Access to this full-text is provided by Wiley.
Content available from Behavioral Sciences & the Law
This content is subject to copyright. Terms and conditions apply.
Behavioral Sciences & the Law
-
RESEARCH ARTICLE
OPEN ACCESS
Risk Perception and Risk Communication: Multi‐Actor
Perspectives on Pretrial Decision‐Making
Ashley E. Rodriguez
1
| Evan M. Lowder
2
| Peyton Frye
2
1
Department of Sociology and Criminology, Pennsylvania State University, University Park, Pennsylvania, USA |
2
Department of Criminology, Law and
Society, George Mason University, Fairfax, Virginia, USA
Correspondence: Ashley E. Rodriguez (aer5839@psu.edu)
Received: 11 January 2024 | Revised: 8 December 2024 | Accepted: 6 January 2025
Funding: This work was supported by George Mason University Ofce of Scholarship, Creative Activities, & Research.
Keywords: decision‐making | pretrial | risk assessment | risk communication
ABSTRACT
As jurisdictions across the United States implement pretrial risk assessments to advance pretrial reform, there has been a
limited research focus on factors affecting risk assessment‐guided decision‐making. To advance this work, this study examined:
(1) differences in perceptions of risk and utility of risk assessment information by criminal‐legal role; (2) whether static or
variable risk assessment presentation affected pretrial release decisions, including the moderating role of offense violence; and
(3) factors affecting risk assessment‐guided decision‐making more broadly. Vignettes were issued to 298 judges, pretrial ofcers,
prosecutors, and defense attorneys across the United States with random assignment to a one‐value probability (30%, 40%, or
50%) or a range of probabilities (30%–50%) risk estimate. Findings showed that risk assessment presentation did not affect
decision‐making, and decision‐makers either subjectively interpreted the risk assessment value or created their own risk
criteria. Results necessitate more training for pretrial decision‐makers on interpreting risk assessment information.
1
|
Introduction
Across the U.S., local criminal‐legal systems are implementing
pretrial risk assessments to reform the current pretrial system
(Lattimore et al. 2020; Pretrial Justice Institute 2019). A study of
pretrial practices across the U.S. revealed that two‐thirds of
counties use risk assessment tools as a part of their pretrial
process (Pretrial Justice Institute 2019). While judges are not
required to use risk assessment information in their decisions,
risk assessments provide structure to decision‐makers who must
make pretrial release decisions in a matter of minutes (Des-
marais, Monahan, et al. 2021; Desmarais and Lowder 2019).
Risk assessments accomplish this task by producing similar
recommended pretrial release conditions for defendants with
similar risk levels. This process creates more uniform pretrial
decisions across judges while also reducing judicial use of
heuristics in decision‐making (Desmarais, Monahan, et al. 2021;
Desmarais and Lowder 2019; Viljoen et al. 2021).
Risk assessments have good accuracy in predicting defendant
behavior (Desmarais, Zottola, et al. 2021; Jung et al. 2020; Vil-
joen et al. 2021), but their use in practice does not necessarily
reduce rates of failure to appear (FTA) or re‐arrest (Bechtel
et al. 2017; Lowder et al. 2021; Viljoen et al. 2018). As such, risk
assessment presentation may not accurately communicate the
uncertainty involved in risk estimates to decision‐makers. The
majority of prior research in this area has focused on other
decision‐makers, such as forensic clinicians, but there are
several ongoing studies assessing the impact of risk communi-
cation in courtrooms. Some researchers argue that categorical
presentations of risk assessments do not accurately convey risk
assessment information to decision‐makers because categories
This is an open access article under the terms of the Creative Commons Attribution‐NonCommercial‐NoDerivs License, which permits use and distribution in any medium, provided the original work
is properly cited, the use is non‐commercial and no modications or adaptations are made.
© 2025 The Author(s). Behavioral Sciences & the Law published by John Wiley & Sons Ltd.
Behavioral Sciences & the Law, 2025; 00:1–17 1 of 17
https://doi.org/10.1002/bsl.2717
obscure variation within categories, and practitioners and re-
searchers do not have a consensus on the denitions behind
each category (Scurich 2018). Overall, this fragmented literature
highlights the lack of research understanding factors that in-
uence decision‐making more broadly, including decision‐
makers’ perception of risk assessment information in decisions.
The goal of the current study is to assess how risk assessment
and risk communication affects the pretrial decision‐making of
judges, pretrial ofcers, prosecutors, and defense attorneys. We
issued online vignettes to judges, pretrial ofcers, prosecutors,
and defense attorneys located across the U.S. Our aims were
three‐fold and sought to: (1) examine the perceptions of risk and
utility of risk assessment information by pretrial role; (2)
examine the way in which presentation of risk assessment in-
formation affected pretrial decisions, including whether offense
type moderated the decision; and (3) exploratorily analyze the
quantitative factors (e.g., offense severity, value of risk assess-
ment tools, etc.) and qualitative rationales (e.g., case facts, risk
appraisal, etc.) that motivate decision‐making broadly, which
we expect would include the role of risk assessment informa-
tion. We randomly assigned pretrial decision‐makers to receive
either a risk assessment one‐value probability or a risk assess-
ment condence interval probability to quantitatively examine
their decision‐making. Additionally, we collected qualitative
data by asking pretrial decision‐makers to explain the rationale
behind their decision. This mixed‐methods analytic approach
provided much needed quantitative and contextual data on
pretrial decision‐making.
2
|
Literature Review
2.1
|
Perceptions of Risk Assessment Tools by
Pretrial Decision‐Makers
Pretrial decision‐makers have mixed views on the inclusion of risk
assessments in pretrial decision‐making. Prior work generally
suggests that judges and pretrial ofcers are more supportive of risk
assessments than prosecutors and defense attorneys (DeMichele
et al. 2019; Terranova et al. 2020). In their mixed‐methods study of
court decision‐makers’ attitudes towards risk assessments in a
Midwestern state, Terranova et al. (2020) found that judges and
pretrial ofcers perceived risk assessments as useful for objective
decision‐making. The authors additionally found that prosecutors
perceived risk assessments as effective for bond arguments. In a
different study, DeMichele et al. (2019) administered a survey to a
convenience sample of court decision‐makers who used the Public
Safety Assessment (PSA) as their risk assessment tool. In this study,
prosecutors viewed risk assessments as inadequate for securing
pretrial detention for high‐risk defendants and believed that risk
assessments overall detained too few defendants. There are also
mixed ndings on the utility of risk assessments among defense
attorneys. DeMichele et al. (2019) found that most defense attor-
neys rely on risk assessment information to support client release.
Conversely, defense attorneys showed the least support for risk
assessments in a study by Terranova et al. (2020).
Despite pretrial decision‐makers holding a generally supportive
view of risk assessments, pretrial actors have expressed concerns
about such tools. For one, many pretrial decision‐makers oppose
the inclusion of risk assessments in their decision‐making
because their decision‐making (i.e., decisions themselves, speed
of their decisions, detention rates) can be monitored (Brayne and
Christin 2021). Another complaint regarding risk assessments is
the face validity of particular risk assessments items for the pre-
diction of FTA or committing misconduct (Terranova et al. 2020).
Court decision‐makers report skepticism about the authenticity
of self‐reported items, particularly interview data (Terranova
et al. 2020). Finally, some pretrial decision‐makers believe that
personal knowledge and prior experience will yield more accu-
rate decisions than a risk assessment's recommendation (Brayne
and Christin 2021). While prior research suggests most decision‐
makers may accept risk assessment tools, less research has
examined court decision‐makers’ attitudes towards risk assess-
ment tools and how their attitudes translate to pretrial decisions.
2.2
|
Risk Communication
Clear communication of risk assessment information can help
criminal‐legal decision‐makers improve their understanding
and utility of risk assessment information. Risk assessment in-
formation can be presented as risk levels or categories (low,
moderate, high), total scores (i.e., score of “2”), frequencies (i.e.,
20 out of 100), probabilities (i.e., 20%), risk levels with predicted
probabilities (i.e., low risk =5%), risk factors (e.g., history,
gender, substance abuse use), or combinations of these risk
presentation formats (Heilbrun, Dvoskin, et al. 1999; Hilton
et al. 2015; Kwartner et al. 2006). Despite the numerous ways in
which risk can be communicated, there are few comprehensive
studies that demonstrate which method is the most effective in
communicating risk to judicial decision‐makers.
2.2.1
|
Forensic Clinicians
Prior research on risk communication mainly focuses on
forensic clinicians, but these ndings provide a foundation for
judicial risk communication research. One approach to assess
the most effective risk presentation method is to survey practi-
tioners on their preferences. In two studies of risk communi-
cation preferences, psychologists and psychiatrists favored risk
factor presentations over numerical likelihoods (Heilbrun
et al. 2000; Heilbrun, Philipson, et al. 1999). However, practi-
tioners' preferences for particular risk communication formats
may not translate into improved risk communication. More
specically, Scurich (2018) argued that categorical risk assess-
ments may not accurately convey or improve risk assessment
accuracy. Thus, practitioners should not use categorical risk
assessments. On the other hand, practitioners may encounter
difculty interpreting numerical risk assessment information
because they interpret this information in a comparative
manner (Monahan and Steadman 1996). Therefore, directly
examining how risk communication affects release decisions
may be more informative in determining the relative leniency
and restrictiveness of risk communication formats.
