Covariation and consistency of activation parameters

Individual differences in activation processes, as well as the consistency and predictability of these differences, constitute a classical issue eliciting much theoretical discussion in this field and as such poses an essential question for any practical application of psychophysiological methods. A typical activation experiment assessing 125 male students on four self-report and 21 physiological measures under five conditions (rest, mental arithmetic, interview, anticipation and blood taking) was performed supplying an empirical basis for a multivariate analysis. A partition of covariance, factor analyses, item analyses and scale construction procedures as well as models engaging an increasing number of components were used to study the covariation and consistency of these activation parameters. Several biometric problems that are generally thought to complicate the evaluation of such data (i.e. non-linear relations, the problem of initial values, differing sensibility curves of physiological response systems, individual response specificities) are considered and tested empirically. Findings suggest that the use of a single variable or a composite measure as an 'indicator' of individual differences in state or reaction aspects of activation is inadequate, due to empirical inconsistency and the lack of predictability between functional subsystems. A multicomponent model or a set of marker variables, having empirically derived discriminative efficiency as well as reliability estimates, seem to be preferable.