Using Latent Class Analysis to Model Socioeconomic Position: Results from Three UK Birth Cohorts

Background: Socioeconomic position (SEP) is a multi-dimensional construct which changes over time as societal norms and behaviours change. Exploring how SEP differs between cohorts requires access to equivalent SEP indicators from the same socio-cultural context, which are often unavailable. However, consistent measures can be derived using retrospective harmonization, allowing researchers to conduct cross-cohort comparisons in relationships between equivalent variables. This study aimed to model SEP and then identify associated characteristics using harmonized data from three UK birth cohorts: the BCS70 (1970s), ALSPAC (1990s), and MCS (2000s).

Method: Latent class analysis (LCA) was conducted in each cohort using harmonized SEP data. The validity of the latent class models was evaluated with a subjective measure of financial stress. Finally, sociodemographic characteristics were tested as covariates of the latent classes to explore cross-cohort differences in the profile of the people within each class.

Results: Although different latent class models emerged in each cohort, the most disadvantaged class within in each cohort experienced the highest subjective financial difficulties. Additionally, certain characteristics were consistently associated with the most disadvantaged class across cohorts including single or cohabiting parents, parents that smoked, larger family sizes, and greater maternal mental distress.

Discussion: Despite using equivalent SEP indicators, the findings suggest that the nature and composition of SEP differs between cohorts. Nonetheless, the most disadvantaged in each cohort consistently experienced greater financial strain and were distinguished by specific sociodemographic characteristics; however, the risks associated with those characteristics fluctuate over time.