Locus of control as moderator: an explanation for additive versus interactive findings in the demand-discretion model of work stress?

Research based on Karasek's (1979) model of work stress has produced conflicting results; although some evidence of the demand X discretion interactions predicted by the model has been reported, most studies have failed to demonstrate interactive effects in relation to mental health outcomes. The present article investigates locus of control (LOC) as a potential moderator of demand/discretion effects. In a sample of civil servants (N = 590), regression analyses demonstrated a three-way LOC X demand X discretion interaction (p less than .01) for affective distress, but not for absence frequency. In longitudinal data from student teachers (N = 147), a similar three-way interaction was observed (p less than .05). In this case, the result applied specifically to anxiety as an outcome, and not to social dysfunction. In each study, the form of the three-way interaction was such that demand and discretion combined interactively to predict outcome for externals (-1SD LOC), in a manner consistent with Karasek's predictions, whereas for internals (+1SD LOC) additive findings were obtained. The implications of these results are discussed with reference to the demand-discretion model, and to the issues of stressor-outcome specificity raised by Broadbent (1985).