AGGREGATION OR DISAGGREGATION - SOME COMM ENTS ON THE DEBATE AROUND THE RESULTS OF DIEKMANN AND PREISENDORFER

The results of Diekmann and Preisendorfer (1992) are discussed with re spect to the importance of the level of data aggregation for statement s about both the attitude-behaviour relation and behavioural consisten cy across situations. It is argued that the reported effects cannot be ascertained to be substantial because the data vary in their level of specificty and aggregation, because of the low reliability of one-ite m indicators, and because temporal stability is confounded with transs ituational consistency. Suggestions are made how these problems may be avoided. Finally, an alternative for the use of measures of variance- accounted-for as estimates of the practical importance of effects is p resented (BESD).