A Caltech Library Service

Institutional Design of Criminal Justice Processes


Huey, Joanna Nanami (2024) Institutional Design of Criminal Justice Processes. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/3gjk-mh19.


This dissertation contains three essays that contribute to ongoing debates about the design of institutions and procedures related to criminal justice.

Chapter 1 investigates how peremptory challenges in the jury selection process affect the diversity of and outcomes from juries. A game-theoretic model of attorneys’ decisions to strike potential jurors finds that the process 1) can lead selected jurors from a majority group to be a skewed sample and 2) can increase minority representation, contrary to common intuition. The first theoretical finding about the skew is supported by empirical analysis of data from jury selection transcripts: a novel measure of the pro-defense lean of jury pool members is developed, and selected White jurors are found to be more pro-defense than the average White pool member.

Chapter 2 develops a game-theoretic model of decisions about the verdict and sentence in a criminal trial, considering both single-actor and two-actor versions of this two-step process. Restrictions on sentencing discretion can lead to nullification where an actor with acquits who would have convicted under full discretion. When actors care about the lawfulness of their own actions, a two-actor process may lead to additional convictions, as the convicting actor can free ride off of a separate sentencing actor who will pay the cost of sentencing away from the lawful sentence. The model also leads to non-monotonic effects on the verdict when lawfulness or the expected sentence change.

Chapter 3 (joint work with Alexander V. Hirsch) uses mechanism design to examine single-threshold information escrows in a workplace setting. In this setting, reports of misconduct by a manager are kept secret until the number of reports exceeds a threshold and the manager is fired. When the firm designing the system wishes to minimize misconduct, a single-threshold mechanism leads to optimal results when misconduct reports are costless. In contrast, costly misconduct reports can make truthful reporting impossible under certain threshold values, raising the threshold above the firm’s ideal or even eliminating the possibility of any truthful mechanism. We find that single-threshold mechanisms are generally worse for the firm than mechanisms that mix two thresholds and can be worse than choosing whether to fire the manager without eliciting any information about misconduct.

Item Type:Thesis (Dissertation (Ph.D.))
Subject Keywords:criminal law; institutions; game theory
Degree Grantor:California Institute of Technology
Division:Humanities and Social Sciences
Major Option:Social Science
Awards:John O. Ledyard Prize for Graduate Research in Social Science, 2019.
Thesis Availability:Public (worldwide access)
Research Advisor(s):
  • Hirsch, Alexander V.
Thesis Committee:
  • Katz, Jonathan N. (chair)
  • Palfrey, Thomas R.
  • Rosenthal, Jean-Laurent
  • Hirsch, Alexander V.
Defense Date:13 May 2024
Record Number:CaltechTHESIS:05282024-055813959
Persistent URL:
Huey, Joanna Nanami0009-0000-5934-9184
Default Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:16429
Deposited By: Joanna Huey
Deposited On:04 Jun 2024 21:54
Last Modified:17 Jun 2024 19:37

Thesis Files

[img] PDF - Final Version
See Usage Policy.


Repository Staff Only: item control page