Date of Award

Winter 1980

Project Type


Program or Major


Degree Name

Doctor of Philosophy


Researchers from diverse disciplines--e.g. sociologists, criminologists, political scientists, legal observers, and most recently also psychologists--have studied judicial sentencing. More than half a century of research, employing multiple methodologies, strongly demonstrates that the unguided use (or abuse) of discretion has frequently led to vast disparity in judicial sentencing; i.e. large variations among sentences given for highly similar offenses and/or offenders. Among the various reform proposals reviewed, sentencing councils have been suggested as a constructive solution to reduce sentencing disparity without abandoning judicial discretion or displacing disparity to other agents in the criminal justice system.

The present research attempts to demonstrate the disparity reducing effect of discussion in sentencing councils in a controlled laboratory setting. From a social psychological perspective, sentencing a criminal offender can be conceptualized as judging an ambiguous stimulus object. Convergence toward the mean of the sentencing decisions (reduction of variability) was predicted as a function of group discussion in line with theories of norm formation processes (e.g. Moscovici, 1974; Sherif, 1935, 1936; Sherif & Sherif, 1969). Discussion of goals of judicial sentencing was expected to further reduce variability. Also of interest was whether or not writing down the sentencing decision prior to discussion would make judges less susceptible to group influence processes. In line with the polarization hypothesis (Moscovici & Zavalloni, 1969; Myers & Lamm, 1976; Lamm & Myers, 1978), penalty shifts (leniency or severity) were also predicted as a function of group discussion.

Extending the Solomon-four-group design, this study crossed three levels of the council factor (no council, council, and extended council) with two pretest conditions (no pretest and pretest). College students (N = 277) simulated mock judges who were presented a description of a case of armed robbery. As predicted, variability within councils was considerably less as a function of discussion in three-member councils and this effect was stronger for mock judges who had not written down their sentences than for those who had committed themselves that way. There were no differences in within council variability between the council and the extended council condition but somewhat less overall variability was observed in the extended council condition as compared to the no-council condition. There was no evidence for leniency and/or severity shifts.

The reduction of variability findings provide strong support for the postulation of norm formation processes as a function of group discussion. The observed commitment effect, and its absence in the no-pretest conditions, are also compatible with this interpretation. Contrasting traditional inductive approaches to validity with more recent deductive approaches, the applicability of these theoretical principles to real world sentencing councils is argued. The implementation of sentencing councils on a trial basis, in which council members do not write down their sentences before discussion, is recommended. Sentencing councils would be expected to reduce sentencing disparity to some extent but would have to be supplemented by other structural and procedural innovations that could also be investigated within the theory-testing interventionist approach promulgated here.