Abstract
The possibilities of drawing causal inferences from cross-lagged panel correlations are considered. Assuming that the effects of a causal state vanish with time, and assuming three-wave panel data, certain patterns of cross-lagged correlations are open to causal interpretations.