Amsterdam Public Health

Researchers in the health and behavioral sciences are often interested in studying change over time. In longitudinal cohort studies for instance, we aim to understand how a particular phenomenon changes across development. In randomized controlled trials, we aim to investigate the effect of a treatment or intervention on (change in) an outcome.
To study such change, it is common practice to measure the construct with the same scale at several time points, and then analyze the data with longitudinal models. But how do we know that the analyzed change is real, and does not represent an artifact because the scale no longer measures the same construct across time? If the latter occurs, the scale is said to lack longitudinal measurement invariance, which could lead to erroneous conclusions about change. Hence, it is important to examine longitudinal measurement invariance before conducting a longitudinal analysis, but this is not always done in practice.
In this tutorial, Marleen de Moor will introduce the basic principles of longitudinal measurement invariance and show how longitudinal measurement invariance can be tested using example datasets. The tutorial is intended for all researchers who are interested in learning about measurement invariance using factor analytic techniques in R-lavaan and Mplus; no prior knowledge about these techniques or software is needed.