In the first half of 2008 three new peer-reviewed articles were published – all supporting the predictive validity of the Structured Assessment of Violence Risk in Youth (SAVRY).

In January, 2008, Jodi Viljoen, Mario Scalora and colleagues published the results of their studying comparing the predictive validity of the SAVRY to that of the Juvenile Sex Offender Assessment Protocol-II (J-SOAP-II) and to the Juvenile Sexual Offense Recidivism Risk Assessment Tool- II (J-SORRAT-II). The participants were 169 male youth in a residential adolescent sex offender treatment program. The J-SOAP and the J-SORRAT were designed to assess risk specifically for sexual offense recidivism, whereas the SAVRY was designed only to assess general violence risk in youth.

In this sample, the authors found that none of the three instruments were very effective in predicting sexual offense recidivism, but that the SAVRY and J-SOAP-II did predict non-sexual violence. Both instruments also performed better with older adolescents (over 15) than with younger ones (15 and younger).

The study examined offenders’ behavior during treatment (in the facility), with the following results:

Offending During Treatment (AUC Values)

Sexual Aggression           Non-Sexual Aggression
During Treatment             During Treatment

SAVRY                    .52 / .51                         .69 / .59
J-SORRAT-II           .59                                  .58
J-SOAP-II               .61                                  .66

They also examined the tools’ ability to predict community recidivism after the youth were discharged with the following findings:

Postdischarge reoffending in the overall sample (AUC Values)

Sexual Offense     Serious Non-Sexual      Any Offense
Violent Offense

SAVRY                    .53 / .51                     .69 / .56                 .58 / .50
J-SORRAT-II           .53                              .55                          .54
J-SOAP-II               .54                              .63                          .56

It is difficult for any instrument to specifically predict sexual offense recidivism (in part due to low base rates of detected offenses), but rates of general recidivism (even nonsexual violence) are typically much higher, than sexual reoffending even among adjudicated JSO. Consistent with other research, the Viljoen et al. study shows the SAVRY’s usefulness in assessing general violence risk – even among sexual offenders.

In the following (March) issue of the journal Criminal Justice and Behavior, Joanna Myers and Fred Schmidt published results of their study of the SAVRY’s predictive validity among 121 juvenile offenders at 1-year and 3-year follow-up periods. They note that the SAVRY’s strong predictive validity was “robust in prediction of violent recidivism across gender and ethnicity,” (p. 354) a finding that contributes to knowledge of its usefulness for both males and females.

Coding within each SAVRY domain, and for the Summary Risk Rating showed a high degree of reliability (done with a subset of the sample), with Intraclass Correlation Coefficients (ICCs) of .96 for the Historical domain, .89 for the Social/Contextual domain,.92 for the Individual domain, .97 for the SAVRY total score, and .95 for the summary risk rating.

For the primary predictive validity analysis, Myers and Schmidt used Receiver Operating Characteristic (ROC) Analysis. According to the authors:

“ROC analyses produce a graph where sensitivity is plotted against specificity in the form of a curve. The AUC reflects the probability that an individual who recidivated will score higher on the measure than an individual who did not recidivate (Douglas & Webster, 1999). An AUC of .50 indicates that prediction is no better than chance, with an increasingly larger AUC value reflecting greater accuracy of prediction.” (p.349)

Based on these analyses, they found the following “Area Under the Curve” (AUC) statistics:

Violent                    General                 Nonviolent
Recidivism              Recidivism            Recidivism
1-Year/3-Year        1-Year/3-Year      1-Year/3-Year

SAVRY              .66 / .77                     .75 / .76              .80 / .68

The authors did not subject the Summary Risk Rating (SRR) to ROC analyses, so not direct comparisons of AUC statistics can be made between scores and risk judgments, however, the authors did report that the risk judgments did significantly predict violent and general recidivism both at 1-year and 3-year follow-up.

