118 undergraduate business students participated in an experiment which required that they manage production and workforce with the objective of minimizing costs using one of two forms of computer-based DSS. Using a 2/spl times/2 design, subjects were randomly assigned to one of four experimental conditions: the (1) presence or (2) absence of a modeling feature, crossed with a (3) "good" or (4) "bad" decision environment implemented as a decreasing or increasing sales trend. Results indicate that subjects with access to a modeling feature performed better than those without such capability; subjects' confidence, however, was not correlated with environmental effects or the availability of a modeling feature. We also found a strong correlation between confidence and actual decision quality for all conditions. All these results are in the opposite direction from those of a previously published study using the same task but a different DSS. Finally, we found no evidence of overconfidence, on average, in any of our conditions.