Testing prospective effects in longitudinal research: Comparing seven competing cross-lagged models

Area of Science:

• Psychological research methods
• Longitudinal data analysis
• Psychometrics

Background:

• The cross-lagged panel model (CLPM) has been a standard for examining prospective associations between psychological constructs.
• Recent critiques have led to the development of alternative longitudinal models.
• Understanding the behavior of these competing models is crucial for accurate psychological research.

Purpose of the Study:

• To compare the performance of seven longitudinal models, including the traditional CLPM and the random intercepts CLPM (RI-CLPM).
• To evaluate model convergence, fit statistics, and parameter estimate consistency using a case study of self-esteem and depression.
• To provide practical recommendations for selecting appropriate models for analyzing prospective associations.

Main Methods:

• Analyzed ten independent samples with at least four waves of data (N = 326–8,259).
• Compared seven longitudinal models, focusing on the CLPM and RI-CLPM.
• Assessed models based on convergence rates, model fit indices, and the consistency of cross-lagged effect estimates.

Main Results:

• The CLPM and RI-CLPM demonstrated robust convergence across all samples.
• RI-CLPM generally showed superior model fit compared to the CLPM.
• CLPM yielded more consistent cross-lagged effect estimates, both within and across samples, than the RI-CLPM.

Conclusions:

• The choice between CLPM and RI-CLPM depends on the specific research question and the nature of the effects being investigated.
• CLPM is recommended for researchers focusing on between-person effects.
• RI-CLPM is suggested for analyses centered on within-person effects.