The subjects were 215 students in introductory psychology classes at Northern Illinois University. The students were asked to judge the guilt of a student accused of cheating on an examination. Individual subjects were given either strong or weak evidence and were told that either they or an authority would choose a punishment from a range of mild to moderate or moderate to severe penalties. Some students were led to believe that their judgments were real and binding; others were told that the study was an experimental similation. Contrary to the findings of Diamond and Seisel, the analysis showed that real jurors were more likely to convict and were more certain of the defendant's guilt. However, the severity of the penalty did not have a substantial positive effect on evaluations of offenders. Students in control of the penalty assigned more severe punishments than did students who recommended punishments to an authority. In addition, conviction votes were lowest when an authority controlled a severe punishment and students believed that their votes would have real consequences. Thus, differences exist between jurors who take part in both phases of a bifurcated trial and those who do not. 3 tables and 36 footnotes.