Investigating Measurement Invariance Across Situations in Intensive Longitudinal Data

Measures of dynamic constructs in everyday life, captured in intensive longitudinal data (ILD), may function differently across clustering dimensions, leading to measurement noninvariance. Cross-classified factor analysis (CCFA) has been used to test invariance across people and time, but additional dimensions remain largely unexplored. Situational characteristics-both perceived and objective aspects-may affect responses, making them an informative clustering dimension beyond time for measurement invariance investigations in ILD. Using ILD from 225 participants with up to 50 observations each, we applied the CCFA approach to test the measurement invariance of vigor across people and time and, newly, to examine situational characteristics as an alternative clustering dimension. Profiles of perceived situations for clustering observations were derived via latent profile analysis, complemented by more objective situational aspects: location, accompanying people, and main activity. For the vigor items, the measurement parameters varied substantially across participants, indicating noninvariance. While invariance held across time and objective situational properties, measurement parameters varied markedly across perceived situations. These findings highlight the need to move beyond time as a clustering dimension and, crucially, consider people's situation perception when evaluating measurement invariance in ILD. The proposed CCFA usage can be copied to incorporate further clustering dimensions in future research.