Combining Static and Dynamic Recidivism Risk Information Into the Five-Level Risk and Needs System: A New Zealand Example

DOI10.1177/00938548211033319
Date01 January 2022
Published date01 January 2022
CRIMINAL JUSTICE AND BEHAVIOR, 2022, Vol. 49, No. 1, January 2022, 77 –97.
DOI: https://doi.org/10.1177/00938548211033319
Article reuse guidelines: sagepub.com/journals-permissions
© 2021 International Association for Correctional and Forensic Psychology
77
COMBINING STATIC AND DYNAMIC
RECIDIVISM RISK INFORMATION INTO THE
FIVE-LEVEL RISK AND NEEDS SYSTEM
A New Zealand Example
DARCY J. COULTER
CALEB D. LLOYD
Centre for Forensic Behavioural Science, Swinburne University of Technology and Forensicare
RALPH C. SERIN
Department of Psychology, Carleton University
Communicating recidivism risk is individualized to each assessment. Labels (e.g., high, low) have no standardized meaning.
In 2017, the Council of State Governments Justice Center (CSGJC) proposed a framework for standardized communication,
but balancing the framework’s underlying principles of effective risk communication (and merging static and dynamic infor-
mation) adds complexity. In this study, we incorporated dynamic risk scores that case managers rated among a routine sample
of adults on parole in New Zealand (N = 440) with static risk scores into the Five-Level Risk and Needs System. Compared
with static risk only, merging tools (a) enhanced concordance with the recidivism rates proposed by CSGJC for average and
lower-risk individuals, (b) diminished concordance for higher-risk individuals, yet (c) improved conceptual alignment with
the criminogenic needs domain of the system. This example highlights the importance of attending to the underlying prin-
ciples of effective risk communication that motivated the development of the system.
Keywords: DRAOR; risk communication; standardized language; dynamic risk; static risk
Across current criminal justice practice, risk tools prompt assessors to communicate risk
for criminal recidivism in idiosyncratic and unstandardized ways. Internationally,
forensic examiners use hundreds of assessment tools with individualized scoring systems
(Neal & Grisso, 2014) and there is low concordance in category placement across tools that
AUTHORS’ NOTE: The views expressed are those of the authors and not necessarily those of the New
Zealand Department of Corrections. Ralph C. Serin is the author and holds the copyright of the Dynamic Risk
Assessment for Offender Re-entry (DRAOR) tool. Caleb D. Lloyd and Ralph C. Serin are co-authors of the
2017 version of the DRAOR scoring manual and co-developers of the DRAOR training program and training
certification. This research was supported by an Australian Government Research Training Program
Scholarship. Correspondence concerning this article should be addressed to Darcy J. Coulter, Centre for
Forensic Behavioural Science, Swinburne University of Technology, Level 1, 582 Heidelberg Road,
Alphington, VIC 3078, Australia; e-mail: dcoulter@swin.edu.au.
1033319CJBXXX10.1177/00938548211033319Criminal Justice and BehaviorCoulter et al. / Combining Static and Dynamic Risk Information
research-article2021
78 CRIMINAL JUSTICE AND BEHAVIOR
purportedly predict the same outcome in the same population (Bourgon et al., 2018).
Furthermore, although clinicians and judges prefer when risk is communicated categori-
cally (e.g., high, medium, or low risk; Evans & Salekin, 2014; Heilbrun et al., 2000), foren-
sic clinicians disagree what numeric degree of risk each category indicates, partly because
the meaning of these labels can legitimately and substantially differ across tools, jurisdic-
tions, and assessment situations (Hilton et al., 2008; Mills & Kroner, 2006).
In this context, the Council of State Governments Justice Center (CSGJC) recently pro-
posed guidelines designed to standardize risk communication, the Five-Level Risk and
Needs System (5-Level System; Hanson, Bourgon, et al., 2017). The system’s five catego-
ries have labels from Level I to Level V, with the latter containing those with the highest
expected recidivism rates. To date, the 5-Level System is arguably the only concerted
attempt to standardize correctional risk assessment communication. It offers a starting
framework that, if adopted widely, provides shared definitions that show promise for guid-
ing tool development and clinical practice toward more precise, understandable risk com-
munication. Yet, collective progress toward standardization is likely to be slow among
adopters of the 5-Level System because correctional agencies (that largely operate as silos
rather than networks) each implement an idiosyncratic suite of tools, making it an idiosyn-
cratic task to reorganize each agency’s risk information to align to the system’s guidelines.
Furthermore, standardization is an ambitious objective because it implies cross-tool concor-
dance across a number of elements that are historically unique to each tool’s developmental
purpose, practical use, and evidence base.
FEATURES OF THE FIVE-LEVEL RISK AND NEEDS SYSTEM
Although not explicitly declared as such, several goals, or principles, of effective risk com-
munication appeared to guide the system developers’ recommendations when defining the
five levels. First, the 5-Level System attempts to communicate risk in a more nonarbitrary
manner. Arbitrary metrics are scales where a score does not communicate its position on the
underlying psychological dimension nor communicates the relationship between a change in
score and the change in magnitude of the underlying dimension (Blanton & Jaccard, 2006).
For example, it is not intrinsically obvious what a score of 4 on the Static-99R communicates
about sexual recidivism risk compared with a score of 3. A user must know that in routine
samples, the predicted 5-year sexual recidivism rate is 11% for people scoring 4 (Hanson
et al., 2016). By contrast, a nonarbitrary recidivism risk tool explicitly communicates the like-
lihood of recidivism. For example, a score on a scale (e.g., 0.55) that is equal to the estimated
likelihood of recidivism for people with that score (e.g., 55%) may be largely deemed nonar-
bitrary. Whereas the 5-Level System retains a categorical structure (an arbitrary metric), it
explicitly defines (some of) its categories by a specified range of absolute recidivism rates,
making two of the five categories (Levels I and V) close to nonarbitrary among knowledge-
able users (and arguably much less arbitrary than current scoring systems).
Partly sacrificing nonarbitrary communication (a criterion-referenced procedure) in ser-
vice of its second goal (achieved through norm-referenced procedures), the developers
designed the 5-Level System to facilitate practical comparisons within the assessed popula-
tion. The structure supports relative risk statements by defining Level III as people with
typical (median) levels of risk and needs in that population (±10% absolute risk), Level II
as people with risk and needs less than average (but still greater than an absolute recidivism

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT