Research Article: Inferring models of opinion dynamics from aggregated jury data

Date Published: July 1, 2019

Publisher: Public Library of Science

Author(s): Keith Burghardt, William Rand, Michelle Girvan, Emőke-Ágnes Horvát.


Jury deliberations provide a quintessential example of collective decision-making, but few studies have probed the available data to explore how juries reach verdicts. We examine how features of jury dynamics can be better understood from the joint distribution of final votes and deliberation time. To do this, we fit several different decision-making models to jury datasets from different places and times. In our best-fit model, jurors influence each other and have an increasing tendency to stick to their opinion of the defendant’s guilt or innocence. We also show that this model can explain spikes in mean deliberation times when juries are hung, sub-linear scaling between mean deliberation times and trial duration, and unexpected final vote and deliberation time distributions. Our findings suggest that both stubbornness and herding play an important role in collective decision-making, providing a nuanced insight into how juries reach verdicts, and more generally, how group decisions emerge.

Partial Text

What mechanisms underlie collective decision-making? Recent research on collective decision making has compared statistical patterns in empirical data to models [1–5] and tested how opinions change in controlled experimental settings [6–10]. Although both methods have provided substantial insight into the dynamics of collective decisions, the mechanism underlying how groups make decisions that do not end in unanimous agreement is underexplored. In addition, it is often difficult to determine through available data whether opinions form independently or if influence plays a role because either mechanism generally displays similar patterns in data [11, 12]. Despite this difficulty, we ask whether clues in data can hint at the role influence (or “herding” [13]) might play in group decision-making. We expect that opinions shift due to influence, but methods to test this intuition is lacking. Our recent work modeling voting behavior suggests that both herding and “increasing stubbornness”, in which individuals increasingly hold onto their opinion the longer they have it, help to explain data on vote distributions [5]. Do related models for other datasets reach similar conclusions? We explore these questions by comparing data of collective decision making in which decisions are made in the absence of complete consensus to a battery of plausible models with and without influence and/or stubbornness.

In the introduction, we laid out four motivations for the current research. First, we want to understand how groups make decisions that do not end in consensus. Second, we want to find ways empirical data can provide insight into the role influence plays in decision-making. Third, we aim to test the hypothesis that opinion formation is impacted by increasing stubbornness [5]. Finally, we want to determine whether we can use our modeling framework to better understand jury deliberation as a case study.