FORUM on Corrections Research

Family violence risk assessment: Evaluating its importance

by Eunice Kim¹ and Vince Roper
Edmonton Institution, Correctional Service of Canada

Research shows that only a small percentage of men who assault their spouses face criminal charges or punishment. Using empirical evidence from several studies, Dutton² calculated conditional probabilities for the rate of report, detection, arrest and conviction of men who assault their wives: approximately 14% are reported, 7% are detected, 1% result in arrest, 0.75% result in conviction, and 0.38% result in punishment by fine or incarceration. That study suggests that more offenders have perpetrated spousal assault than have been identified. In a recent random file review of 935 offenders,³ 33.7% of the sample had been violent toward a family member. Of all the offenders’ files that contained evidence of family violence, 80.2% were charged for the offence. This is a high number, but remember that only a very small proportion of family violence and spousal assault offences lead to charges, and an even smaller percentage of those that were charged were punished with a fine or incarceration. Research suggests that the number of spousal assault offenders that are incarcerated is not an accurate representation of the actual number of offenders who have committed spousal assault. Further research indicates a high risk for male federal offenders to commit violence against their spouses and family members.⁴ A comparison between male federal offenders and men who assault their wives revealed many similarities. The researchers outlined four risk indicators for family violence specifically pertaining to federal offenders:

1. they have witnessed or experienced abusive family origins
2. they have committed past assaults against family members
3. they have a record of violence against strangers or acquaintances
4. they exhibit personality disorders with anger, impulsivity or behavioural instability (e.g., borderline, narcissistic, antisocial)

Because these risk factors are prevalent in an incarcerated population, we likely underestimate the number of offenders that have actually committed family violence; we also probably underestimate the number of offenders at risk for committing family violence.

The Service must assess an offender’s risk for family violence if psychologists, case management officers and institutional staff are to make informed decisions regarding private family visits and conditional release. Identifying an offender’s risk as a case need also targets the offender’s treatment needs.

Family Violence Risk Assessment pilot project

Concern with the likely under-identification of offenders at risk led a project team to recommend changes to existing assessment procedures. The team developed a simple, three-phase Family Violence Risk Assessment. The team also recommended training in the use of family violence assessment tools and workshops in family violence awareness. To evaluate this new approach, a pilot project was launched in October 1995. To narrow the focus of the pilot, the definition of family violence was restricted to spousal assault by men against their female partners.

Phase One: Case management officers assessed family violence risk during the assessment interview and reported their findings in the new Family Violence History section of the Criminal Risk Summary.

Phase Two: Psychologists assessed family violence risk during the assessement interview and reported their findings in the new Family Violence section of the Psychology Screening Report.

Phase Three: Parole officers assessed family violence risk based on information from the offender’s family and reported their findings in the new Family Violence section of the Community Assessment.

Methodology

The pilot sample consisted of 210 male offenders who went through the Intake Assessment Unit of a federal correctional institution in Alberta; 108 were admissions from July to September 1995 and 102 were admissions from January to February 1996. Inmate consent was obtained by the signing of the Intake Assessment Unit Offender Contract.

Only the first two phases of the pilot projectwere evaluated. Frequency of positiveidentifications of family violence risk in thepre-pilot project sample were compared withthe frequency of positive identifications forfamily violence risk in the post-pilot projectsample. Four markers were used to measurethe impact of the new procedures introducedby the pilot project (See Table 1).

Table 1

All four variables were coded to indicate the presence or absence of family violence risk. Positive identifications for family violence risk were scored as 1. An evaluator (case management psychologist) had to state that the offender was at risk to score the offender as a positive identification. If, for example,the evaluator mentioned that the offender was witness to spousal assault between his parents, but did not the offender’s risk to commit family violence, this could not be scored as 1.

Results

To test for differences between the pre-pilot and the post-pilot group,demographic variables that could confound results were analysed. The mean age of the subjects in the pre-group was 36.5 years. The mean age the subjects in the post-pilot group 35.2 years. There was no significant difference (t= 0.74, p < .05). Chi-square analyses conducted on race, marital status and current convictions for spousal assault variables indicated that there was no association between these variables and group membership.

Discussion

OMS Flag: As seen in Figure 1, there was an extremely low frequency of flags inboth the pre- and post-evaluations.Inconsistencies between case management officers were found. Some were unaware that the Family Violence Flag existed, while others did not recall being given instructions to fill in the Family Violence Flag. Retraining staff for consistent and reliable documentation of family violence risk on OMS seems to be needed.Case Needs Marital/Family: Differences inthe number of offenders identified for family violence risk in the pre-pilot project compared with the post-pilot project were not significant. Answers to Case Need questions were self-reported by the offender. However, reliance onself-reports could result in significant departures from the truth. It is possible this particular question did not evaluate risk, but only the presence or absence of a behaviour pattern. Assuming that the rate of family violence had not drastically increased over time, the number of positive identifications would not have increased for this variable because case management officerswere simply tabulating the incidence of spousal assault.

Figure 1

Figure 2

Criminal Risk Summary: Analysis of criminal risk showed that 37.3% of the post-pilot project sample were positively identified for family violence risk. It is likely that by adding a specific family violence assessment section to the summary, case management officers were forced to confront the issue. This finding is similar to Robinson’s study⁵ where 33.7% of the offender sample, in a national file review, had some history of perpetrating family violence.

Psychology Screening Report: Results of the Psychology Screening Report show that 69.6% of offenders in the post-pilot sample were identified for family violence risk. Of those positively identified for family violence risk, 51.2% were identified for family violence risk based on actual behavioural evidence (see Figure 2). Conversely, 47.9% of the sample identified for family violence risk were identified on the basis of two theoretical risk factors alone (history of child abuse, personality disorder) — not behavioural evidence.

Has family violence risk been over-identified? Should risk factors be eliminated from family violence risk assessment? One reason to keep theoretically excluded risk factors in family violence risk assessment is concern for possible victims. By identifying offenders likely to commit family violence, potential victims can be protected. Another reason to keep theoretical risk factors is to allow for refutability or testability. However, unless risk factors are empirically tested, we will never know whether they are predictive or not.

On the other hand, there is an important reason why we should consider removing theoretical risk factors from family violence risk assessment. Over-predicting violent behaviour results in false labelling, thereby violating the offender’s rights (e.g., an offender may be wrongfully denied the privilege of private family visits). Over-estimating an offender’s risk for family violence could also jeopardize his chances for release on parole.

Conclusion

Federal offenders are at high risk for committing family violence. Therefore, it is prudent to develop a comprehensive family violence risk assessment. Evaluation of this new risk assessment strategy reveals an overall increase in identifications of family violence risk, but the inclusion of theoretical risk factors dramatically increases the number of offenders identified for family violence risk. Follow-up studies may determine whether those offenders with only theoretical risk factors actually did commit violence against their spouses and family members, thereby validating the predictive power of such theoretical risk factors.

1. Both authors can be reached at 21611 Meridian Street, Edmonton, Alberta T5J 3H7.

2. D. G. Dutton, “The criminal justice response to wife assault,” Law and Human Behaviour, 11, 3 (1987): 189–206.

3. D. Robinson, “Federal offender family violence: Estimates from a national file review study,” Forum on Corrections Research, 7, 2 (1995): 15–18.

4. D. G. Dutton and S. D. Hart, “Risk markers for family violence in a federally incarcerated population,” International Journal of Law and Psychiatry, 15 (1992):101–112.

5. Robinson, “Federal offender family violence: Estimates from a national file review study.”