A statistical technique used in longitudinal research in which the relative influences of two variables upon one another are examined, while the correlations between each within time and within each across time are taken into account. More technically, it is a linear regression model where scores on at least two variables (X and Y) are measured on at least two occasions (t=1 and 2) and the scores at the later occasion (X[t+1]) are predicted by the same variable ‘lagged’ in time (X[t]) and the other variable ‘crossed’ in time (Y[t+1]).
See Cross-sectional design, Longitudinal design, Longitudinal studies