Statistical Methods for Investigating the Ethnic Achievement Gap in Colombia

Abstract

In industrialised countries like the US and the UK, the substantial body of literature about the problem of ethnic achievement gaps has informed a series of policies attempting to tackle it. Multiple methods have been applied, often borrowed from disciplines that have refined them to predict academic achievement or examine school effectiveness. In Colombia, only until 2018, the ethnic achievement gap became part of the policy agenda, but there is only one study describing the gaps, offering limited evidence to inform policy. This thesis views the ethnic achievement gap as a multilevel phenomenon of averages (reflecting the effect of student, school and local authority characteristics) and has both substantive and methodological aims. The substantive aim is to improve the current understanding of ethnic achievement gaps in Colombia (between White-mixed European descendants, the largest ethnic group- and Afrocolombian, Indigenous, and other ethnic minorities), including how they vary over time, within different subpopulations, and across different levels of the education system: students, schools and local authorities. The methodological aim is to study the potential of tailored measurement and statistical modelling techniques for improving the understanding of the nature of ethnic achievement gaps. The analysis applies single-level and multilevel modelling, as well as multivariate analysis methods to the results of a nation-wide compulsory examination for 11th grade (age 16/17) students between 2008 and 2013. Substantively, it is shown that the ethnic achievement gap of each minority group reflects different underlying processes at each level of the education system. Methodologically, it is argued that multilevel modelling is preferable to single-level modelling in some, but not all circumstances and each measurement method impacts the analysis of ethnic achievement gaps in different ways. The implications for policy and the consolidation of methods that contribute to the understanding of ethnic achievement gaps are discussed.

Date of Award	28 Nov 2019
Original language	English
Awarding Institution	University of Bristol
Sponsors	Economic and Social Research Council
Supervisor	George B Leckie (Supervisor) & William J Browne (Supervisor)