Table 7.

Implications for Practice, Policy, and Research.

Implications for research
• More research should be conducted on the reliability of the SARA. Reliability statistics such as Cronbach’s alpha, Macdonald’s Omega, mean inter-item correlation, and inter-rater reliability need to be assessed and reported. • Efforts should be made to standardize how predictive validity is analyzed and reported in research. Although most studies report the AUC, some use different less robust approaches making comparison difficult. • Studies should also systematically report recidivism rates, as well as the mean SARA score of the sample, recidivists, and non-recidivists. This would allow for meta-analysis to be conducted allowing for a more comprehensive understanding of the literature. • More studies should test the individual contributions of the measure’s items, not just in terms of predictive validity and IRR, but also utilizing item-specific analyses such as Item Response Theory. • More research needs to be conducted on the SARA-V3 to establish its validity as only three studies have investigated this since its publication. • Researchers should also focus on verifying the field validity of the SARA measure as most studies use the SARA in a research context, coded from records by researchers with sections or items omitted. Few studies have validated the psychometric properties of the SARA measures according to their intended use. • Future research should evaluate predictive validity at multiple time points to establish the ability of the SARA measures to assess imminent, short, and long-term risk. • More effort should be made to validate the SARA measures in diverse groups such as ethnic minorities, native populations, women, and the LGBTQ+ community.
Implications for practice and policy
• Although the SARA is overall empirically supported, its psychometric properties are somewhat more modest than its widespread use would imply. Its limitations should be kept in mind during assessment. • The SARA had acceptable but modest predictive accuracy. Clinicians should bear in mind that there remains a certain margin of error when assessing clients. • The SARA is an SPJ measure meant to assist with case management. The use of SRR was found to be equivalent to when items are summed up actuarially in terms of predictive validity. SRRs can therefore be used to clinically inform case management without compromising predictive accuracy. • Although the SARA had acceptable predictive accuracy, many of its items did not. Only a handful of items showed a strong association with recidivism. This should be kept in mind when prioritizing treatment goals. • Clinicians should be wary of switching over to the SARA-V3 until more validation research has been conducted. • Clinicians should keep in mind that the SARA measures are not currently validated in minority groups (i.e., ethnic minorities, native populations, women, LGBTQ+). These groups can present risk factors in a different manner. Clinicians utilizing these measures in these groups should be aware of their realities and take this into account when using these measures, while still maintaining a SPJ approach.

Note. AUC = area under the curve; IRR = inter-rater reliability; SARA = Spousal Assault Risk Assessment; SPJ = Structured Professional Judgment; SRR = summary risk ratings.