 12 (18%) 9 (14%) Bachelor’s degree 16 (24%) 16 (24%) Post-Graduate degree or qualification 18 (27%) 20 (30%) Income per year before tax, Lower income (Up to £39,999)a, n (%) 29 (46%) 32 (51%) Higher income (£40,000 and more), n (%) 34 (54%) 31 (49%) IMDb (Mean (SD)), 12.6 (7.7) 11.6 (7.2) Hunger (Mean (SD)), n = 134 52.4 (25.8) 50.1 (23.7) Fullness (Mean (SD)), n = 134 26.7 (24.5) 27.9 (21.8) aLower/higher income categorisation is equivalent to average household income below, and above, ~$50,000 or €45,000 bIndex of Multiple Deprivation CONSORT Flow Diagram
There was no clear evidence of a difference in calories (kcal) consumed between the plate size groups:
The only meal-microstructure measure that differed between the two groups was the amount of food left on the plate after the meal, with participants using large plates leaving 8.6 g (95% CI [1.1, 16.0]) more food. No other group differences were found for meal micro-structure measures ( Primary outcome and meal micro-structure Small plate Large plate Mean ( Mean ( Mean Difference 95% CI Cohen‘s Kilocalories consumed (kcal) 624.9 (292.3) 644.1 (265.0) 19.2 −76.5, 115.0 0.692 0.07 Meal micro-structure Meal duration (min) 8.6 (4.7) 8.7 (4.3) 0.0 −1.5, 1.6 0.953 0.01 Number of servings (n) 1.8 (0.8) 1.7 (0.7) −0.2 - 0.4, 0.1 0.186 −0.24 Amount self-served (g) 422.6 (192.7) 443.1 (174.0) 20.5 −42.5, 83.5 0.521 0.11 Amount of food left on the plate (g) 4.9 (10.5) 13.5 (28.9) 8.6 1.1, 16.0 0.024 0.40 Number of bites (n) 41.6 (23.6) 38.3 (17.0) −3.3 −10.5, 3.9 0.366 −0.16 Average bite size (g) 11.6 (5.0) 11.7 (4.3) 0.2 −1.5, 1.8 0.852 0.03 Average bite duration (duration/ bites), (min/n) 0.2 (0.1) 0.2 (0.1) 0.0 −0.0, 0.0 0.573 0.10 Bites per minute (bites/duration), (n/min) 5.2 (2.1) 4.8 (1.8) −0.4 −1.1, 0.3 0.216 −0.22
There was no evidence of an interaction between plate size and any of the measured effect modifying variables on consumption (response inhibition (
We found, in the most robust study to date of the effect of plate size on consumption, no clear evidence of a difference in the energy consumed between groups that self-served lunch using larger or smaller plates. The effect observed was in the hypothesised direction, but very small, with 19 kcal greater consumption from larger plates. Confidence intervals around the effect include the possibility of a small to medium effect in either direction. In addition, while more food was left on the larger plates, this concerned very small absolute differences, and there was no evidence of impact of plate size on the amount of food served or any other elements of the meal micro-structure. There was also no evidence of modification by individual characteristics, namely executive function, socio-economic position, and sensitivity to perceptual cues. While we would not predict an interaction with the first two of these, and the third was speculative in the absence of existing evidence, it would be unlikely for interactions to be detected in the absence of main effects [
A particular strength of the current study is that it adheres to recommended practices for open and reproducible science, including pre-registration of both the study protocol and statistical analysis plan, and an appropriate sample size calculation. Unlike most studies on this topic, it also complies with guidance for conducting randomised controlled trials, including ensuring adequate randomisation procedures, allocation concealment and blinding, as well as analysis being conducted by an external analyst unaware of group allocation. Finally, it also purposefully recruited a broadly representative, general population sample. These factors in combination mean that this study provides the most robust evidence to date to address the current uncertainty about the potential impact of altering plate size.
