Bennett (1955) reported that a critical condition for producing Lewin's (1958) classic group discussion-decision effect (is., high compliance with a request from an authority figure) was a group consensus strongly favoring compliance. This finding prompted investigation of variables potentially affecting such a consensus. A first study found that group consensus is significantly less likely to favor compliance if subjects are allowed to make a majority rather than an individual decision. Group size had no significant effects on decision-making. A second study replicated this effect and investigated several explanations for it. These results suggest that, at least when compliance is not in the subjects' best interest, the Lewinian group discussion-decision effect will be less likely to be found if a majority decision role is followed as opposed to an individual decision rule. The implication of these data for social engineering is discussed.