Applying a critical quantitative lens to examine the learning outcome differences between Indigenous and non-Indigenous children in Peru

Abstract

Across the world, Indigenous Peoples disproportionally suffer from persistent historical inequalities. In Peru, Indigenous children obtain lower scores on academic tests than their non-Indigenous peers. Despite efforts to tackle these inequalities and a growing consciousness that pejorative societal views towards Indigenous populations need to change, discrimination against these groups continues to be salient in Peruvian society. A deficit discourse that perpetuates views on Indigenous individuals as less able or ignorant is still present. Furthermore, structured efforts within academic research to tackle this narrative are lacking. Unfortunately, few quantitative researchers take steps to assess biases in data collection or analyses that can mask or exacerbate group differences. Based on these issues, this thesis has two aims. First, to examine the extent to which the choice of indigeneity definition and outcome measure influence Indigenous to non-Indigenous learning outcome differences. Second, to gauge the extent to which child, household and structural factors could explain differences between Indigenous and non-Indigenous children and whether this varies depending on how indigeneity is defined. The motivation of this thesis is to showcase that a deeper and more critical engagement with statistical data can provide new insights on the processes that may explain educational disadvantages of Indigenous children in Peru. Following equality of opportunity theory, this thesis is grounded on the notion that children should have the same chances to succeed in education regardless of their background. The first aim is addressed through four research questions (RQs). RQ1 is to what extent have previous studies that examine education outcome differences between Indigenous and non-Indigenous children in Peru critically reflected on their choice of indigeneity operationalisation and outcome measures? Indigeneity is a complex social construct and operationalising it through survey data is difficult. This warrants the question of whether different operationalisations of indigeneity matter when researching Indigenous groups’ education outcomes. Similarly, when researching the education outcomes of diverse groups comparatively it is important to assess the cross-cultural validity of the data. Inbuilt biases in data collection tools can systematically advantage one group over another and magnify or mask group differences. To provide an empirical example, this thesis examines survey data from Peru to answer RQ2, to what extent does the way that indigeneity is operationalised influence the magnitude of the observed between group differences? and RQ3, to what extent are available education outcome measures appropriate to study the learning of Indigenous children comparatively? When examining RQ3 focus is also placed on RQ5, are findings consistent across different indigeneity operationalization? To address the second aim, two research questions are examined. Namely, RQ4 to what extent are group differences attributable to personal attributes as compared to structural and educational opportunity factors? And RQ5. RQ4 is grounded on the Quantitative Critical Race Theory (QuantCrit) and Indigenous Statistics principle that indigeneity does not drive educational performance or pose an innate disadvantage. Rather it is assumed that observed educational differences between Indigenous and non-Indigenous children must be driven by other factors. These can be structural barriers or biases that are built into the research process. Findings from RQ3 inform the choice of outcome measure to examine this question. Finally, RQ5 extends the data exploration from RQ2. It seeks to empirically show whether the categories used to define Indigenous children influence research conclusions. This thesis’ methodology follows a critical quantitative framework based on QuantCrit and Indigenous Statistics principles. These emphasise that the categories used to collect and analyse statistical information shape research conclusions, and therefore researchers should be reflexive about their research choices. Hence, reflexivity is built into the integrative review and the empirical analyses of the Peru Young Lives survey. To address aim one, an integrative review of the literature, t-tests and Differential Item Functioning analyses are undertaken to examine different indigeneity definitions and the validity of outcome measures. Reflexivity is incorporated in the integrative review by examining whether researchers explained their rationale for and critically evaluated their analytic choices. Reflexivity is integrated in the empirical analyses by using a multidimensional operationalisation of indigeneity alongside binary ones. Using a multidimensional definition helps this research move away from a simplistic binary conceptualisation of indigeneity and enables capturing to some extent the complex social and evolving nature of indigeneity in Peru. Comparing results to those of binary definitions further enables showing how conclusions may vary depending on how groups are categorised. To address the second aim the Oaxaca-Blinder decomposition technique is used. The Oaxaca-Blinder decomposition enables breaking down the difference between groups into an explained portion attributed to differences in endowments and an unexplained portion that captures differences in the returns that groups obtain from a factor. Reflexivity is built into these analyses through the careful selection of the main outcome measure. The mathematics test is chosen because it shows no between-group bias. Furthermore, the variables included in the analysis are categorised as either personal attributes or structural and educational opportunity factors. The categorisation of some variables into these groups deviates from mainstream approaches. For example, while maternal education is often considered a personal attribute of the mother that is driven by personal choices, in this thesis it is considered a structural variable because it reflects educational access opportunities the mothers had. Finally, analyses are undertaken using a multidimensional and two binary indigeneity definitions, to show how different definitions can influence research conclusions. The integrative review reveals a clear trend. Authors of past studies rarely engage critically with the data they use, nor do they reflect on their research choices. Most reviewed studies use a single factor to define indigeneity, commonly the language of the mother, despite other information being available. In addition, most use the vocabulary test as the primary outcome. However, this thesis’ empirical analyses show that the choice of indigeneity definition does matter. Definitions based on ethnicity yield smaller between group differences than those based on the language of the child. Further, the cross-cultural validity analyses of the outcome data, identified biases in the vocabulary data, the outcome most used in the literature. However, bias is not found for the mathematics data, suggesting that it is more suitable for undertaking group comparisons. Lastly, questions flagged for bias vary with the indigeneity definition. Together, the analyses for aim one reveal a pattern of linguistic disadvantage affecting Indigenous children. Further, the Oaxaca-Blinder decomposition analyses show that the score difference in mathematics is entirely attributable to the explained portion. This means that differences are due to difference in endowments between groups. Only structural and educational opportunity factors contribute to the explained gap. Maternal education is consistently the most important factor, covering almost half of it. This is followed by grade repetition, rurality, region and wealth index. Importantly, child characteristics do not contribute to this portion. In terms of the unexplained portion, results are not statistically significant. This suggests that there is insufficient evidence that the score gap is due to differences in returns between groups. These results are consistent across indigeneity definitions. This thesis highlights the need for a more critical engagement with quantitative data to study the education outcomes of Indigenous children in Peru. It makes a unique contribution by using a critical quantitative framework. It has shown that the choice of indigeneity definition and outcome can influence the observed differences between Indigenous and non-Indigenous children. Finally, this thesis corroborates that score differences between Indigenous and non-Indigenous children are mostly attributable to structural and educational barriers. These findings emphasise that the observed outcome differences are not due to being Indigenous and highlight the importance of framing the identified differences as educational debts society owes Indigenous children. Therefore, addressing inequalities that affect Indigenous children early on has the potential to improve their long-term educational prospects. Supporting them throughout their educational journey, addressing linguistic disadvantages and promoting school completion have the potential to equalise outcomes.