Low Risk   Moderate Risk   High Risk
Rating       Rating                Rating

% Violent Offending
at 3-year follow-up             3% (1/34)     26% (14/54)      56% (19/34)

With regard to general recidivism as an outcome, the authors concluded that SAVRY total score and SRR were both significant predictors for 1-year and 3-year follow-up periods. They state specifically:

“It is useful to note that of those deemed low risk on the SAVRY, only 3% committed any reoffense during 1-year follow-up. This suggests that a low-risk rating is not only associated with a negligible rate for VR but also reveals a low probability for GR” (p.352).

Also in the March, 2008 issue of the journal Assessment, Jennifer Welsh and colleagues report the results of a study comparing the predictive and incremental validity of three adolescent risk assessment instruments among a sample of 133 youth court referrals. The study compared the Structured Assessment of Violence Risk in Youth (SAVRY), the Youth Level of Service/Case Management Inventory (YLS/CMI), and the Psychopathy Checklist:Youth Version (PCL:YV), for assessing risk of general and violent recidivism.

Examining basic correlations, they found the following relationships between the instruments and recidivism:

Violent                  General
Recidivism             Recidivism

SAVRY             .45                           .46
PCL:YV             .33                          .40
YLS/CMI           .21                          .18

Using Receiver Operating Characteristic (ROC) Analysis, the instruments showed the following “Area Under the Curve” (AUC) statistics:

Violent                  General
Recidivism            Recidivism

SAVRY                 .81                       .77
PCL:YV                .73                       .74
YLS/CMI              .64                       .60

The final test in the comparison was that of incremental validity – the extent to which any given instruments accounts for an accurate prediction over and above the information provided by the other instruments. The authors concluded that:

“Of the three instruments investigated, the SAVRY offered the most incremental predictive validity for GR and VR, whereas the PCL:YV followed closely behind” (p. 111)

These results on incremental validity were consistent with those that the SAVRY authors found in their initial SAVRY validation study.

Of particular note was Welsh, et al’s finding that the SAVRY high degree of predictive accuracy for violent recidivism also held for predicting general recidivism:

“The values in the current study suggest that SAVRY classification accuracy is highest (i.e., R2 change of .14) when it is used for violence risk assessment, the intended purpose of the instrument, but can also be a significant predictor for general risk assessment. Because there are few independent studies that have evaluated the SAVRY in a psychometric manner, these results are important in confirming the value of this instrument for which it was originally designed (i.e., VR), and also for more (i.e., GR).”

The SAVRY authors – in collaboration with the National Youth Screening Assistance Project (NYSAP) – have been exploring whether, and the extent to which, the SAVRY might reasonably be extended to assessing risk for general recidivism in adolescents, including a possible “guide” for making a differential Summary Risk Rating for each outcome.

The National Youth Screening Assistance Project (NYSAP) is a technical assistance center founded by Thomas Grisso and based at the University of Massachusetts Medical School. NYSAP is privately funded to provide assistance to juvenile justice programs nationwide that wish to implement evidence-based tools to improve identification of youths’ needs related to mental health problems, substance use, and aggression risk. Based on its comprehensive review of all available instruments, NYSAP has selected the SAVRY as one of its top choices for risk/needs assessment for risk of reoffending among adolescents.


Meyers, J. & Schmidt, F. (2008). Predictive validity of the Structured Assessment for Violence Risk in Youth (SAVRY) with juvenile offenders. Criminal Justice and Behavior, 35, 344-355.

Viljoen, J. L., Scalora, M., Ullman, D., Cuadra, L., Bader, S., Chavez, V., & Lawrence, L. (2008). Assessing risk for violence in adolescents who have sexually offended: A comparison of the J-SOAP-II, SAVRY, and J-SORRAT-II. Criminal Justice and Behavior, 35, 5-23.

Welsh J., Schmidt F, McKinnon L, Chattha H., Meyers J. (2008). A comparative study of adolescent risk assessment instruments: predictive and incremental validity.
Assessment, 15, 104-15.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: