B4534 - Measurement Invariance in predictors of dropout from cohort studies - 12/02/2024
Individuals who take part in voluntary studies provide valuable data for health scientists across a range of health-related outcomes. These studies are extremely useful for understanding relationships between health related exposures and health outcomes. Results from studies such as ALSPAC (Children of the 90s) often rely on statistical methods to apply such relationships to the wider population (as the people who take part in these studies are often unusual in some way). One of the most commonly used statistical methods for this is called weighting, where we statistically "up-weight" people who are less likely to take part in a study, and "down-weight" people who are more likely to take part in a study, so that we can apply our findings to populations which are dissimilar to the studied (ALSPAC) population. It is not well understood how these methods perform in cases where the weights are constructed from a complex measure such as mental health - which may be measured differently for people who remain in a study, and those who go on to drop out. This study will provide valuable insights into how we might approach this.