Associations may be different at the between- and within-subjects level.
A random-effects approach can be used to estimate correlations at the two different levels.
Large correlations were found between positive and negative affect at both levels.
The discrepancy between the estimates was larger with more repeated observations available.