https://orcid.org/0000-0002-9314-2774
![Sample our Social Sciences journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110822/TOC-Subject-Sample-2021-Social-Sciences.jpg"})
Abstract
It has long been argued that defendants detained pretrial face more severe case outcomes than released defendants. Considering the magnitude, directionality, and significance of these findings, this article uses systematic review and meta-analysis to examine the average effect of pretrial detention on a series of case processing outcomes: conviction, guilty plea, dismissal, charge reduction, incarceration, and sentence length. Assessing 143 effect sizes across 57 studies that met the inclusion criteria, findings indicated that detained defendants face more severe outcomes, with the strongest effect on their likelihood of incarceration. Pretrial detention had a medium effect on convictions, guilty pleas, and dismissals, a smaller effect on sentence length, and a non-significant, small effect on charge reductions. The studies’ effect sizes were heterogeneous, highlighting the importance of jurisdictional differences in policies and practices. Moderator analyses were used to assess this variation. Future research should examine how disparity cumulates via the pretrial detention penalty.
Acknowledgments
I would like to thank Dr. Natasha Frost, Dr. Gregory Zimmerman, and Dr. John Wooldredge for their helpful comments on earlier drafts of this paper. I also thank the meta-analysts who answered my methodological questions, especially Dr. Ojmarrh Mitchell, and the authors who provided the statistics needed to calculate the effect sizes.
Disclosure Statement
No potential conflict of interest was reported by the author(s).
Correction Statement
This article has been corrected with minor changes. These changes do not impact the academic content of the article.
Notes
1 Unpublished studies, particularly theses and dissertations, may still suffer from publication bias. However, this bias is typically of a lesser extent than that of peer reviewed articles as evident by related meta-analyses. For example, in his review of the impact of race on sentencing outcomes, Mitchell (Citation2005) found that effect sizes from published studies “exhibited substantively and statistically larger effect sizes” than those from unpublished studies. While including unpublished works begins to address publication bias, it does not resolve it entirely.
2 I used the following to search for studies across each of these databases. Search 1 of 2: (“pretrial detention” OR “pre-trial detention” OR “pre-sentence detention” OR “presentence detention”) AND (adjudicat* OR convict* OR plea OR “charge reduction” OR dismiss* OR sentenc* OR cumulative OR processing OR downstream OR “case outcomes”). Search 2 of 2: (“pretrial release” OR “pre-trial release” OR “pre-sentence release” OR “presentence release”) AND (adjudicat* OR convict* OR plea OR “charge reduction” OR dismiss* OR sentenc* OR cumulative OR processing OR downstream OR “case outcomes”).
3 Despite extensive efforts to identify studies that met the eligibility criteria, it is possible that some works were unintentionally excluded. Authors may have used phrases such as “held in custody pretrial” rather than “pretrial detention,” which would not have appeared in the search. The exclusion of these papers is unlikely given the entire papers were screened (e.g., literature review), and other steps were taken to identify eligible studies (e.g., eligible studies’ references, experts). If relevant studies were excluded from the search, it is possible that given the overlap of data sources, they used the same data as identified studies and would have been omitted later in the process.
4 While many studies examine the ‘in/out’ decision, where both prison and jail sentences are considered ‘in,’ other scholars have argued that prison and jail sentences are fundamentally different and should be analyzed separately (Holleran & Spohn, Citation2004). As such, I conduct sub-analyses for both sentence type and sentence length to examine only prison sentences and only jail sentences. However, the results of these analyses were largely similar to those of the combined prison/jail analyses (see Appendix D). A meta-analysis focusing on jail sentence length could not be conducted as only three effect sizes from six studies were available.
5 For example, Freiburger (Citation2011) conducts two logistic regression analyses predicting the ‘in/out’ decision. These analyses feature the same sample and include the same number of variables. The first includes three variables related to caretaker roles and the second replaces those with three variables related to financial roles. I averaged the effect size of pretrial detention on the incarceration decision across the two models.
6 However, if the studies analyze different years of the data, both were included (as long as there were no overlapping years). Moreover, if one study focused on only male defendants and a second study focused on only female defendants, both were included. If one study focused on drug defendants and another on male defendants, only one would be included as there would be overlap between the two samples (involving some statistical dependency). In short, multiple studies analyzing the same underlying data could be included if they focused on entirely different samples of this data.
7 One study that used a Poisson count model to examine sentence length in months was therefore excluded as the estimates could not be converted to Cohen’s d or odds ratios with the information available.
8 This involved converting the odds ratios and 95% confidence intervals to logged odds and standard errors, reversing the directionality of the coefficient, and converting the reversed logged odds and standard errors back to odds ratios and 95% confidence intervals.
9 References below 85% tended to have excluded abstracts and were therefore only matched by title and author. To err on the side of caution, 85% was selected so that only references with duplicate titles, authors, and abstracts, were flagged.
10 The number of identified works was rather large due to the many different dependent variables examined in the current review. A recent meta-analysis by Zane and Pupo (Citation2021) also examined multiple case processing stages when assessing the impact of a juvenile’s race on their case processing. Given the wealth of research on racial disparity, they collected 39,109 references using their search strategies and ultimately included 67 studies (0.17%). This is similar to the current review where 0.41% of the screened references were ultimately included in the meta-analysis.
11 Although type of analysis was significantly associated with charge reduction, the finding is not substantively significant given that 16 studies used regression analyses and only one instrumental variable analysis. Future works should further explore this variation given the conflicting directionality of the findings.
12 The studies utilize a variety of binary criminal history variables, such as prior arrest, prior conviction, prior prison, and on probation/parole. Moreover, some studies use 0 vs 1+ coding whereas others use 0-1 vs 2+ coding. Other studies specify the felony level, time frames, or certain types of offenses. Future meta-analysis with a larger sample size should investigate these differences further.