Previous experimental investigations have shown that expectations are not perfectly rational due to bias. Traditional adaptive models, however, in many cases do not perfectly describe the formation of expectations either. This paper makes two contributions to the experimental literature on the formation of expectations: First, we investigate whether subjects who have more information about the economic model than in previous studies also form biased expectations. Second, we argue that in some cases macroeconomic expectations might be formed by the anchoring-and-adjustment heuristic, which is well known in psychology. We find that subjects' expectations are biased although the design might be more favorable to rational expectations. The anchoring-and-adjustment model of expectations gets some support by our data, but the best model encompasses both the anchoring-and-adjustment model and the traditional adaptive model.