<p>Based on the assumptions of the RI/EM we hypothesized as follows:
1. The REM part of the model predicts a) positive effects from earlier achievement (verbal and math) to later self-concept within a given domain (skill development effects) and b) positive effects from earlier academic self-concept to later achievement in matching domains (self-enhancement effects). In direct comparison, the skill development effects are presumably much stronger than the self-enhancement effects.
2. The I/E part predicts a) positive within-domain effects between verbal/math achievement and the corresponding self-concept (external frame of reference effects), and b) negative cross-domain effects between verbal/math achievement on contrasting subsequent academic self-concepts (internal frame of reference effects). The I/EM was originally formulated for the cross-sectional case, but in the RI/EM, the predictions are extended to the longitudinal case.</p>