Author
Abstract
Longitudinal surveys are a cornerstone of social research, shaping key policy decisions. Yet, over time, certain groups are far more likely to drop out, raising concerns about representation and analysis bias. Attrition is especially high among respondents with ethnic minority and immigrant backgrounds – but why? Are they just more exposed to common risk factors, or do specific combinations of demographic and socio-economic characteristics contribute to increasing attrition rates in these groups? This paper tackles this question using data from Understanding Society, the UK’s largest longitudinal survey. First, the paper examines whether well-known predictors of attrition – such as type of housing tenure, form of economic activity, and level of education – operate differently across ethnic and immigration groups. While some differences emerge, in many instances, their effect size is small. Second, the paper employs MAIHDA (Multilevel Analysis of Individual Heterogeneity and Discriminatory Accuracy) to partition the attrition risk into additive (mean predictors’ effects) and multiplicative (effects of predictors’ interactions). The findings show strong multiplicative effects, primarily driven by the interaction among age, economic activity, and having dependent children. The analysis demonstrates that the distribution of these multiplicative effects varies strongly by ethnic and migration backgrounds, disproportionately increasing the likelihood of attrition of second-generation immigrants and sample members from Bangladeshi, mixed, and other Asian backgrounds. These findings carry two significant contributions for survey research. First, they demonstrate that the associations between ethnic minority and immigration backgrounds and attrition are largely explained by the effects of nonresponse predictors and by their complex interactions. Second, it proposes MAIHDA as a method for the study of survey nonresponse as it can offer important insights for statisticians and survey researchers seeking to understand the structure and the role of multiplicative effects on attrition in their studies.
Paper download