Measurement Invariance in Longitudinal Bifactor Models: Review and Application Based on the p Factor

Area of Science:

• Psychometrics
• Developmental Psychology
• Statistical Modeling

Background:

• Bifactor models are widely used for studying lifespan changes in psychopathology and cognition.
• Longitudinal measurement invariance (MI) testing is essential for valid interpretation of construct change over time.
• Current MI testing in bifactor models is infrequent and inconsistent, with limited simulation research.

Purpose of the Study:

• To review existing literature on MI testing in bifactor models.
• To provide recommendations for assessing MI in bifactor models based on simulation studies.
• To demonstrate the application of these recommendations using an empirical example.

Main Methods:

• Systematic review of MI simulation literature for bifactor and related models.
• Development of guidelines for assessing MI in bifactor models.
• Application of guidelines to an empirical dataset examining the general psychopathology factor (p).

Main Results:

• Only one prior study examined MI in bifactor models under restricted conditions.
• The study proposes recommendations for MI assessment, considering estimator choice and missing data.
• An empirical example demonstrated residual MI across gender and time for the general psychopathology factor (p).

Conclusions:

• Established MI guidelines enhance the confidence and comparability of findings from bifactor models.
• Further research is needed to refine MI guidelines, especially concerning model complexity and indicator count.
• The proposed guidelines facilitate more reliable interpretation of longitudinal changes in latent constructs studied with bifactor models.