Dissertations, Theses, and Capstone Projects

Date of Degree


Document Type


Degree Name





Steven Penrod

Committee Members

Saul Kassin

Mark Fondacaro

Miko Wilford

Marissa Bluestine

Subject Categories

Criminology and Criminal Justice | Law and Psychology | Law and Society | Other Psychology | Social Psychology


Plea bargaining, defense attorneys, criminal justice, predictive accuracy, law, shadow of the trial


Over the past few decades, the largely hidden, secretive, and widely used system of plea bargaining has caught the fervent attention of scholars. The Shadow of the Trial model has been central to much of the plea-bargaining literature, despite significant critiques about its oversimplification. The model posits that defendants and their attorneys make plea decisions based largely on the estimated probability of conviction and the severity of the sentence to which the defendant could be exposed at trial.

The model, however, assumes that all actors are rational, equally risk averse, have no competing interests, and possess high predictive accuracy. It also does not take into account the nuanced relationship between defendants and their attorneys. In two studies, we examine the foundations of the Shadow of the Trial model by providing practicing defense attorneys with case files containing evidence and asking them to predict jury outcomes and make plea recommendations. Juror verdicts from a mock trial containing the same evidence were used to measure attorneys’ accuracy. We measured attorneys’ perceptions of client guilt and asked them whether they believed their clients’ assertions of guilt or innocence. Attorneys were also asked about their perceptions of the fairness of plea bargaining. Finally, we examined how real-world constraints influenced attorneys’ judgments, perceptions, and recommendations.

Study 1 was a 2 (client assertion: guilty/innocent) x 3 (order of evidence viewed: increasingly exculpatory/increasingly inculpatory/randomized order) x 4 (case lean: incomplete file overall guilty/incomplete file overall innocent/complete file overall guilty/complete file overall innocent) between-subjects factorial design. Our model revealed that client assertion, perception of client guilt, and predictive inaccuracy significantly influenced plea recommendations.

Study 2 was a 2 (race: White/Black) x 3 (inadmissible evidence: present and discovered early/present and discovered late/not present) x 2 (client assertion: guilty/innocent) between-subjects factorial design. We found no race effects, but our conceptual model revealed a strikingly similar pattern to the one in Study 1: client assertion of guilt or innocence, perceptions of guilt, and predictive inaccuracy played key roles in plea recommendations.

Theoretical and practical implications of the findings are discussed.