Blackwell Reference Online

Extract

With two variables, each measured at two points in time, one can compute six correlations: two synchronous correlations between the twovariables at a given time; two lagged correlations between the same variables over time; and two correlations between different variables at different time points, i.e., cross-lagged panel correlations. If the relationship between the two variables is a spurious one (i.e., their association is due to some underlying third variable that is a cause of them both), then the cross-lagged correlations should be equal to each other, assuming equal stabilities of the underlying two constructs. Thus, unequal cross-lagged correlations can be used to argue against spuriousness ( Kenny, 1975 ). Although one can argue against spuriousness with unequal cross-lagged correlations, they do not permit the estimation of casual effects. It is a misuse of such correlations to argue that their relative magnitude indicates the “causal predominance” of one variable over the other ( Rogosa, 1980 ). causality cannot be established by particular patterns of correlations. See also: methodology . ( 1975 ). Cross-lagged panel correlation: A test for spuriousness . Psychological Bulletin , 82 , 887 – 903 . ( 1980 ). A critique of cross-lagged correlation . Psychological Bulletin , 88 , 245 – 58 . ... log in or subscribe to read full text