Multilevel Analysis of Individual Heterogeneity and Discriminatory Accuracy (MAIHDA) is a recent multilevel regression modeling approach, rooted in intersectionality theory, for examining social inequalities across intersections of multiple social identities (e.g., gender, ethnicity, social class). The method was proposed in social epidemiology, but is increasingly applied in social science research. In this talk I will first give an introduction to MAIHDA including an empirical example drawn from published education research. I will then use this to motivate and present my own recent methodological research which explores the claim that MAIHDA’s predicted intersectional means are statistically superior to simple means from descriptive statistics and conventional regression models. Specifically, I will discuss the bias, variance, and mean squared error properties of these competing predictions. The findings show that MAIHDA-based means outperform simple means. However, the relative advantage of each MAIHDA predictor depends on the nature of intersectional inequalities and intersection sizes. MAIHDA’s benefits are most pronounced when inequalities are subtle or when data on certain intersections, such as those for marginalized groups, are sparse—conditions common in practice, highlighting the practical significance of our findings.
Booking is required for people outside of the Department of Social Policy and Intervention (DSPI).
DSPI Members do not need to register.