Few studies compare the relative leniency and restrictiveness of
various numerical risk communication formats. In one such
study, psychologists were presented with a vignette of a hospi-
talized mental health patient that was eligible for release but had
2 of 17 Behavioral Sciences & the Law, 2025
a risk of committing a violent action, and this risk was manipu-
lated to display either a frequency or probability value (Slovic
et al. 2000). The authors found that psychologists perceived the
patient to be a lower risk as opposed to a medium risk when
presented with a probability risk assessment value over a fre-
quency value. Additionally, respondents were more willing to
release the patient when presented with a probability value
compared to a frequency value. Monahan et al. (2002) replicated
this study with an additional manipulation of providing either a
vivid (gory detail of violent event) or pallid (non‐descriptive and
impartial reporting of violent event) description of a violent event
that occurred between a patient and a stranger and discovered a
more nuanced decision‐making situation. Specically, the prob-
ability value format with a pallid description and a respondent
who worked in a forensic facility resulted in more lenient release
decisions. Together, these few ndings highlight the importance
of risk communication in inuencing decision‐making and un-
derscore the need for more comprehensive research on how risk
communication formats affect decision‐making.
2.2.2
|
Judges
Relative to research on forensic clinicians, fewer studies have
examined risk communication in a judicial context, with no
published research in the pretrial setting. In two surveys of pref-
erences for risk communication, most judges preferred to have
risk assessment values presented categorically over numerically
(Evans and Salekin 2014; Kwartner et al. 2006), although judges
also reported that having both numerical and categorical infor-
mation would be useful (Kwartner et al. 2006). Unfortunately, we
lack research on how these risk communication preferences
translate to courtroom decision‐making.
Only three studies have examined the effectiveness of numerical
risk communication formats in a judicial context. In a survey of
judges, Monahan and Silver (2003) asked judges to decide the
minimum likelihood of violence threshold for short‐term civil
commitment, nding no signicant differences between the
judges' decisions when presented with frequency versus per-
centage formats of risk. In a different survey of judges, Rachlinski
et al. (2012) showed that judges were more likely to decide that
DNA evidence came from the defendant and issue a guilty verdict
when presented with a probability score than a frequency score.
Despite this nding, the communication format did not affect the
defendant's conviction, but the authors noted that this could be
due to a lack of statistical power (Rachlinski et al. 2012). Finally,
Evans and Salekin (2014) asked judges presiding over involuntary
commitment proceedings to decide how to release a patient who
met every criteria except for the “danger to others” criterion
necessary for civil commitment. Similar to previous ndings,
frequency and probability risk communication formats did not
produce different release decisions.
Thus far, researchers in the judicial and forensic clinician space
have focused on probability and frequency numerical presenta-
tion formats. However, this body of work overlooks potentially
important differences in static (i.e., 20%) and variable (i.e., 10%–
30%) probability risk communication formats. It may be more
benecial for decision‐makers to present risk in a variable format
over a static format because the variable format provides more
information concerning the uncertainty inherent in predicting
risk. Conversely, decision‐makers with low numerical literacy
may not understand how to use variable probability information
and decide to use a probability value towards the middle of the
probability range to aid their decision‐making. Unfortunately, no
study to date has examined variation in decisions based on
different probability risk communication formats. Based on the
lack of research on risk communication format effectiveness and
pretrial decision‐making, this paper attempts to assess the impact
of risk communication in the pretrial decision‐making setting.
2.3
|
Utility of Risk Assessments in Decision‐
Making
Although some jurisdictions provide decision‐makers with risk
assessment information to inform decision‐making, the extent
to which decision‐makers utilize risk assessment information in
the decision‐making process remains unclear. However, previ-
ous studies attempt to proxy the use of risk assessments through
different strategies. These strategies include self‐reports, statis-
tical tests to determine differences in decision‐making pre‐ and
post‐risk assessment implementation, and general adherence to
risk assessment information.
Little research exists on how court decision‐makers self‐report
their use of risk assessment information in their decisions, with
only two studies identied in the literature. DeMichele
et al. (2019) discovered that among judges, most (98%) at least
sometimes used risk assessments in their decisions, but only half
reported risk assessments being useful for their decisions. Half of
prosecutors reported using pretrial risk assessment information
when they wanted to secure a pretrial detention decision. How-
ever, when prosecutors wanted the defendant to be released, they
rarely or never used pretrial risk assessment information to sup-
port their argument. Defense attorneys held the opposite position,
with most defense attorneys reporting that risk assessment in-
formation was useful to secure client release. Finally, Brayne and
Christin (2021) conducted ethnographic eldwork in a southern
state. Their observations of daily activity and interviews indicated
that many decision‐makers expressed resistance towards the
utilization of risk assessments to inform their decision‐making.
Types of resistance included professionals not wanting to use risk
assessments in their jurisdictions, delays in providing risk
assessment information to judges, and pretrial ofcers and pros-
ecutors refusing to share data with the county's data analytics
team.
Although self‐report data provides insight into pretrial decision‐
makers’ attitudes toward and use of risk assessment information,
these studies lack empirical information on whether risk assess-
ments inuence decision‐making. In Kentucky, two pretrial risk
assessment implementations were associated with higher rates of
pretrial releases, but release rates returned to baseline levels after
a short period of time (Stevenson 2018). This nding suggests that
pretrial risk assessment implementation may have immediate
effects on decision‐making that fades over time. A different study
within a large southeastern U.S. jurisdiction found that risk
assessment implementation was associated with higher pretrial
release rates for felony cases and lower rates for misdemeanor
cases for all types of release, including money bail, supervised
3 of 17
own recognizance (SOR), and own recognizance (OR; Copp
et al. 2022). The authors speculated that this change in the pretrial
release decisions could be due to risk assessment consideration or
perhaps judicial discretion. Additionally, researchers found that
the rate of pretrial releases via OR and SOR increased slightly after
risk assessment implementation, but these releases were often
accompanied with a nancial condition (Copp et al. 2022). Thus,
pretrial risk assessments were not necessarily associated with a
decrease in the use of nancial conditions in favor of uncondi-
tional release. This evidence contradicts DeMichele et al.'s (2019)
self‐report ndings, highlighting the difculty in empirically
demonstrating decision‐makers use risk assessments and that
their decisions result in less restrictive pretrial release conditions,
which has been found in some prior research (Lowder et al. 2021).
In a similar vein, research must not only examine whether
decision‐makers self‐report adherence to risk assessment rec-
ommendations but also whether their decisions actually align
with risk assessment recommendations. A study by Copp
et al. (2022) revealed that less than half of judges followed the
risk assessment recommended pretrial outcome. Instead, judges
tended to depart upwards, typically from a release on own
recognizance (ROR) outcome to supervised pretrial release or
nancial bond (Copp et al. 2022). Research by Copp et al. (2022)
highlights the importance of examining the attitudes and actual
decisions for discrepancies, but future research should continue
to investigate whether decision‐makers deviate from guidelines
and the context in which they do so. Our study contributes to
this small but growing literature through examining how
decision‐makers use risk assessment information.
2.4
|
Study Purpose
Overall, the sparse and incohesive ndings on risk perception
and risk communication in the pretrial context necessitate the
need for research on the perception and effectiveness of a va-
riety of risk communication formats (Hilton et al. 2015) and the
drivers of pretrial decision‐making more broadly. To address
this gap, we surveyed criminal‐legal decision‐makers from
across the U.S. and examined their pretrial release decisions and
rationales in response to a pretrial case vignette. These vignettes
varied in terms of risk assessment presentation and offense. Our
aims were three‐fold: (1) to examine the perceptions of risk and
utility of risk assessment information by pretrial role; (2) to
examine how presentation of risk assessment information
affected pretrial decisions, including whether offense type
moderated the decision; and (3) to exploratorily examine the
quantitative and self‐reported factors motivating decision‐
making, including the role of risk assessment information.
3
|
Method
3.1
|
Participants
We recruited a broad range of pretrial decision‐makers, in part
due to the difculty in recruiting a judge‐only sample (Nir 2018),
but also due to the important role of other decision‐makers in the
context of pretrial decision‐making. In all U.S jurisdictions, pre-
trial ofcers prepare and submit recommendations for release to
the judge and in some counties have deferred release authority,
which allows them to release a defendant prior to an initial
hearing (e.g., Riley 2024; West 2017). Both prosecutors and de-
fense attorneys are active participants in the initial hearing, where
both sides make arguments and negotiate a defendant's pretrial
release terms.
We recruited 298 criminal‐legal decision‐makers, and our
sample primarily consisted of defense attorneys (33.8%) and
pretrial ofcers (31.7%), followed by judges (13.2%), prosecutors
(12.8%), and other decision‐makers (8.5%). Descriptive statistics
are presented in Table 1. Decision‐makers had an average of
16.3 (SD =11.4) years of experience in the criminal justice
system and 11.2 (SD =11.1) years of experience handling pre-
trial cases. Decision‐makers had an average weekly exposure to
risk assessment tools (M=4.04, SD =1.27). Most participants
reported that risk assessment tools are somewhat helpful
(M=3.88, SD =1.09). Decision‐makers also believed that the
goals of the criminal‐legal system were equally rehabilitative
and punitive (M=3.45, SD =1.26). The decision‐makers were
primarily White (83.9%), male (51.5%), and 44.9 (SD =12.93)
years old. Most decision‐makers issued ROR (69.5%) as opposed
to monetary bond. On average, there was a 55.6% (SD =36.92)
probability of releasing the defendant with additional condi-
tions. These decision‐makers practiced in 30 U.S. states with a
breakdown of participants by state presented in Table A1.
