Twenty subjects worked first as individuals, then as members of a group, in assessing the probative strength of evidence in favoring either the guilt or innocence of "defendants" in contrived criminal court cases. Past research in group related phenomena, such as group polarization, has indicated that group judgments tend to be an enhancement of the average individual judgments. In this study, however, no evidence of group polarization was found. When the subjects considered evidence items one at a time, the group post-deliberation judgments were weaker than the average individual pre-deliberation judgments, when all of the judgments favored guilt. When the judgments all favored innocence and when the evidence was considered all together, no significant differences were found between the group and individual assessments. When pre-deliberation responses Were not homogeneous, the minorities appeared to have the ability to shift the majority toward neutrality. Whereas past research has indicated that a minority of one has no ability to shift the majority, in this study, when evidence items were considered separately, a minority of one, favoring innocence, was able to shift the majority favoring guilt toward neutrality. The method of aggregation of evidence items was examined using a Baysian and an Additive Model. Neither model usefully described the data. In both models, the aggregations made by individuals were widely dispersed, indicating that individuals used varying strategies to aggregate the evidence. The group aggregations, however, tended to be less variable. This indicated that the group served to make the aggregations more homogeneous, thus reducing individual differences.