SERN Reliability: Indicators Contributing to Risk and Needs Ratings
Research Highlights: The SERN demonstrated internal consistency reliability with static and dynamic risk indicators.
Why we are doing this study
In 2019, the Romanian Probation Service (RPS) nationally implemented a ‘Scale for the Evaluation of Risk and Needs’ (or SERN) that was modeled earlier on the static and dynamic factors components of Correctional Service of Canada’s (CSC) Offender Intake Assessment process.Footnote 1 This project continues the technical assistance being provided under the auspices of a 2022 joint statement of cooperation between the Probation Directorate of Romania and CSC. The present study explores the reliability of the SERN or the extent to which its items are consistent with an overall 5-level risk and needs rating.
Publication
RIB-23-41
2023
Research in Brief - PDF
SERN Reliability: Indicators Contributing to Risk and Needs Ratings
What we did
A snapshot sample was gathered of 792 probationers (737 men and 55 women) ranging in age 18 to 80 (M= 39.2 years; SD=12.9 years) who were SERN-assessed while under the supervision of RPS, across 42 sites over the course of a
one-month period. SERN level is coded as Low=1, Low-Medium=2, Medium=3, Medium-High=4 or High=5 rating made by probation officers. SAS / ETS software was used for analyzing the internal consistency of the SERN data.
What we found
Table 1 presents the measures of association statistics for the 9 static risk indicators in relation to the overall risk and needs SERN level rating. Cronbach’s α (.80) showed good internal reliability consistency for the static risk indicators with SERN level.
Most of the Chi-square and Cramer’s V statistics between the static risk indicators and the SERN rating were statistically significant and indicated good to moderate associations with several indicators showing some very strong relationships (>.60); previous adult convictions (0.741), earlier probation (0.664) and prison terms (0.606).
It is also observed that for probationers there were no occurrences of a previous history of escape/attempt/unlawfully at large which implies possible elimination from the SERN. As well, there were very few observations and/or statistical significance of the static risk indicator pre-release failure-technical which suggests perhaps collapsing this indicator with pre-release failure new crime.
Indicators (12) | M | SD | Χ2 | p | V |
---|---|---|---|---|---|
Under age | 0.029 | 0.168 | 177.85 | <0.0001 | 0.474 |
Previous adult convictions | 0.179 | 0.384 | 435.69 | <0.0001 | 0.741 |
Violent criminal offences | 0.117 | 0.322 | 130.58 | <0.0001 | 0.406 |
Earlier probationer | 0.138 | 0.345 | 348.97 | <0.001 | 0.664 |
Pre-release failure-technical | 0.003 | 0.050 | 8.66 | <0.070 | 0.105 |
Pre-release failure-new crime | 0.008 | 0.087 | 43.39 | <0.0001 | 0.126 |
Earlier prison term | 0.076 | 0.264 | 291.16 | <0.0001 | 0.606 |
Escape/attempt/unlawfully at large | n/a | n/a | n/a | n/a | n/a |
Confinement for disciplinary reasons | 0.003 | 0.050 | 77.40 | <0.0001 | 0.313 |
Less than six months since last incarceration | 0.002 | 0.050 | 25.63 | <0.0001 | 0.180 |
No crime-free period of one year or more | 0.040 | 0.197 | 33.04 | <0.0001 | 0.204 |
Co-convicted | 0.160 | 0.367 | 57.02 | <0.0001 | 0.368 |
Notes: M = Mean, SD = Standard Deviation, X2= Chi-square, p = p-value, V= Cramer’s V., n/a=not applicable, aCronbach’s α = .80 |
Table 2 presents the measures of association statistics for the 18 dynamic risk indicators in relation to the overall risk and needs SERN level rating. Cronbach’s α (.89) showed good internal reliability consistency for the dynamic risk indicators with SERN level.
All of the Chi-square and Cramer’s V statistics between the dynamic risk indicators and the SERN rating were statistically significant and indicated good to moderate associations with the most robust being difficulty with self-control (0.425), disrespect for persons or property (0.411), and low levels of education (0.407).
Domains (9) Items (18) | M | SD | Χ2 | V | |
---|---|---|---|---|---|
Education | |||||
Less than grade 10 | 0.346 | 0.476 | 131.23 | 0.407 | |
Less than grade 12 | 0.622 | 0.485 | 82.35 | 0.323 | |
Employment | |||||
Unemployed | 0.143 | 0.349 | 91.79 | 0.340 | |
Lacks work skills | 0.043 | 0.203 | 89.86 | 0.337 | |
Family Background | |||||
Family attachment limited | 0.045 | 0.208 | 63.11 | 0.282 | |
Members criminally active | 0.085 | 0.278 | 71.83 | 0.277 | |
Social Contacts | |||||
Affiliated with crime groups | 0.480 | 0.214 | 60.80 | 0.341 | |
Has criminal friends | 0.135 | 0.342 | 87.82 | 0.333 | |
Substance Abuse | |||||
Abuses alcohol | 0.121 | 0.327 | 81.51 | 0.279 | |
Abuses drugs | 0.029 | 0.168 | 14.74 | 0.136 | |
Community Living | |||||
No fixed accommodation or changes addresses | 0.027 | 0.161 | 41.40 | 0.229 | |
Difficulties with financial mgt. | 0.090 | 0.286 | 91.77 | 0.340 | |
Thinking/Behavior | |||||
Has difficulty with self-control | 0.254 | 0.435 | 143.34 | 0.425 | |
Linking actions to consequences is limited | 0.551 | 0.498 | 121.57 | 0.391 | |
Criminal Attitude | |||||
Negative attitude legal system | 0.064 | 0.250 | 89.52 | 0.336 | |
Denies crime/ minimize | 0.144 | 0.351 | 100.51 | 0.356 | |
Values Orientation | |||||
Disrespect for persons/prop | 0.064 | 0.246 | 133.74 | 0.411 | |
Support use of violence | 0.035 | 0.185 | 87.10 | 0.331 | |
Notes: M = Mean, SD = Standard Deviation, X2= Chi-square, all p-values <.0001, V= Cramer’s V. bCronbach’s α = .89 |
What it means
Statistical analyses on a new sample have reaffirmed the value of systematically assessing Romanian probationers with a structured set of static and dynamic risk indicators. Overall, Cronbach’s α (.83) showed good internal reliability consistency for the SERN level. Moreover, it appears that a number of static risk (e.g., previous convictions, probation periods and prison terms) and dynamic risk (e.g., thinking/behaviour, value orientation, education) indicators strongly contribute to the final SERN rating. These findings support the overall risk and needs SERN ratings being made by probation officers. While the internal consistency of SERN instrument is evident in this study, some data-informed revisions and streamlining may be considered with further validation work.
For more information
Please e-mail the Research Branch.
You can also visit the Research Publications section for a full list of reports and one-page summaries.
Prepared by: Larry Motiuk, Ben Vuong, Marian Badea and Valentin Nᾰstase