Main content



Loading wiki pages...

Wiki Version:
Many studies have investigated the association between self-esteem and depressive symptoms in longitudinal research designs with the goal of understanding the nature of this association. A popular method applied to investigate longitudinal temporal associations is the Cross-Lagged Panel Model (CLPM). In the CLPM, temporal associations are investigated while controlling for stability of the constructs through the inclusion of autoregressive associations. The CLPM has been used in many studies examining the association between self-esteem and depressive symptoms. Overall, these CLPM studies found negative temporal associations from self-esteem to depressive symptoms more often than the other way around and with double the effect size. The results of CLPM studies are often interpreted in a way that suggests that found associations tell something about how self-esteem and depressive symptoms are associated over time within individuals, i.e. that it provides information about possible causal mechanisms. Unfortunately, such inferences are not warranted due to a major limitation of the CLPM approach. The CLPM does not differentiate between-person effects from within-person effects. Hamaker, Kuiper and Grasman (2015) showed that within-person effects may differ from between-person effects in presence, sign and strength. This means that results found in CLPM studies may not reflect associations found within-persons, and thus may not represent possible causal mechanisms. This raises the question, do associations between self-esteem and depressive symptoms found in a CLPM hold when between-person effects are separated from within-person effects? The Random Intercept Cross-lagged Panel Model (RI-CLPM) is an extension of the standard CLPM model which enables to separate between-person effects from within person effects through the inclusion of random intercepts. Although it seems unlikely that the sign of cross-lagged associations between self-esteem and depressive symptoms will change when separating between-person effects from within-person effects, it is very well conceivable that the presence and strength of associations differ. In two initial studies among adolescents we found mixed findings with regard to the similarity of results when either CLPM or RI-CLPM is used. In Study 1 we saw strong associations from depressive symptoms to self-esteem in the CLPM disappear in the RI-CLPM, leaving only small effects from self-esteem to depressive symptoms. Because Study 1 had the limitation that self-esteem was measured with only one item, it is possible that reliability issues of this measure contributed to the outcomes. In Study 2, with multi-item assessment of self-esteem and depressive symptoms, we found fairly similar associations in the CLPM and RI-CLPM, which were in line with most previous research findings. Considering our mixed findings, we aim to compare whether associations between self-esteem and depressive symptoms found in the CLPM differ from the RI-CLPM in another sample of pre-adolescents and adolescents. In sum, we aim to: Investigate whether associations between self-esteem and depressive symptoms found in a CLPM are also found within-persons when using a RI-CLPM.