3.2
|
Materials
To address Aim 2, we tested two independent variables, which
were experimentally manipulated. Risk assessment probability
was dened as a static 30%, 40%, or 50% value, or as a variable
30%–50% value. Offense violence was dened as either receiving
a violent or nonviolent offense scenario. In Aim 3, we tested 10
independent variables, in multivariable models. These included
offense violence (i.e., nonviolent or violent offense). Pretrial
decision‐maker role was dened as having a criminal‐legal
profession as either a pretrial ofcer, judge, prosecutor, de-
fense attorney, or other role. We measured professional expe-
rience in the criminal‐legal system and professional experience
in the pretrial system on a continuous scale of years. Exposure
to risk assessment tools was measured on a 5‐point Likert‐type
scale from never (1) to daily (5). The value of risk assessment
tools was measured as the decision‐maker's agreement with the
idea that pretrial risk assessment tools are useful in pretrial
decision‐making on a 5‐point Likert‐type scale from strongly
disagree (1) to strongly agree (5). Perceived goal of the criminal‐
legal system was measured on a 7‐point Likert‐type scale from
completely rehabilitative (1) to completely punitive (7). Race
was dened as White, Black, or Hispanic, due to low numbers of
individuals endorsing Hispanic ethnicity separately from race.
Gender was dened as male or female. Finally, we measured
age on a continuous scale of years.
To address Aims 2 and 3, we measured two dependent vari-
ables. Release decision was dened as whether the pretrial
decision‐maker decided to release the defendant on monetary
bond or ROR. The likelihood of release with conditions was
dened as the decision‐makers’ perceived likelihood that they
would release the defendant with conditions (e.g., pretrial
4 of 17 Behavioral Sciences & the Law, 2025
supervision, electronic monitoring, drug testing, etc.) on a scale
of 0%–100%.
3.3
|
Procedure
3.3.1
|
Survey Structure
We developed a vignette‐based survey containing a brief
decision‐making vignette, manipulation questions, closed‐ and
open‐ended questions regarding the pretrial release decision,
and professional background and demographic questions. The
vignette opened with a statement that the participant needed to
take on the role of a judge at an initial hearing. The participants
were instructed to make a pretrial release decision based on the
defendant's likelihood of (1) appearing to their next hearing and
(2) abstaining from new criminal activity. The vignette next
simulated information presented to a judge at the initial hearing
by describing defendant characteristics (e.g., race, gender, age),
case characteristics (e.g., charge, crime description, criminal
history), and risk assessment value. The decision‐makers also
received a risk estimate which included a statement that the risk
estimate predicts the defendant's chance of failing to appear or
being arrested for a new offense during the pretrial period.
We included pretrial release arguments made by the prosecutor
and defense attorney to simulate arguments that a prosecutor
and defense attorney would typically make in a pretrial hearing.
The prosecutor argued that based on the defendant's criminal
history, the current charge, and the threat the defendant posed
to the community, the decision‐maker should release the
defendant on a high monetary bond amount. To counter the
prosecutor's argument, the defense attorney highlighted how
the defendant had ties to the community (i.e., employment,
family, home) and the importance of maintaining those ties, and
requested an ROR decision.
The vignette also contained an adapted Pretrial Services Report
used by Indiana jurisdictions (Indiana EBDM Pretrial Work
TABLE 1 |Descriptive statistics.
Measure N%M SD
Outcome variables
Monetary bond (ROR) 91 30.54
Likelihood of release with conditions 55.62 36.92
Predictor variables
Risk assessment presentation
30% 75 25.17
40% 72 24.16
50% 78 26.17
30%–50% 73 24.50
Offense violence (nonviolent) 151 50.67
Role
Pretrial ofcer 89 31.67
Prosecutor 36 12.81
Judge 37 13.17
Defense attorney 95 33.81
Other 24 8.54
Professional experience in the criminal‐ legal system 16.30 11.36
Professional experience in the pretrial system 11.23 11.13
Exposure to risk assessment tools 4.04 1.27
Value of risk assessment tools 3.88 1.09
Goals of the criminal‐legal system 3.45 1.26
Race
White 234 83.87
Black 17 6.09
Other 28 10.04
Gender (male) 134 48.55
Age 44.91 12.93
Note: N =298.
5 of 17
Group 2018). The Pretrial Services Report does not explicitly
provide information as to how the risk assessment tool pro-
duced a risk assessment value. Thus, we did not indicate how
the defendant scored on each of the seven items in our simu-
lated report. However, throughout the vignette we included
information that the risk assessment tool would use to produce
a risk estimate (number of prior failure to appears, prior con-
victions, employment, and residential stability) to provide more
context about the defendant to the decision‐makers. We did not
provide substance use information, although it is a factor in our
risk assessment tool, because in a typical proceeding where a
defendant does not have a history of substance use, none of the
court actors would mention the lack of substance use.
After reading the vignette, participants answered three manipu-
lation questions. These questions veried whether participants
read the vignette, their social desirability, and their perceived
likelihood of the defendant committing misconduct if released
during the pretrial phase. To address Aim 1, we asked partici-
pants to state their exposure to risk assessment tools during their
criminal‐legal career. We also asked them to rate their agreement
with the statement: Pretrial risk assessment tools are helpful for
pretrial decision‐making. We asked participants to identify
whether the goal of the criminal‐legal system is more punitive or
rehabilitative. Finally, we asked participants to state their role in
the criminal‐legal system. For Aim 2, we asked participants to
decide how they would release the defendant and how likely they
were to release the defendant with additional conditions. To
address Aim 3, we used information from the previous questions
and also collected additional information such as the partici-
pant's professional experience in the criminal‐legal system and
the pretrial system. We then requested their race, gender, and
age. We additionally asked participants to describe their decision‐
making rationale and to list unprovided information that would
have assisted with their decision‐making.
3.3.2
|
Vignette Manipulation
Vignettes varied based on risk estimate and offense scenario.
Respondents were randomly assigned to one of four estimates:
30%, 40%, 50%, or 30%–50% value. This manipulation only
changed the risk estimate values across the eight vignettes. We
designed the risk estimate of 30%–50% to have a 20% range in
order to detect variation in decision‐making that may not be
discernible in a 5% range. Further, this range allowed us to
examine whether decision‐makers made decisions that aligned
with the lower, middle, or upper section of the interval when
compared to the decisions of the participants that received a static
risk estimate.
Risk estimates are one piece of information that pretrial
decision‐makers use to make their decisions. Decision‐makers
tend to also use legally relevant factors such as the charged
offense and criminal history. To ensure ndings would gener-
alize across offense types and to additionally explore the role of
offense type as a moderator, we randomly assigned respondents
to one of two offense violence scenarios: violent or non‐violent.
Both scenarios began with the defendant and victim sitting on a
bus bench, waiting for their bus to arrive. For the non‐violent
theft scenario, the victim forgot to grab their workbag when
they boarded the bus. The defendant waited for the bus doors to
close and then ed the scene with the victim's workbag. The
victim estimated that the workbag contained over $3000 worth
of items. In the violent battery scenario, the bus arrived and the
impatient defendant shoved the victim while waiting in line to
board the bus. The defendant immediately ed the bus stop. The
shove caused the victim to hit their head on the bus step,
resulting in ve stitches and a mild concussion at the emergency
room. Although both scenarios had different case facts with
different levels of violence, across multiple states, both scenarios
warranted felony level 6 charges. This is due to the value of the
stolen items (within the range of 750–50,000 dollars) for
the theft charge and the “moderate bodily injury” sustained by
the victim for the battery charge.
3.3.3
|
Survey Review
We collaborated with various stakeholders to improve the
external validity of our vignette. To ensure that the vignette and its
materials were as authentic as possible, we consulted with a local
director of pretrial services who has overseen risk assessment
implementation in his Indiana jurisdiction. We next had the
entire survey reviewed by two academic researchers who are
specialized in risk assessment outcomes, and one judge who has
expertise in risk assessment tool usage in Indiana. Modications
included creating four risk assessment presentations, and
removing pretrial detention as a release option because it is often
not an option for a low‐level felony arrest. We additionally
received assistance in creating clearer offense descriptions that
would match felony level 6 offenses while varying the level of
violence. The George Mason University Institutional Review
Board reviewed and approved this study protocol (IRB #1865910).
3.3.4
|
Data Collection
Our target sample size was guided by an a priori power analysis.
We conducted the power analysis using a 4 �2 contingency table
and χ
2
statistic to ensure sufcient power for a global effect of
study conditions on our primary outcome of pretrial release. We
needed 121 participants to detect a medium effect (power level of
0.80, phi coefcient effect size of 0.30, and alpha of 0.05), and 273
participants for a small‐medium effect (power level of 0.80, phi
coefcient effect size of 0.2, and alpha of 0.05; Cohen 1992).
We used snowball sampling to recruit judges, pretrial ofcers,
prosecutors, and defense attorneys from across the U.S. We con-
tacted various government organizations (e.g., County and State
Pretrial Departments, etc.) and professional associations (e.g.,
American Judges Association, etc.) to forward an email contain-
ing the online vignette link to eligible participants. We began our
recruitment by emailing all national government and profes-
sional organizations that had members who overlapped with our
target population. We then emailed individual state government
and professional organizations. Finally, we employed a cluster
random sampling strategy by using a random county generator to
6 of 17 Behavioral Sciences & the Law, 2025
randomly select ve counties from each state to contact at the
government and professional association level. In total, we sent
342 emails to various government and professional organizations
across the U.S. All data was collected anonymously via Qualtrics.
3.4
|
Analytic Strategy
A small proportion of responses contained missing data (n=21,
7.0%). The variables that contained the most missing data were role
in the criminal‐legal system (n=21), age (n=21), race (n=19), and
gender (n=19), reecting attrition over the course of the survey.
We conducted chi‐squared tests and t‐tests on all the independent
and dependent variables to detect patterns of missingness with
excluded responses. Due to no signicant patterns of missingness,
we did not omit incomplete cases from our main analyses.
For our analytic strategy we utilized a convergent parallel mixed
methods approach. We rst employed quantitative methods to
have a base‐line quantitative understanding of how perceptions of
risk assessment tools and the criminal‐legal system vary across
criminal‐legal professions (Aim 1), how risk communication, in
combination with offense violence, impact pretrial release
decision‐making (Aim 2), and the factors that generally inuence
decision‐making (Aim 3). This type of analysis creates general-
izability of our ndings to other criminal‐legal actors within the
U.S. (Dawadi et al. 2021). We concurrently addressed Aim 3 by
using content analysis to have a more nuanced understanding of
what criminal‐legal professionals believe were important
decision‐making factors. As such, we used quantitative and
qualitative research methods to triangulate our ndings (Dawadi
et al. 2021). This analytic strategy helped us to not only validate
ndings from the other strategies, it also helped to explain null
ndings, which in whole created a complete picture of the usage
of risk assessment tools in pretrial decision‐making (Creswell and
Clark 2018; Dawadi et al. 2021). For Aim 1, we conducted one‐way
ANOVA tests with post‐hoc Bonferroni tests. For Aim 2, we
conducted hierarchical linear and logistic regression models. In
the rst step, we tested the main effect of risk presentation. In the
second step, we added offense violence as a main effect. In the
nal step, we modeled the interaction effect between risk pre-
sentation and offense violence. To address Aim 3, we created
linear and logistic regression models using offense violence,
professional characteristics, and demographic characteristics to
model how these characteristics inuence pretrial decision‐
making and the likelihood of release with conditions.
To further explore the factors that motivated decision‐making, two
members of the research team (Ashley E. Rodriguez and Peyton
Frye) conducted content analysis of two open‐response survey
questions using an inductive coding approach. Coders took an
initial sample of 15 responses to develop main domains and sub‐
domains of items. Coders then coded all responses for presence
or absence of these main domains and sub‐domains. New items
were discussed and agreed upon by both coders before recoding
responses for these new items. Coders had an overall inter‐rater
agreement score of 96.9%, indicating strong agreement.
4
|
Findings
4.1
|
Aim 1: Perceptions of Risk Assessments and
the Criminal‐Legal System
Table 2contains the one‐way ANOVA results for our three
measures of how court decision‐makers perceive pretrial risk
assessment tools. First, pretrial decision‐makers signicantly
varied in their exposure to pretrial risk assessment tools
(p<0.001, η
2
=0.14). Bonferroni post‐hoc tests revealed that
only pretrial ofcers (M=4.72, SD =0.67) had more exposure to
pretrial risk assessment tools than judges (M=3.64, SD =1.40,
p<0.001), prosecutors (M=3.70, SD =1.66, p<0.001), defense
attorneys (M=3.78, SD =1.20, p<0.001), and other court
actors (M=3.67, SD =1.43, p=0.001). Second, pretrial
decision‐makers also varied in their perceived value of risk
assessment tools (p<0.001, η
2
=0.22). Specically, pretrial
ofcers (M=4.46, SD =0.81) valued risk assessment tools more
highly than prosecutors (M=3.62, SD =0.76, p<0.001).
Additionally, defense attorneys (M=3.25, SD =1.08) held a
lower perceived value of risk assessments compared to pretrial
ofcers (p<0.001), judges (M=4.08, SD =1.20, p<0.001) and
other court actors (M=4.25, SD =0.94, p<0.001). Finally,
decision‐makers also differed in their perceptions of the goals of
the criminal‐legal system (p<0.001, η
2
=0.22). Pretrial ofcers
(M=3.53, SD =0.98, p=0.014), judges (M=3.75, SD =0.84,
p=0.009), and prosecutors (M=4.11, SD =0.94, p<0.001)
viewed the criminal‐legal system as more punitive than defense
attorneys (M=2.96, SD =1.53).
TABLE 2 |Perceptions of risk and utility of risk information by role.
Measure
Pretrial
ofcer Judge Prosecutor
Defense
attorney Other
F(4,276) p η
2
M SD M SD M SD M SD M SD
Exposure to risk
assessment tools
4.72 0.67 3.64 1.40 3.70 1.66 3.78 1.20 3.67 1.43 10.78 <0.001 0.14
Value of risk
assessment tools
4.46 0.81 4.08 1.20 3.62 0.76 3.25 1.08 4.25 0.94 19.81 <0.001 0.22
Goals of the
criminal‐legal
system
3.53 0.98 3.75 0.84 4.11 0.94 2.96 1.53 3.63 1.28 7.55 <0.001 0.10
Note: N =281.
7 of 17
4.2
|
Aim 2: Risk Communication and Pretrial
Decisions
Hierarchical logistic regression models examining the effect of
risk communication on pretrial decisions are presented in Table 3.
Block 1 shows that there were no differences in the odds of being
released on monetary bond for court decision‐makers who
received the 30% (p=0.775), 40% (p=0.749), or 50% (p=0.157)
risk assessment value when compared to the reference category of
30%–50%. In Block 2, the addition of offense violence signicantly
improved the model t (p=0.010). Specically, presentation of a
violent offense relative to a nonviolent offense nearly doubled the
odds of court decision‐makers releasing the defendant on mon-
etary bond (OR =1.94, p=0.010). In Block 3, the addition of a risk
communication by offense violence interaction did not improve
the overall model t (p=0.656). Further, the risk assessment
value remained non‐signicant when compared to a 30%–50%
risk probability with a nonviolent offense, and offense violence
became non‐signicant (p=0.075).
Table 3additionally contains the hierarchical linear regression
models for the likelihood of pretrial release with conditions. In
Block 1, none of the risk assessment values signicantly con-
tributed to the prediction of the likelihood of pretrial release. The
addition of offense violence in Block 2 signicantly improved
model t (p=0.002). More specically, a violent offense resulted
in a 12.99% point higher likelihood of release with conditions
(p=0.002). In Block 3, the interaction between risk assessment
information and offense violence did not improve model t
(p=0.304). Further, none of the main effects, including offense
violence (p=0.176) or the interaction effects signicantly pre-
dicted the likelihood of release with conditions.
We conducted a sensitivity analysis, not presented but available
upon request, to compare the 40% risk presentation against the
30%–50% risk assessment presentation. We wanted to assess
whether decision‐makers made decisions that fell in the middle
of the risk assessment presentation interval. In all three blocks
of the hierarchical logistic regression model, risk assessment
presentation, offense violence, and the interaction between risk
assessment presentation and offense violence did not signi-
cantly predict whether the court decision‐maker released the
defendant on monetary bond over ROR. The hierarchical linear
regression model revealed that only the addition of offense
violence in Block 2 raised the likelihood of release with condi-
tions by 17.18% points (p=0.005).
4.3
|
Aim 3: Factors Driving Pretrial Release
Decisions
4.3.1
|
Multivariable Results
The multivariable logistic regression model for the bond decision is
presented in Table 4. We omitted risk assessment presentation
from the multivariable models due to its lack of effect in previous
models. Similar to the hierarchical logistic regression model, the
multivariable logistic regression model showed that decision‐
makers who were exposed to the violent offense were twice as
likely (p=0.022) to release a defendant on monetary bond over
ROR. Prosecutors were 4.25 times more likely to recommend
monetary bond than ROR (p=0.001) compared to pretrial ofcers.
Finally, each additional year of professional experience in the
criminal‐legal system resulted in a lower likelihood of recom-
mending monetary bond over ROR, by a factor of 0.04 (p=0.050).
Table 4also contains our multivariable model for the likelihood of
release with conditions. When pretrial decision‐makers were
exposed to a defendant with a violent offense as opposed to a
nonviolent offense, the likelihood of release with conditions was
11.14% points higher (p=0.009). Additionally, compared to
TABLE 3 |Regression models of risk communication and offense violence on pretrial release decisions.
Conditional effect by block
Pretrial release decision
Likelihood of releasing the
defendant with conditions
OR SE p95% CI ΒSE p95% CI
Block 1
Risk assessment probability (30%–50%)
30% 0.90 0.33 0.775 0.44, 1.85 −7.87 6.05 0.195 −19.78, 4.04
40% 0.89 0.33 0.749 0.43, 1.84 −6.11 6.11 0.319 −18.14, 5.93
50% 1.63 0.57 0.157 0.83, 3.22 3.53 5.60 0.557 −8.27, 15.33
Block 2
Offense violence (nonviolent) 1.94 0.51 0.010 1.17, 3.24 12.99 4.21 0.002 4.70, 21.28
Δ−2LL χ
2
(1) =6.70, p=0.010 χ
2
(1) =9.52, p=0.002
Block 3 0.384
Risk (30%–50%) �offense violence (nonviolent) 11.96 0.351
30% �violent 1.04 0.82 0.963 0.22, 4.86 −10.43 12.08 0.667 −33.96, 13.10
40% �violent 0.46 0.36 0.316 0.10, 2.10 11.27 11.85 0.384 −12.50, 35.03
50% �violent 0.61 0.44 0.492 0.15, 2.52 5.10 11.96 0.351 −18.22, 28.41
Δ−2LL χ
2
(3) =1.62, p=0.656 χ
2
(3) =3.63, p=0.304
Note: N =298.
8 of 17 Behavioral Sciences & the Law, 2025
pretrial ofcers, defense attorneys were 15.62% points (p=0.013)
less likely to release with conditions. Finally, for every one unit
increase in the self‐reported value of risk assessment tools, par-
ticipants' self‐reported likelihood of ordering release with condi-
tions was 2.05% points higher (p=0.038).
4.3.2
|
Qualitative Findings
Coders produced 11 main theme domains for the decision‐
makers’ rationale and these themes and their subthemes are
presented in Table 5. The most prevalent theme was case facts
(52.7%), which included discussion of the offense violence, harm
to the victim, defendant aggression, the felony‐level nature of the
crime, and the defendant eeing the crime scene. The second
most prevalent theme was risk appraisal (51.3%). Under this
theme, decision‐makers primarily discussed whether they
believed the defendant was likely to commit misconduct,
explicitly referenced risk assessment information, and expressed
concerns about victim or public safety. Community stability was
the third most discussed theme (48.3%) where decision‐makers
highlighted the importance of the defendant's employment,
family ties, and the defendant's mental health or substance use.
Decision‐makers also discussed the importance of the defendant's
criminal history (47.0%) which included how the defendant had
no prior FTA, had a limited criminal history, and had a mix of
prior misdemeanor and felony offenses with occasional mention
of the distal nature of the criminal history. Many pretrial
decision‐makers wrote about pretrial monitoring (45.3%), with
many citing pretrial monitoring as a useful supervision strategy
for the defendant's scenario. The next most common theme was
monetary bond (25.8%). Most decision‐makers believed that
monetary bond is a benecial pretrial release strategy for this
particular defendant, although a few decision‐makers expressed
their opposition towards the use of monetary bond. Some
decision‐makers mentioned legal considerations (15.4%) that
constrained their decision‐making such as their local statues or
guidelines, the assumption that the defendant is presumed
innocent, and the assumption that the defendant is culpable.
Approximately 14.4% of decision‐makers wanted to provide other
pretrial services to the defendant such as anger management
classes, mental health treatment, or drug testing. Some decision‐
makers felt constrained by our pretrial release options (9.7%) and
mentioned other conditions they would want to issue before
releasing the defendant such as administrative monitoring,
electronic monitoring, and home detention. Our 10th theme was
anticipated outcomes (9.7%), which primarily consisted of
whether the decision‐maker believed that the defendant would be
able to comply with their release conditions. Our nal theme was
pretrial detention (6.4%). All participants who mentioned pretrial
detention believed that pretrial detention is harmful to the
defendant and unnecessary in this scenario. We more closely
examined risk appraisal and community stability due to the
varied ways in which decision‐makers discussed these topics.
Risk appraisal was the second most common theme in the deci-
sion‐makers’ responses and there was considerable variation in
discussed factors. Table A2 contains a table detailing the variation
in these themes by decision‐makers’ roles. Across decision‐maker
roles, 51.7% of pretrial ofcers, 55.6% of judges, 67.6% of prose-
cutors, 46.3% of defense attorneys, and 50.0% of other criminal‐
legal professionals discussed the defendant's risk to the commu-
nity. Although the purpose of Aim 2 was to examine the effect of
risk assessment presentation on decision‐making, only 16.1% of
respondents explicitly mentioned the risk assessment probability.
TABLE 4 |Multivariable regression models of offense violence and pretrial decision‐maker characteristics on pretrial release decisions.
Measure
Pretrial release decision
Likelihood of release with
conditions
OR SE p95% CI ΒSE p95% CI
Offense violence (nonviolent) 1.99 0.60 0.022 1.11, 3.58 11.14 4.25 0.009 2.76, 19.51
Role (pretrial ofcer)
Judge 1.66 0.83 0.310 0.62, 4.41 −9.66 7.73 0.212 −24.87, 5.55
Prosecutor 5.25 2.51 0.001 2.06, 13.38 3.14 7.38 0.671 −11.38, 17.66
Defense attorney 0.97 0.43 0.940 0.41, 2.30 −15.62 6.28 0.013 −27.99, −3.26
Other 0.95 0.56 0.935 0.30, 2.99 −14.74 8.33 0.078 −31.14, 1.66
Professional experience in the criminal‐legal system 1.01 0.03 0.642 0.96, 1.06 −0.26 0.37 0.491 −1.00, 0.48
Professional experience in the pretrial system 0.96 0.02 0.050 0.93, 1.00 −0.37 0.29 0.195 −0.94, 0.19
Exposure to risk assessment tools 0.86 0.11 0.240 0.68, 1.10 2.05 1.83 0.263 −1.55, 5.64
Value of risk assessment tools 1.00 0.16 0.979 0.74, 1.37 4.73 2.26 0.038 0.27, 9.18
Goals of the criminal‐legal system 1.26 0.16 0.065 0.99, 1.61 3.07 1.79 0.087 −0.45, 6.58
Race (White)
Black 0.65 0.45 0.534 0.16, 2.55 4.63 9.02 0.609 −13.14, 22.39
Hispanic 2.22 1.08 0.100 0.86, 5.76 −0.44 7.41 0.953 −15.03, 14.16
Gender (male) 0.99 0.30 0.975 0.54, 1.81 −5.34 4.48 0.234 −14.16, 3.48
Age 1.02 0.02 0.379 0.98, 1.06 0.58 0.30 0.052 −0.01, 1.18
Note: N =276.
9 of 17
TABLE 5 |Themes considered in pretrial decision‐making across all pretrial decision‐makers.
Theme n%
Case facts 157 52.68
Nonviolent 59 19.80
Crime of opportunity 42 14.09
Violent 39 13.09
Harm to victim 25 8.39
Unknown victim 25 8.39
Defendant aggression 22 7.38
Felony crime 21 7.05
Fled crime scene 8 2.68
Risk appraisal 153 51.34
Not likely to have misconduct 69 23.15
Explicitly mentions risk assessment information 48 16.11
Likely to have misconduct 45 15.10
Concern about victim safety 20 6.71
Concern about public safety 18 6.04
No concern about victim safety 10 3.36
Confusion about risk assessment information 6 2.01
Risk assessment bad 5 1.68
Community stability 144 48.32
Employed 115 38.59
Family ties 27 9.06
Suspected drug use 17 5.70
Mental health 16 5.37
Housing 11 3.69
No evidence of drug use 11 3.69
Criminal history 140 46.98
No prior FTA 46 15.44
Limited 45 15.10
Felony and misdemeanor 25 8.39
Distal 14 4.70
Felony 12 4.03
Pretrial monitoring 135 45.30
Helpful 123 41.28
Unnecessary 11 3.69
Monetary bond 77 25.84
Benecial 47 15.77
Not needed 27 9.06
Cannot afford 11 3.69
Legal considerations 46 15.44
Local statue/guidelines 19 6.38
Defendant presumed innocent 13 4.36
Defendant is culpable 12 4.03
(Continues)
10 of 17 Behavioral Sciences & the Law, 2025
These respondents included 19.1% of pretrial ofcers, 19.4% of
judges, 32.4% of prosecutors, 6.3% of defense attorneys, and 20.8%
of other criminal‐legal professionals. Thus, risk assessments are
not as important in decision‐making as other factors such as
offense violence (32.9%). We designed our survey using extreme
risk assessment values to increase variation in responses. In
practice, having a 30% or higher risk assessment score indicates a
high likelihood of committing misconduct or failing to appear.
Many individuals who mentioned the risk assessment value
interpreted values as less risky than these values are in risk
assessment practice. One judge wrote:
A 30% to 50% chance of a violation is also a 50% to 70%
chance of no violation… a little better than a coin toss.
In place of using risk assessment scores to determine a de-
fendant's risk to the community, most decision‐makers (46.0%)
discussed risk in more general terms. In particular, decision‐
makers typically identied their own risk factors such as case
facts, criminal history, and community stability. Further, many
decision‐makers made subjective statements about the de-
fendant's risk to the public or the victim. For example, a pretrial
ofcer wrote:
The defendant is before the Court for bond on a vio-
lent felony with prior felony and misdemeanor con-
victions…. The defendant has no prior failure to
appear offenses and has local ties to the community
through his residence and employment, indicating
that he is not an apparent risk of failing to appear in
Court. However, considering the nature of the instant
offense along with the defendant's prior criminal re-
cord, this Pretrial Ofcer would submit that pretrial
supervision may be the added measure of account-
ability needed to mitigate the defendant's potential
risk to public safety.
Many decision‐makers, particularly defense attorneys (30.5%),
pretrial ofcers (25.8%), and other criminal‐legal professionals
(29.2%) as opposed to judges (11.1%) and prosecutors (13.5%),
determined that the defendant was not likely to commit
misconduct during the pretrial period (23.2%). Conversely 15.1%
of respondents believed that the defendant would commit
misconduct, especially prosecutors (35.1%), judges (19.4%), and
pretrial ofcers (16.9%) relative to defense attorneys (3.2%), and
other criminal‐legal professionals (4.2%). Across roles, most
decision‐makers did not consider either victim safety (6.7%) or
public safety (6.0%), and some even expressed no concerns with
victim safety (3.4%).
A few respondents (2.0%) desired more information about the
risk assessment tool which may have hindered their use of the
risk assessment probability in their decision‐making. These re-
spondents included one pretrial ofcer (1.1%), two judges
(5.6%), two prosecutors (5.4%), and one defense attorney (1.1%).
More specically, a few of our respondents expressed confusion
about what the risk assessment probability predicts, the infor-
mation that goes into calculating a defendant's probability, and
how we produced the risk assessment probability.
Finally, a few respondents (1.7%), two judges (5.6%) and three
defense attorneys (3.2%), discussed how they do not use risk
assessment scores in their decision‐making. A defense attorney
wrote,
While defendant has prior convictions he is not
currently under supervision and it appears he suc-
cessfully completed whatever was required of him for
the two priors so he is a good risk for supervision. He
does not appear to be a ight risk and pretrial super-
vision will sufce to cover concerns for danger to the
community. Money bail should not be used as an
impediment to release and I am all too aware of the
shortfalls of RAT‐ even validated RATs‐ so I look for
all ways to promote release, particularly where as
here, he appears to have paid his debt to society,
successfully completed prior sentences AND is
employed and has ties to the community. Additionally
as is often lost‐ he is presumed innocent.
Although only a minority of defense attorneys expressed how
they do not believe in the validity of risk assessment tools, we do
not completely know why they hold a skeptical view of the tool
and how to address their concerns.
TABLE 5 |(Continued)
Theme n%
Other pretrial services 43 14.43
Needs services 32 10.74
Other pretrial release conditions 29 9.73
Anticipated outcomes 29 9.73
Compliance with release conditions 21 7.05
Pretrial detention 19 6.38
Harmful to defendant 14 4.70
Unnecessary 7 2.35
Bad 1 0.34
Note: N =298.
11 of 17
Many respondents (48.3%) mentioned community stability or
wanted more information about the defendant's life in the com-
munity. A breakdown of these themes by decision‐maker role is
presented in Table A3. Community stability factors appeared to
be an important decision‐making factor for defense attorneys
(60.0%), other criminal‐legal professionals (54.2%), pretrial of-
cers (49.4%), and judges (47.2%), and of lesser importance to
prosecutors (24.3%). Decision‐makers considered employment to
be the most important community stability factor to assist in
decision‐making (38.6%), particularly amongst defense attorneys
(47.4%), other criminal‐legal professionals (50.0%), pretrial of-
cers (39.3%), and judges (33.3%) when compared to prosecutors
(21.6%). As an example, a defense attorney wrote:
It was a theft charge and did not note any victim. It is
more benecial to allow the defendant to earn income
and attempt to support his family than it is to detain
him and have the state foot the bill pending a hearing.
Family ties were the second most prevalent community stability
theme, but a smaller proportion of court decision‐makers
mentioned family ties (9.1%). More specically, only 7.9% of
pretrial ofcers, 8.3% of judges, 12.6% of defense attorneys,
16.7% other criminal‐legal professionals, and one prosecutor out
of 36 prosecutors (2.7%) referenced familial connections. Even
fewer decision‐makers mentioned defendant housing (6.7%),
despite housing being an important condition of pretrial release
conditions such as pretrial supervision. These respondents
included pretrial ofcers (5.6%), and defense attorneys (6.3%).
Finally, we did not mention or intend for our ctional defendant
to have substance use or mental health concerns, but a few
respondents inferred that the defendant had these concerns.
Some participants (5.7%) suspected that the defendant used
drugs or had a substance abuse use that affected the defendant's
behavior, and typically suggested that the defendant receive a
drug screening, have a pretrial release condition of drug testing,
and/or receive treatment for their substance abuse. Fewer par-
ticipants (3.7%) highlighted that the scenario did not mention
substance use. Similarly, 5.4% of respondents expressed con-
cerns that the defendant may have mental health concerns and
stated that the defendant likely committed the crime due to
their mental health issues. As such decision‐makers suggested
that the defendant should receive a mental health screen, and/
or receive appropriate treatment.
5
|
Discussion
Pretrial risk assessments are an important element of pretrial
reform, but the eld lacks research on which factors inuence
the consideration of risk assessment information in pretrial
decision‐making. We contributed to this literature by examining
the perceptions and utility of risk assessments, their effects on
pretrial release decisions, and self‐identied decision‐making
rationale in a sample of criminal‐legal professionals. We found
that pretrial ofcers had the most exposure to risk assessment
tools and more highly valued risk assessment tools relative to
prosecutors. Defense attorneys, in contrast, had the lowest rat-
ing of these tools but also held a more balanced view of the
criminal‐legal system relative to their fellow participants who
viewed the system as more punitive. Second, risk assessment
presentation alone and the interaction between risk assessment
presentation and offense violence had a null effect on pretrial
release outcomes. Finally, pretrial decision‐makers largely
considered legal factors in their decisions, but they also
considered risk appraisal through subjective interpretation of
risk assessment values or creating their own risk criteria.
Further, the role of the decision‐maker, their experience in the
criminal‐legal system, and their value of risk assessment tools
inuenced pretrial decisions.
Similar to previous research, we found differences in perceived
values of pretrial risk assessments across pretrial decision‐
makers. More specically, our ndings showed that pretrial of-
cers highly value risk assessments more than prosecutors and
defense attorneys. Previous research found a similar trend
(DeMichele et al. 2019; Terranova et al. 2020), noting that pretrial
ofcers tend to prefer risk assessments for their objective evalu-
ation of the defendant compared to prosecutors and defense at-
torneys who hold less favorable views of risk assessments because
risk assessment results do not always align with their pretrial
arguments. Considering that court decision‐makers work
together to create pretrial decisions, it may be benecial to pro-
vide defense attorneys and prosecutors training on how risk
assessment information can be used in the context of pretrial
arguments. Although not a focus of the present study, greater
understanding of the information that is—and is not—included
in risk assessment data (e.g., the presence or absence of specic
criminogenic risks and needs) may help defense attorneys and
prosecutors advance pretrial arguments even in the face of
extreme risk assessment results (e.g., “high risk” or “low risk”
defendants). Such training may help improve local court cultures
around risk assessment use in order to dissuade resistance to the
use of such tools more broadly (Brayne and Christin 2021).
We found no effect of variable risk assessment presentation on
pretrial release decision, similar to prior research on risk
communication in a judicial context (Evans and Salekin 2014;
Monahan and Silver 2003; Rachlinski et al. 2012). Perhaps
decision‐makers who were presented with the variable risk as-
sessments may have tended to deviate toward the middle of this
estimate (i.e., 40%) to make their decision. The phenomenon of
attribute substitution may help explain this result, where
decision‐makers simplify the provided information to be able to
balance all of the presented information (e.g., case summary,
pretrial release options, their decision‐making rationale; Kah-
neman and Tversky 1979). Although no research has examined
numerical literacy among courtroom decision‐makers, this nu-
merical shorthand is plausible due to the poor numerical liter-
acy of the general population (Peters 2012). Conversely, we
found some difference between the static risk presentations of
30% and 50%, suggesting extreme differences in perceived risk
may impact release decision‐making as opposed to minor dif-
ferences in risk estimates or more certain versus less certain
estimates. Considering that the majority of defendants are low
to moderate risk, this nding raises concerns about how
decision‐makers interpret risk assessment information at lower
risk levels. Specically, low risk defendants who are placed at
disproportionately higher levels of supervision can lead to
increased rates of pretrial supervision failure (Keebler 2009).
12 of 17 Behavioral Sciences & the Law, 2025
Our qualitative ndings suggest that the lack of a risk commu-
nication effect may be due to decision‐makers ignoring or
deprioritizing risk assessment information in their decisions,
which aligns with previous research on risk assessment use (Copp
et al. 2022). Although few respondents expressed concerns with
risk assessment tools, some respondents may have placed less
priority on risk assessment information because they valued their
knowledge over a risk assessment recommendation (Brayne and
Christin 2021), were skeptical of the face validity of risk assess-
ments (Terranova et al. 2020), or believed risk assessments limited
their discretion (DeMichele et al. 2019). Further, prosecutors and
defense attorneys held a less favorable view of risk assessments
than pretrial ofcers, judges, and other decision‐makers, which
may have contributed to the lack of pretrial risk assessment
utility. Alternatively, the underrepresentation of prosecutors,
judges, and other criminal‐legal professionals in the sample may
have resulted in non‐representative perspectives on risk assess-
ment tools.
Among respondents who used risk assessment information in their
decision‐making, our qualitative results revealed that these re-
spondents may have interpreted risk probabilities to represent
lower risk levels than detailed in most pretrial tools. In contrast, risk
probabilities over 30% typically indicate “high” levels of pretrial
misconduct risk among normed tools. We did not provide re-
spondents with instructions on how to interpret risk assessment
information, meaning that decision‐makers relied on their previ-
ous training—or potentially lack thereof—to interpret this infor-
mation. Previous research has found that court decision‐makers
seek training on how to interpret and use risk assessment infor-
mation (DeMichele et al. 2021). However, the extent to which
decision‐makers receive training, the quality of such training, and
the effects of the training on risk assessment application are un-
known (Scott‐Hayward and Fradella 2019). Nevertheless, consid-
ering that the implementation of risk assessments affects pretrial
release decisions (Copp et al. 2022; Stevenson 2018) and that more
jurisdictions are implementing risk assessments (Lattimore
et al. 2020; Pretrial Justice Institute 2019), pretrial decision‐makers
need more specialized training on how to interpret and utilize risk
assessment information. In particular, decision‐makers may
benet from a structured decision‐making framework that pro-
motes use of and adherence to risk assessment recommendations
(Garrett and Monahan 2020).
Lastly, our ndings highlight how court decision‐makers follow
their own criteria in their decision‐making. Decision‐makers
often referenced legal characteristics such as offense violence
and criminal history or other risk factors such as an individual's
likelihood to commit misconduct and risk to the victim or the
public. Prior research shows criminal‐legal professionals gener-
ally agree that legal factors and risk factors are important factors
to consider for pretrial release decisions (Campbell et al. 2021;
DeMichele et al. 2019). However, our results also suggest that the
importance of these factors varies by court decision‐maker role, as
previously found by DeMichele et al. (2019). Many risk assess-
ments consider these factors in their risk assessment evaluations
(Mayson 2017), which—combined with training on how to use
pretrial assessments (Latessa and Lovins 2010)—may aid in the
buy‐in for decision‐makers to use risk assessment information.
However, one less discussed caveat of using legal characteristics
and other risk assessments in courtroom decision‐making is the
potential duplicate consideration of risk factors. At this time,
there is no guidance on whether decision‐makers should use risk
assessment information plus other factors mentioned in the tool
(e.g., criminal history, offense violence, substance use, etc.). This
situation raises questions about whether the duplication of risk
factors in decision‐making could contribute to over‐estimating
risk and therefore less accurate risk estimation, while at the
same time exacerbating issues of predictive bias.
Overall, our ndings suggest various ways to improve pretrial
decision‐making. For one, risk assessment creators need to
provide more transparent information (e.g., items considered,
item weights, predicted pretrial outcomes) about their risk
assessment tool to decision‐makers. This practice would pro-
mote a more common understanding of risk and its applicability
in the pretrial eld. Second, decision‐makers need more training
on how to use risk assessment information and incorporate
pretrial risk assessments into their role. For example, Advancing
Pretrial Policy and Research (APPR) provides transparent in-
formation about the items the Public Safety Assessment (PSA)
uses, how that PSA weighs that information, and how to
interpret the risk assessment information (APPR 2023). Addi-
tionally, APPR hosts training sessions on how to use the PSA,
and how to integrate risk assessments into pretrial decision‐
making. To increase transparency about the development,
validation, and utilization of risk assessment tools, Stanford Law
School created factsheets about the ve most popular risk as-
sessments (Stanford Law School 2023). Demystifying risk as-
sessments through trainings and transparency may assist in
improving risk assessment use in a pretrial context.
5.1
|
Limitations
Our study design limits our ndings. First, simulated case sce-
narios may have produced different decisions than those court
decision‐makers make in a real courtroom. More specically,
some information that is commonly available to judicial
decision‐makers, such as race, has shown to produce different
outcomes for defendants (Demuth and Steffensmeier 2004;
Hissong and Wheeler 2019; Petersen 2020; Sacks et al. 2015;
Wooldredge et al. 2015). We omitted the defendant's race from
the pretrial services report to reduce race as a cofounding var-
iable in our analyses. Second, the small sample size limited our
statistical power, reducing our ability to detect signicant effects
in quantitative analyses. However, given the limited previous
research on risk assessment guided decision‐making, the pur-
pose of this study was largely to advance a limited evidence
base. Finally, our snowball sampling strategy may have reduced
the external generalizability of ndings, despite a broad repre-
sentation of states. In particular, the sample may have been
overrepresented by individuals who felt strongly about pretrial
decision‐making and risk assessment and self‐selected into the
sample. Additionally, there was some overrepresentation of
certain states due to the sampling strategy and the authors'
connections to court actors. Despite these limitations, our study
showed issues within the current implementation of pretrial
risk assessment tools across the U.S.
13 of 17
5.2
|
Future Directions
Future studies should consider testing different risk communi-
cation formats such as categorical and numerical presentations to
determine the relative punitiveness of these presentations in a
pretrial context. Other studies may consider issuing a similar
vignette and questionnaire followed by a test of the decision‐
makers’ knowledge and application of risk assessments to better
grasp their understanding of risk assessments. Future research
could test the effects of different risk communication formats
among decision‐makers that have taken a course on risk assess-
ment tools to ensure that decision‐makers understand how to use
a risk assessment value in their decisions. Additional research
could focus on decision‐making within each role to tease out how
risk assessment implementation and use varies across roles.
Finally, the broader eld of pretrial risk assessment may benet
from observational studies of risk assessment use in pretrial
courtrooms to increase the external validity of ndings on how
decision‐makers use and consider risk assessments in practice.
6
|
Conclusion
Jurisdictions increasingly rely on pretrial risk assessment tools to
improve pretrial decision‐making. However, our ndings suggest
some court decision‐makers greatly vary in their interpretations
of risk assessment information or may not rely on risk assessment
information in decision‐making. This variation or lack of use can
hinder the ability of risk assessment information tools to guide
pretrial decisions. Pretrial reform efforts champion risk assess-
ment guided decision‐making, but jurisdictions must optimize
the use of risk assessments in decision‐making in order for risk
assessments to meaningfully improve pretrial outcomes.
Acknowledgments
We thank the George Mason University Ofce of Scholarship, Creative
Activities, & Research for funding this study.
Conicts of Interest
The authors declare no conicts of interest.
References
APPR. 2023. Advancing Pretrial Justice. Advancing Pretrial Policy &
Research (APPR). https://advancingpretrial.org/.
Bechtel, K., A. M. Holsinger, C. T. Lowenkamp, and M. J. Warren. 2017.
“A Meta‐Analytic Review of Pretrial Research: Risk Assessment, Bond
Type, and Interventions.” American Journal of Criminal Justice 42, no. 2:
443–467. https://doi.org/10.1007/s12103‐016‐9367‐1.
Brayne, S., and A. Christin. 2021. “Technologies of Crime Prediction:
The Reception of Algorithms in Policing and Criminal Courts.” Social
Problems 68, no. 3: 608–624. https://doi.org/10.1093/socpro/spaa004.
Campbell, C., K. Henderson, and B. Renauer. 2021. “Report on Exam-
ining Pretrial Detention in Oregon: A Qualitative Analysis of Decision
Making.” Criminal Justice Policy Research Institute 1–55. https://doi.org/
10.13140/RG.2.2.23157.29926.
Cohen, J. 1992. “A Power Primer.” Quantitative Methods in Psychology
112, no. 1: 155–159. https://doi.org/10.1037/0033‐2909.112.1.155.
Copp, J. E., W. Casey, T. G. Blomberg, and G. Pesta. 2022. “Pretrial Risk
Assessment Instruments in Practice: The Role of Judicial Discretion in
Pretrial Reform.” Criminology & Public Policy 21, no. 2: 329–358. https://
doi.org/10.1111/1745‐9133.12575.
Creswell, J. W., and V. L. P. Clark. 2018. Designing and Conducting
Mixed Methods Research. SAGE Publications.
Dawadi, S., S. Shrestha, and R. A. Giri. 2021. “Mixed‐Methods Research:
A Discussion on Its Types, Challenges, and Criticisms.” Journal of
Practical Studies in Education 2, no. 2: 2–36. https://doi.org/10.46809/
jpse.v2i2.20.
DeMichele, M., P. Baumgartner, K. Barrick, M. Comfort, S. Scaggs, and
S. Misra. 2019. “What Do Criminal Justice Professionals Think About
Risk Assessment at Pretrial?” Federal Probation 83, no. 1: 32–41.
DeMichele, M., M. Comfort, K. Barrick, and P. Baumgartner. 2021. “The
Intuitive‐Override Model: Nudging Judges Toward Pretrial Risk
Assessment Instruments.” Federal Probation 85, no. 2: 22–31.
Demuth, S., and D. Steffensmeier. 2004. “Ethnicity Effects on Sentence
Outcomes in Large Urban Courts: Comparisons Among White, Black,
and Hispanic Defendants.” Social Science Quarterly 85, no. 4: 994–1011.
https://doi.org/10.1111/j.0038‐4941.2004.00255.x.
Desmarais, S. L., and E. M. Lowder. 2019. Pretrial Risk Assessment Tools:
A Primer for Judges, Prosecutors, and Defense Attorneys. Safety and
Justice Challenge, John D. and Catherine T. MacArthur Foundation.
http://www.safetyandjusticechallenge.org/wp‐content/uploads/2019/
02/Pretrial‐Risk‐Assessment‐Primer‐February‐2019.pdf.
Desmarais, S. L., J. Monahan, and J. Austin. 2021. “The Empirical Case
for Pretrial Risk Assessment Instruments.” Criminal Justice and
Behavior 49, no. 6: 1–10. https://doi.org/10.1177/00938548211041651.
Desmarais, S. L., S. A. Zottola, S. E. Duhart Clarke, and E. M. Lowder.
2021. “Predictive Validity of Pretrial Risk Assessments: A Systematic
Review of the Literature.” Criminal Justice and Behavior 48, no. 4: 398–
420. https://doi.org/10.1177/0093854820932959.
Evans, S. A., and K. L. Salekin. 2014. “Involuntary Civil Commitment:
Communicating With the Court Regarding ‘Danger to other’.” Law and
Human Behavior 38, no. 4: 325–336. https://doi.org/10.1037/lhb0000068.
Garrett, B. L., and J. Monahan. 2020. “Judging Risk.” California Law
Review 108, no. 2: 439–494.
Heilbrun, K., J. Dvoskin, S. Hart, and D. McNiel. 1999. “Violence Risk
Communication: Implications for Research, Policy, and Practice.” Health,
Risk & Society 1, no. 1: 91–105. https://doi.org/10.1080/1369857990840
7009.
Heilbrun, K., M. L. O’Neill, L. K. Strohman, Q. Bowman, and J. Phili-
pson. 2000. “Expert Approaches to Communicating Violence Risk.” Law
and Human Behavior 24, no. 1: 137–148. https://doi.org/10.1023/A:
1005435005404.
Heilbrun, K., J. Philipson, L. Berman, and J. Warren. 1999. “Risk
Communication: Clinicians’ Reported Approaches and Perceived
Values.” Journal of the American Academy of Psychiatry and the Law 27,
no. 3: 397–406.
Hilton, N. Z., N. Scurich, and L.‐M. Helmus. 2015. “Communicating the
Risk of Violent and Offending Behavior: Review and Introduction to
This Special Issue.” Behavioral Sciences & the Law 33, no. 1: 1–18.
https://doi.org/10.1002/bsl.2160.
Hissong, R. V., and G. Wheeler. 2019. “The Role of Private Legal Repre-
sentation and the Implicit Effect of Defendants’ Demographic Charac-
teristics in Setting Bail and Obtaining Pretrial Release.” Criminal Justice
Policy Review 30, no. 5: 708–730. https://doi.org/10.1177/088740341771
4560.
Indiana EBDM Pretrial Work Group. 2018. Pretrial Practices Manual, 1–
79. https://www.in.gov/courts/iocs/les/pretrial‐work‐group‐practices‐
manual.pdf.
14 of 17 Behavioral Sciences & the Law, 2025
Jung, J., C. Concannon, R. Shroff, S. Goel, and D. G. Goldstein. 2020.
“Simple Rules to Guide Expert Classications.” Journal of the Royal
Statistical Society: Series A 183, no. 3: 771–800. https://doi.org/10.1111/
rssa.12576.
Kahneman, D., and A. Tversky. 1979. “Prospect Theory: An Analysis of
Decision Under Risk.” Econometrica 47, no. 2: 263–291. https://doi.org/
10.2307/1914185.
Keebler, G. 2009. “Pretrial Risk Assessment in the Federal Court.”
