The Development of the Brief Spousal Assault Form for the Evaluation of Risk (B-SAFER): A Tool for Criminal Justice Professionals

4. Pilot Testing

4. Pilot Testing

4.1 Quantitative Analyses

Six police agencies, representing five cities, volunteered to pilot the B-SAFER. One of the B-SAFER developers (P. Randall Kropp) delivered half-day training sessions to selected officers at all of these agencies. Each officer was then provided with a draft B-SAFER manual and asked to complete the B-SAFER coding form and a checklist of recommended risk management strategies on current and recent spousal violence cases. The following police agencies participated in the pilot project yielding a total of 50 completed B-SAFER forms:

Vancouver (B.C.) Police Department.

Twenty-nine (29) B-SAFER forms were completed by officers with the Domestic Violence and Criminal Harassment (DVACH) unit of the Vancouver Police Department.

Nelson (B.C.) Police.

Six (6) B-SAFER forms were completed by officers with the Nelson City Police.

Royal Canadian Mounted Police, Nelson (B.C.) Detachment.

Nine (9) B-SAFER forms were completed by officers with the Nelson RCMP Detachment.

Charlottetown (P.E.I.) Police Department.

Four (4) B-SAFER forms were completed by officers with the Charlottetown Police Department.

Summerside (P.E.I.) Police Service.

Two (2) B-SAFER forms were completed by officers with the Summerside Police Department.

Calgary (Alberta) Police Service.

Training on the B-SAFER was conducted for the Domestic Conflict Unit (DCU) of the Calgary Police Service in May 2003. The Director of the DCU is pleased with the B-SAFER approach and scheduled further training for May 2004. However, at the time of this report, no B-SAFER forms have been forwarded for analysis.

Repeated attempts were made to recruit a law enforcement agency in a francophone or bilingual community in Québec, New Brunswick and Manitoba, however, testing the tool in French remains to be done.

Training on the B-SAFER was also conducted for the Swedish National Police. Pilot testing in the counties of Kalmar, Växjö, and Blekinge was supervised by Professor Henrik Belfrage, a B-SAFER co-author. The Swedish National Police subsequently forwarded data for 283 cases to BCIFV for analysis. We deemed this data to be directly relevant to this report because: (a) the Swedish criminal justice system is similar to Canada's with the presence of a proactive spousal assault policy; (b) as in Canada, police officers in Sweden are required to make recommendations regarding detention and supervision prior to trial; (c) the B-SAFER was developed in collaboration with academics and police agencies in Sweden, so the risk factors were considered directly applicable; (d) previous research on the SARA-PV (Police Version) in Sweden indicated that the structural professional judgment approach could be successfully applied.

Quantitative analysis of the pilot data forwarded to BCIFV by police in Canada and Sweden is summarized in Tables 1 through 6. All analyses of the Canadian data combined the cases from British Columbia and Prince Edward Island, as the PEI sample was too small to make worthwhile a separate analysis. Tables 1 and 2 report the presence of B-SAFER risk factors for the Canadian and Swedish samples, respectively. All of the B-SAFER items were present in at least some cases from both countries, and many were present in a large percentage of cases.

Table 3 reports the average number of Current (past 4 weeks up to and including the incident under investigation) and Past risk factors in each case. In general, the cases from Canada had more risk factors than did those from Sweden, suggesting that the Canadian cases were higher risk. The higher risk of the Canadian cases probably reflects the fact that they came primarily from a specialized investigative unit in Vancouver established to deal exclusively with high risk or difficult to manage cases, whereas those from Sweden came from regular patrol officers.

Table 3 - Number of Risk Factors Ratings (Mean, Standard Deviation)*
  Canada Sweden
Current Risk Factors** 10.14 (3.94) 7.15 (4.15)
Past Risk Factors 10.34 (5.26) 6.09 (4.87)

The finding that the Canadian cases were higher risk is also borne out in Table 4, which summarizes the distribution of risk ratings made using the B-SAFER in Canada and Sweden. The B-SAFER required users to consider the risk to intimate partners if no intervention was taken. Consistent with the structured professional judgement approach, these ratings were made at the discretion of the officers. Police officers were asked to rate:

  • (a) imminent risk (within less than two months) for violence;
  • (b) long-term risk (beyond 2 months) for violence; and
  • (c) risk for extremely serious assault or death.

In each case risk was rated as Low, Moderate, or High (L, M, H).

The findings summarized in Table 4 indicate that in the Canadian sample roughly one third of the cases were considered a high risk for imminent violence, close to half were considered a high long-term risk for violence, and one quarter were consider high risk for severe assault or death. These results should not be over interpreted, however, due to the small sample size and the unrepresentative nature of the cases referred to the specialized Vancouver unit which, as mentioned above, was established to deal exclusively with high risk or difficult to manage cases.

Table 4 - Distribution of B-SAFER Risk Ratings
  Canada Sweden
  Low Mod High Low Mod High
Risk for Imminent Assault (Next 2 months) 35% 27% 39% 44% 47% 9%
Long-Term Risk of Assault (Beyond 2 months) 27% 29% 45% /8% 55% 8%
Risk for Severe Assault / Death Low Moderate High 47% 29% 25% 83% 17% 1%

Table 5 reports the average number of management strategies used in each case in Canada and Sweden. Although more management strategies were recommended by Swedish police than by Canadian police, this appears to be a result of the fact that detention was recommended in about 25% of the Canadian cases but in none of the Swedish cases. The management strategies options were different in the Canadian and Sweden studies, so direct comparisons could not be made. However, the most commonly employed management recommendations in the Canadian sample were:

  • "no contact with victim" (86% of cases),
  • "no go within bounded area" (71%),
  • "no possession of weapons" (51%),
  • "abstain from drugs and alcohol" (37%),
  • "do not attempt to locate victim" (35%),
  • "report to bail supervisor" (29%), and
  • detention (25%).

