Collecting objective biological measures (or biomeasures) in population-based social surveys, including Understanding Society (UKHLS), is critical to informing research and public health policy. However, biomeasures are not free of measurement errors emerging from the data collection process, especially when they are collected by different actors in mixed-mode surveys, such as nurses, interviewers, or self-collected by the respondents themselves.

This fellowship aims to better understand the measurement properties of biomeasure data collected by nurses, interviewers, and respondents, and test their measurement equivalence. To do this, multilevel modelling and Multi-Group Confirmatory Factor Analysis will be applied to the biomeasure data collected as part of a mode experiment conducted in Wave 12 of the UKHLS Innovation Panel.

The findings will inform research, policy, and survey practice in several ways. First, knowing which biomeasures are susceptible to nurse/interviewer effects may lead to improvements in data collection and training procedures in the main UKHLS study and other biosocial surveys. Second, the findings will inform data users about the potential consequences of ignoring nurse/interviewer effects in their statistical analyses as well as possible remedies. Third, testing the measurement equivalence across modes will contribute to academic discourse about “mode effects” and the comparability of health data collected in mixed-mode surveys. Lastly, the creation and archiving of a dataset that includes the corrected biomeasure scores on the latent variable will enable practitioners to analyze the scores after correcting for measurement error and mode effects, leading to more reliable inferences that feed into policy discussions.