Federal Probation 73, no. 2: 3–29.
Kwartner, P. P., P. M. Lyons, and M. T. Boccaccini. 2006. “Judges’ Risk
Communication Preferences in Risk for Future Violence Cases.” Inter-
national Journal of Forensic Mental Health 5, no. 2: 185–194. https://doi.
org/10.1080/14999013.2006.10471242.
Latessa, E. J., and B. Lovins. 2010. “The Role of Offender Risk Assess-
ment: A Policy Maker Guide.” Victims and Offenders 5, no. 3: 203–219.
https://doi.org/10.1080/15564886.2010.485900.
Lattimore, P. K., S. Tueller, A. Levin‐Rector, and A. Witwer. 2020. “The
Prevalence of Local Criminal Justice Practices.” Federal Probation 84,
no. 1: 28–37.
Lowder, E. M., C. L. Diaz, E. Grommon, and B. R. Ray. 2021. “Effects of
Pretrial Risk Assessments on Release Decisions and Misconduct Out-
comes Relative to Practice as Usual.” Journal of Criminal Justice 73:
101754. https://doi.org/10.1016/j.jcrimjus.2020.101754.
Mayson, S. G. 2017. “Dangerous Defendants.” Yale Law Journal 127, no.
3: 490–569.
Monahan, J., K. Heilbrun, E. Silver, E. Nabors, J. Bone, and P. Slovic. 2002.
“Communicating Violence Risk: Frequency Formats, Vivid Outcomes,
and Forensic Settings.” International Journal of Forensic Mental Health 1,
no. 2: 121–126. https://doi.org/10.1080/14999013.2002.10471167.
Monahan, J., and E. Silver. 2003. “Judicial Decision Thresholds for
Violence Risk Management.” International Journal of Forensic Mental
Health 2, no. 1: 1–6. https://doi.org/10.1080/14999013.2003.10471174.
Monahan, J., and H. J. Steadman. 1996. “Violent Storms and Violent
People: How Meteorology Can Inform Risk Communication in Mental
Health Law.” American Psychologist 51, no. 9: 931–938. https://doi.org/
10.1037/0003‐066X.51.9.931.
Nir, E. 2018. “Approaching the Bench: Accessing Elites on the Judiciary for
Qualitative Interviews.” International Journal of Social Research Method-
ology 21, no. 1: 77–89. https://doi.org/10.1080/13645579.2017.1324669.
Peters, E. 2012. “Beyond Comprehension: The Role of Numeracy in
Judgments and Decisions.” Current Directions in Psychological Science
21, no. 1: 31–35. https://doi.org/10.1177/0963721411429960.
Petersen, N. 2020. “Do Detainees Plead Guilty Faster? A Survival
Analysis of Pretrial Detention and the Timing of Guilty Pleas.” Criminal
Justice Policy Review 31, no. 7: 1015–1035. https://doi.org/10.1177/
0887403419838020.
Pretrial Justice Institute. 2019. Scan of Pretrial Practices (No. 2019‐10‐02KN).
https://university.pretrial.org/HigherLogic/System/DownloadDocumentFi
le.ashx?DocumentFileKey=b2bd6339‐8201‐60f4‐c262‐a6317a409b82&force
Dialog=0.
Rachlinski, J. J., A. J. Wistrich, and C. Guthrie. 2012. “Altering Atten-
tion in Adjudication.” UCLA Law Review 60, no. 6: 1586–1619.
Riley, S. 2024. “Overriding (In)justice: Pretrial Risk Assessment
Administration on the Frontlines.” In The 2024 ACM Conference on
Fairness, Accountability, and Transparency, 480–488. https://doi.org/10.
1145/3630106.3658920.
Sacks, M., V. A. Sainato, and A. R. Ackerman. 2015. “Sentenced to
Pretrial Detention: A Study of Bail Decisions and Outcomes.” American
Journal of Criminal Justice 40, no. 3: 661–681. https://doi.org/10.1007/
s12103‐014‐9268‐0.
Scott‐Hayward, C. S., and H. F. Fradella. 2019. Punishing Poverty: How
Bail and Pretrial Detention Fuel Inequalities in the Criminal Justice
System. University of California Press.
Scurich, N. 2018. “The Case Against Categorical Risk Estimates.”
Behavioral Sciences & the Law 36, no. 5: 554–564. https://doi.org/10.
1002/bsl.2382.
Slovic, P., J. Monahan, and D. G. MacGregor. 2000. “Violence Risk
Assessment and Risk Communication: The Effects of Using Actual
Cases, Providing Instruction, and Employing Probability Versus Fre-
quency Formats.” Law and Human Behavior 24, no. 3: 271–296. https://
doi.org/10.1023/A:1005595519944.
Stanford Law School. 2023. Stanford Pretrial Risk Assessment Tools Fact-
sheet Project. Stanford Law School. https://law.stanford.edu/pretrial‐risk‐
assessment‐tools‐factsheet‐project/.
Stevenson, M. T. 2018. “Assessing Risk Assessment in Action.” Minne-
sota Law Review 58: 303–384.
Terranova, V. A., K. Ward, J. Slepicka, and A. M. Azari. 2020. “Per-
ceptions of Pretrial Risk Assessment: An Examination Across Role in
the Initial Pretrial Release Decision.” Criminal Justice and Behavior 47,
no. 8: 927–942. https://doi.org/10.1177/0093854820932204.
Viljoen, J. L., D. M. Cochrane, and M. R. Jonnson. 2018. “Do Risk
Assessment Tools Help Manage and Reduce Risk of Violence and
Reoffending? A Systematic Review.” Law and Human Behavior 42, no. 3:
181–214. https://doi.org/10.1037/lhb0000280.
Viljoen, J. L., L. M. Vargen, D. M. Cochrane, M. R. Jonnson, I. Goossens,
and S. Monjazeb. 2021. “Do Structured Risk Assessments Predict Vio-
lent, Any, and Sexual Offending Better Than Unstructured Judgment?
An Umbrella Review.” Psychology, Public Policy, and Law 27, no. 1: 79–
97. https://doi.org/10.1037/law0000299.
West, B. S. 2017. “The Next Step in Pretrial Release Is Here: The
Administrative Release Program.” Advocate.https://dpa.ky.gov/wp‐
content/uploads/2020/11/2017‐01.pdf.
Wooldredge, J., J. Frank, N. Goulette, and L. Travis. 2015. “Is the Impact
of Cumulative Disadvantage on Sentencing Greater for Black De-
fendants?” Criminology & Public Policy 14, no. 2: 187–223. https://doi.
org/10.1111/1745‐9133.12124.
15 of 17
Appendix A
TABLE A1 |Survey participant representation by state.
State n%
Alaska 18 6.41
Arizona 7 2.49
Arkansas 8 2.85
California 22 7.83
Colorado 13 4.63
Florida 24 8.54
Georgia 20 7.12
Hawaii 2 0.71
Idaho 1 0.36
Illinois 2 0.71
Indiana 63 22.42
Iowa 1 0.36
Louisiana 5 1.78
Maine 4 1.42
Maryland 4 1.42
Michigan 10 3.56
Montana 7 2.49
New Hampshire 10 3.56
New Mexico 1 0.36
New York 3 1.07
North Carolina 3 1.07
North Dakota 2 0.71
Ohio 1 0.36
Oklahoma 4 1.42
Oregon 28 9.96
Pennsylvania 1 0.36
Texas 1 0.36
Vermont 4 1.42
Virginia 9 3.20
Wisconsin 1 0.36
Wyoming 2 0.71
Note: N =281 because 17 participants did not identify a state.
16 of 17 Behavioral Sciences & the Law, 2025
Appendix BAppendix BAppendix B
TABLE A2 |Risk appraisal themes considered in pretrial decision‐making by all pretrial decision‐maker roles.
Measure
Perceptions of risk and utility of risk information by role
Pretrial
ofcer Judge Prosecutor
Defense
attorney Other
n%N%n%n%n%
Risk appraisal 46 51.69 20 55.56 25 67.57 44 46.32 12 50.00
Not likely to have misconduct 23 25.84 4 11.11 5 13.51 29 30.53 7 29.17
Explicitly mentions risk assessment information 17 19.10 7 19.44 12 32.43 6 6.32 5 20.83
Likely to have misconduct 15 16.85 9 25.00 13 35.14 3 3.16 1 4.17
Concern about victim safety 6 6.74 4 11.11 3 8.11 3 3.16 2 8.33
Concern about public safety 6 6.74 2 5.56 6 16.22 3 3.16 1 4.17
No concern about victim safety 3 3.37 0 0.00 0 0.00 7 7.37 0 0.00
Confusion about risk assessment information 1 1.12 2 5.56 2 5.41 1 1.05 0 0.00
Risk assessment bad 0 0.00 2 5.56 0 0.00 3 3.16 0 0.00
Note: N =298.
TABLE A3 |Community stability themes considered in pretrial decision‐making by all pretrial decision‐maker roles.
Measure
Perceptions of risk and utility of risk information by role
Pretrial
ofcer Judge Prosecutor
Defense
attorney Other
n%N%n%n%n%
Community stability 44 49.44 17 47.22 9 24.32 57 60.00 13 54.17
Employed 35 39.33 12 33.33 8 21.62 45 47.37 12 50.00
Family ties 7 7.87 3 8.33 1 2.70 12 12.63 4 16.67
Suspected drug use 7 7.87 4 11.11 1 2.70 4 4.21 0 0.00
Mental health 5 5.62 3 8.33 1 2.70 7 7.37 0 0.00
Housing 5 5.62 0 0.00 0 0.00 6 6.32 0 0.00
No evidence of drug use 1 1.12 3 8.33 1 2.70 6 6.32 0 0.00
Note: N =298.
Appendix C
17 of 17
