In survey research, panel analysis has the same function as the experiment in psychological research: the determination of causal relations. Here, the historical development in the techniques of panel analysis is sketched. The "cross-lagged panel-correlation technique" & the "sixteen-fold-table technique" are compared, & some extensions & corrections are given. Path analysis, one of the most recent developments in the causal analysis of (panel) data, is introduced. Path analysis has the advantage of giving extra possibilities in testing the causal model, eg, in the case of correlated error-terms. This leads to "conditional" estimations of the model parameters. The techniques are illustrated throughout, on the basis of the well-known panel data on vote-intention & attitude. 8 Tables, 3 Diagrams, 2 Figures. AA.