While conducting the study in laboratory conditions has advantages in controlling how participants are exposed to the intervention and reliably measuring their responses, giving high internal validity, there are inherent limitations conferred by this setting. In particular, external validity is compromised as the setting can never completely replicate a complex, real-world environment or eating occasion. To minimise these concerns, we used a naturalistic lounge environment within a purpose-built eating behaviour facility, which allowed an environment that would closely reflect real-world conditions. Participants were free to move around the room as they wished, were not rushed – as the time allowed was intentionally more than was needed for the meal to be consumed – and they could return to the meal as many times as they liked. A further limitation of the study was that while we were able to measure the total weight of the food consumed and observe eating behaviour throughout via video recordings, we were unable to assess all meal micro-structure characteristics over the duration of the session, such as the size of each individual serving where a participant served themselves on multiple occasions. This would have required monitoring food weight continually for which we did not have an unobtrusive method sufficiently inconspicuous to participants.
The finding of no or a very small effect on consumption suggests that previous meta-analyses of a low quality body of evidence may considerably overestimate effects [
Given the varying effect size estimates accompanied by wide confidence intervals seen in single studies and meta-analyses to date, uncertainty remains around potential effects of plate size on consumption. An additional complication is that a wide variety of plate sizes have been compared in previous studies. While it is possible that effects could differ for other absolute or relative size comparisons, to our knowledge there is no clear evidence that a given magnitude of increase or reduction in plate size would lead to a commensurate effect on consumption. For example, Rolls and colleagues [
The current evidence suggests that it is not feasible to conduct a single, laboratory-based study of this manipulation that is adequately powered, because extremely large sample sizes would be required. While prior effect size estimates – based on meta-analyses derived largely from poor-quality laboratory studies – suggest achievable sample sizes, these likely substantially overestimate effects. We are therefore reliant on cumulative meta-analyses of high-quality studies that minimise risk of bias, in order to reduce the uncertainty around these effects. Crucially, however, whether this has the potential to be an effect relevant to public health efforts to reduce consumption is ultimately predicated on whether effects seen in the laboratory will be observed in real-world field settings.
The effects of tableware size on food consumption in field settings are not currently known. A small number of studies in free-living conditions have been conducted, largely by the Cornell Food and Brand Lab [
Finally, we note that the use of smaller plates is widely promoted as a strategy for losing weight, including by reputable sources of healthcare advice (
The most robust study to date of plate size and consumption suggests that previous meta-analyses of a low-quality body of evidence may have considerably overestimated effects. However, the possibility of a clinically significant effect – in either direction – cannot be excluded. Well-conducted trials of tableware size in real-world field settings are now needed to determine whether changing the size of tableware has potential to contribute to efforts to reduce consumption at population-level.
Collaborative Award in Science from Wellcome Trust (Behaviour Change by Design: 206853/Z/17/Z) awarded to Theresa Marteau, Paul Fletcher, Gareth Hollands, Marcus Munafò.
The authors thank the staff of the Wellcome–MRC IMS Translational Research Facility, Cambridge, UK, for their help in conducting the study: Professor Sadaf Farooqi, Sarah Kelway, Kieran Ryan, Carol Keel and Lesley Wood. We also thank Dr. Hisham Ziauddeen for providing medical support to the study.
GJH, TMM and PF conceived the study. All authors collaborated in designing the procedures. GJH and DK coordinated the study and intervention procedures. DK, IL, EP, and NC collected study data. KD, RM and LK performed the data analyses. GJH and DK drafted the manuscript, with all authors providing critical revisions. All authors read and approved the final manuscript.
The datasets generated and analysed during the current study are available on the Open Science Framework project page
Approved by the Psychology Research Ethics Committee of the University of Cambridge (Reference Number: PRE.2018.036). All participants provided written informed consent.
Not applicable.
The authors declare that they have no competing interests.
Study set-up diagram. (DOCX 170 kb) Interactions and main effects of the plate size and
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.