Longitudinal attrition, the sample members’ decision to leave the sample over time, represents a strong source of bias in longitudinal studies because it does not happen at random. Indeed, attrition rates are particularly high for some subgroups, such as sample members with an ethnic minority or immigrant background. This study sought to further the understanding of how longitudinal attrition varies by ethnic or immigration backgrounds among sample members recruited in the first wave of Understanding Society.

First, this study examined whether factors influencing longitudinal attrition, such as tenure or level of education, have different effects across ethnic or immigration groups. Overall, we found evidence of variations in the sample, but they tended to have a very small effect size, leading to minimal variations in the probability of leaving the study sample.

Second, the study shifted the attention to the intersectional effects that emerge in the data, and the association between intersectional effects and ethnic or immigration background, using a MAIHDA approach. The analysis focused on six variables (age, economic activity, having dependent children, level of formal education, housing tenure, and area type) and revealed the presence of strong intersectional effects. These were found to be associated with ethnic and immigration backgrounds. One of the strongest intersectional effects could be found in the relationship between economic activity, age, and having dependent children. These findings have implications for survey design and weighting strategies.