In the Swedish sample, the most common interventions were:

  • "gather security information" (79%),
  • "initiate a security discussion with victim" (73%),
  • "contact social services for victim" ( 50%),
  • "no-contact order" (49%), and
  • "contact safe house" (11%).
Table 5 - Number of Management Strategies Used by Police
  Mean Number(Standard Deviation) Maximum Number
Canada 5.35 (4.20) 25
Sweden 5.44 (1.77) 17

Perhaps the most important findings thus far are reported in Table 6 below. Table 6 provides the associations (correlations) among the total number of current and past risk factors present on the B-SAFER; risk ratings made using the B-SAFER; and the management strategies recommended in the cases. The correlations suggest that B-SAFER risk factors and risk ratings were substantially associated with the number of management strategies recommended by police, as well as recommendation for detention made in Canada [3]. Simply put, more intervention was recommended in cases perceived to be high risk than in cases perceived to be low risk. For example, risk for imminent violence was correlated at .38 with the total number of management strategies both in the Canadian and Swedish samples. In both countries the correlation was statistically significant, suggesting that it is extremely unlikely that the findings occurred by chance.

Table 6 - Correlations Among B-SAFER Risk Factors, Risk Ratings, and Management Strategies

Overall, the findings of these quantitative analyses on the validity of the B-SAFER Brief Spousal Assault Form indicated the following:

  1. All of the risk factors provided were coded as "present" in a substantial proportion of cases. For example, in the Canadian sample the percentages of cases rated "currently" present ranged from a low of 28% for Mental Disorder to 62% for Relationship Problems. Importantly, there was a low rate (less than 10%) of items "omitted" or unable to be evaluated due to missing information. This suggests that the B-SAFER tool includes relevant risk factors present in spousal assault cases and that the tool can be coded easily by police officers in the course of routine investigations.
  2. Overall or summary ratings of risk were diverse, distributed almost normally in the Canadian samples. This suggests that police officers were able to use the B-SAFER coding instructions to make discriminations among perpetrators.
  3. There was a limited association between B-SAFER ratings and recommended management strategies, and there was substantial variability both within and among officers in their recommendations regarding management. This suggests that police officers' recommendations regarding case management were influenced by their judgements of risk (both the presence of individual risk factors and the overall level of risk), but also that B-SAFER ratings were not highly "prescriptive" with respect to management recommendations.

4.2 Qualitative Feedback

Following the pilot testing, we asked officers from each agency to answer six questions regarding the content and process of the B-SAFER. Eleven (11) of the 50 officers replied. Overall, the feedback was positive. Indeed, the officers in charge of specialized domestic violence units in Calgary and Vancouver have approached their respective provincial governments to recommend or support province-wide use of the B-SAFER in release decision- making by police. Yet some useful suggestions for improvement were offered. The main themes of the officers' responses to each of the six questions are summarized below:

  1. What did you like best about the B-SAFER?

    Overall, officers' said that they found the B-SAFER to be simple and easy to use. Some noted that it encouraged investigators to think about risks in specific and identifiable areas that might otherwise have been overlooked. Others appreciated the item indicators and examples listed on the coding form. Yet others said that the B-SAFER caused investigators to do more standardized and formalized risk assessments. Of note was the following comment: "The B-SAFER provided us with a consistent tool to use in each case, which improved our service to victims."

  2. What did you like least about the B-SAFER?

    Although many officers replied "nothing" to this question, others provided constructive criticism.

    One investigator expressed a concern that in many cases the B-SAFER was completed without the knowledge of the victim. There was also some concern that police officers may have limited knowledge about some of the risk categories, such as those referring to mental disorder.

    Some officers responded that they were uncomfortable completing the risk ratings section of the B-SAFER, indicating that it was difficult to make these determinations. Certain officers were particularly concerned that they would be required to disclose in court the B-SAFER information.

    One officer found the 4-point risk factor rating system complicated. The same officer thought the process required him to make "judgments and assumptions" about the offender and victim that went beyond his role as a police officer.

  3. Would you use the B-SAFER in your own work, or would you recommend it to others?

    Most of the feedback here was very positive. Only one officer answered "no" to this question. He believed that his agency's current investigation procedure meets the needs of the offenders and victims.

  4. Does the B-SAFER contain any risk factors that you think should be changed or deleted?

    Most officers indicated that the B-SAFER was comprehensive and the risk factors appropriate. One respondent indicated that the indicators for risk factor 5, "Negative Attitudes About Spousal Assault," could be expanded to include additional controlling behaviours, such as financial control, verbal and emotional abuse, and manipulative behaviour.

  5. Are there any risk factors missing from the B-SAFER that you think should be added?

    The responses to this question were universally positive. None of the officers indicated that any additions are needed.

  6. Is there anything that could be done to make the B-SAFER easier or more convenient to use?

    One officer noted that the item rating procedure should be simplified, but no specific recommendations were offered. Another suggested that the risk rating section be removed. We received several suggestions that software to assist administration and report writing would greatly facilitate routine use of the B-SAFER, as well as quality assurance.


[3] No recommendations for detention were made by the Swedish police.

Date modified: