The Violent Behavior Vignette Questionnaire (VBVQ): A Measure of Violent Behavior for Research in Forensic and Non-Forensic Settings and Populations

Abstract

The measurement of violent behavior presents serious challenges for research on violence. In the current article, we present initial tests of the construct validity of scores on the Violent Behavior Vignette Questionnaire (VBVQ), which consists of a series of interpersonal conflict vignettes with response options in a multiple-choice format designed to measure current violent behavior. Violent responses on the initial version of the VBVQ generally corresponded to independent indicators of physical aggressiveness and violent behavior among male university students, men in the community, and incarcerated male offenders. We then refined the VBVQ and again tested the validity of its scores in new samples of men in the community and incarcerated male offenders. In both samples, men who selected a violent response option on the VBVQ generally had much higher levels of physical aggressiveness and violent behavior than did men who selected non-violent response options. However, VBVQ responses were not associated with the number of violent offenses in offenders’ official criminal records. Our findings provide some support for the use of the VBVQ in lab and correctional/forensic research, but further research is required to determine whether it offers advantages over other measures.

Keywords

Violence vignettes analog construct validity offenders

The measurement of violent behavior presents serious challenges for research on violence. Criminal violence is (thankfully) a low baserate event that usually requires large samples and long follow-up times to detect adequate numbers of incidents for research (e.g., Harris et al., 2015). This has slowed the progress of research on the causes of violence, how best to assess and treat those factors, and, ultimately, how to reduce violence. A measure that is quick, safe, versatile, and a valid analog of violent behavior would allow research to advance more quickly. In the current article, we conduct initial tests of the construct validity of a self-report measure designed to assess current male violent behavior, for use with forensic and non-forensic populations.

There can be mutual skepticism between researchers who study violence in forensic contexts and those who study aggression in the lab (for discussion of this tension between the different perspectives, see Anderson & Bushman, 1997; Anderson et al., 1999). Lab studies use questionnaires or analogs to measure aggressive behavior, such as ostensibly administering electrical shocks, noise blasts, or hot sauce to a confederate; taking points or resources away from someone; self-reported or role-play responses to hypothetical provocative scenarios/vignettes; and sticking pins in a voodoo doll (Buss, 1961; Cherek, 1992; DeWall et al., 2013; Lieberman et al., 1999; Liebowitz, 1968; O’Connor et al., 2001; Taylor, 1967; Zeichner et al., 1999). This approach has many advantages, such as speed, convenience, and flexibility, which makes it ideal for use in high internal validity randomized experiments.

Researchers who study violent behavior in forensic contexts, however, can be skeptical about the relevance of lab research on aggression for understanding criminal violence (Anderson & Bushman, 1997; Anderson et al., 1999). One of the main concerns is the extent to which the typical operationalization of aggression in lab studies reflects criminal violent behavior; that is, construct validity. Though there is some evidence that some of these self-report and analog measures correlate with indicators of physical aggression (Cherek et al., 1997; Coccaro et al., 2017; Ferguson & Rueda, 2009; Giancola & Parrott, 2008; Giancola & Zeichner, 1995; Golomb et al., 2007; Lieberman et al., 1999; Lim et al., 2011; O’Connor et al., 2001; J. F. Williams et al., 1967; Zeichner et al., 2003), there are some legitimate grounds for skepticism. First, many lab measures of aggression are designed to assess aggression broadly defined and, accordingly, include behaviors beyond criminal violence, such as verbal insults or taking away points (e.g., Cherek, 1992; O’Connor et al., 2001). Second, these measures have rarely been tested for their convergence with indicators of criminal violent behavior (i.e., history of or propensity for actual physical violence). Third, responses on many of these measures are only interpretable relative to other participants rather than in an absolute sense; for example, how much hot sauce, or how many pins should be considered analogous to a violent response? An additional issue may be that many of these measures would be difficult to use in research in correctional/forensic settings. For example, among inmates living together in a prison, it would probably be difficult to keep most potential participants naïve about the true nature of the procedure because they would likely hear about it from the first few participants (e.g., there is no real opponent/antagonist; nobody is actually subjected to your shocks, hot sauce, noise blasts, or other aggressive responses; Cherek, 1992; Lieberman et al., 1999; Taylor, 1967; Zeichner et al., 1999).

In contrast, forensic studies usually use official records of criminal violent behavior (e.g., violent convictions) to operationalize violence (e.g., Campbell et al., 2009). The most obvious advantage of this approach is that official records seem to more obviously reflect the violent behavior that is of interest to forensic researchers, practitioners, policy makers, and the general public; that is, official records appear to have greater construct validity. Lab researchers, however, can be skeptical about the value of many studies of forensic populations for understanding aggression (Anderson & Bushman, 1997; Anderson et al., 1999). The main concern is the lack of experimental control afforded by most studies of criminal violence (e.g., Papalia et al., 2019). For example, randomized experiments are difficult to conduct because of the low baserates and slow timelines for officially documented criminal violence. Further, though official records of violence are perhaps more obviously credible than lab analogs, they are also an imperfect approximation of violent behavior (e.g., Rice et al., 2006). For example, official records of violence do not capture all violent behavior (e.g., assault for which one is not arrested; Theobald et al., 2014). Although there have also been some self-report measures of violent behavior developed specifically for use in forensic settings (Murphy & Preston, 1999; Quinsey et al., 1983), to the best of our knowledge there are no published studies testing their correspondence with indicators of criminal violent behavior.

Thus, we have two literatures each with obvious relevance to the other, but with each seeming to often ignore the other. One reason for this may be the absence of a measure of violent behavior that would satisfy the standards, sensibilities, foci, priorities, and restrictions of research in forensic and non-forensic settings. A measure of current violent behavior that could be used in research in both forensic and non-forensic settings would allow for greater symmetry between these literatures (Anderson & Bushman, 1997). Though researchers would still be examining different populations, they would have the option of using the same criterion, which would make the findings more comparable and mutually relevant.

In the four studies in the current article, we conduct initial tests of the construct validity of scores on a self-report measure designed to assess current violent behavior—the Violent Behavior Vignette Questionnaire (VBVQ)—with samples of men from a university, the community, and prisons. The VBVQ is similar in some ways to other vignette measures of aggression (e.g., Coccaro et al., 2017; Hilton et al., 2003; O’Connor et al., 2001), but differs in that the VBVQ was designed to be a more concentrated assessment of unambiguously criminal violent behavior by men against men. We focus here on male participants and violence perpetrated by men against men for two reasons: (1) the majority of non-sexual criminal violence is committed by men against other men (e.g., Gannon & Mihorean, 2005; Perrault & Brennan, 2010), and (2) we wanted to maximize the precision and efficiency of our studies. Inclusion of other types of violence, such as intimate partner violence, could introduce error because there is some degree of distinctiveness between some types of violence; for example, some men exclusively commit domestic violence (e.g., Dixon & Browne, 2003). The data and methodology that would be required to adequately measure more diverse violence and test the validity of such a measure is beyond the scope of this paper. This is not to minimize the problem of violence committed by and against other groups (e.g., women, adolescents, children), but rather an effort to maximize the impact of our limited resources by focusing them more narrowly for now.

Study 1

In Study 1 we present the initial version of our measure (vignettes) and explore its construct validity with samples of male university students and men from the community. More specifically, the purpose of this study was to explore the baserates of violent responses on our measure, and the relationship between our measure and other measures of violent behavior and physical aggressiveness. Consideration of baserates is relevant to the potential utility of our measure in future research; very low or high baserates for violent responses would be problematic for statistical analyses, and very high baserates would also raise concerns about construct validity because baserates for criminal violence are generally low. With regard to the relationship between measures, if our vignette measure really does assess violent behavior, then it should be associated with other measures of violent behavior and physical aggressiveness.

Method

Participants

Participants included in the main analyses were 226 male university students in the Canadian province of Ontario and 302 men from the community in Canada and the United States. Community participants were recruited from a Qualtrics Panel, which is a pool of participants that Qualtrics has recruited (through a partner company) to participate in research. For the community participants, all recruitment and data collection were performed by Qualtrics and its partners. Study researchers do not directly compensate participants, but rather we pay Qualtrics and its partners, who then provide token awards to their panelists for completing the study. Each partner has its own approach to compensating panelists, but awards can be gift cards, loyalty points, entry into draws for prizes, and cash awards.

These samples excluded a total of 20 additional students and 174 community participants who did not report their gender as male (6 students; 141 community participants), did not report they could understand written English (1 student; 88 community), did not correctly answer all of the quality control items (15 students; 4 community), or did not answer more than 20% of the items on the self-report measures of past violent behavior (Violent Behaviour Scale; VBS; 7 students; 2 community) or physical aggressiveness (Physical Aggression scale of the Aggression Questionnaire; PA-AQ; 9 students; 4 community)—these exclusion categories were not mutually exclusive.

For the students, the median age was 19 years old and most were single (62.0%). For the community participants, the median age was 50 to 59 years old and most were married (48.7%). Average time to complete the survey was 10.39 minutes (SD = 10.35) for the student participants and 10.01 minutes (SD = 9.09) for the community participants.

Measures

Vignettes

Fifteen vignettes depicted interpersonal conflicts with a male antagonist; e.g., a man in line behind you is talking to his friend loudly and rudely and repeatedly bumping into you. See Appendix S1 in the supplemental material for more information about the development of the vignettes and response options. There were two stages per vignette, such that participants would be presented with the initial scenario, provide a response, and—except when they chose other on the first round—they would be presented with a provocative reaction from the antagonist, and would then provide a second response. Participants were instructed to select what they would really do if they were in that situation right now. After each round, participants were asked “What do you do?” and presented with seven response options. The wording of some responses was tailored for each vignette, but they reflected the following generic options and were always shown in the following order:

Ignore it

Without being rude, say something to try to deal with it peacefully

Rudely say something to him

Threaten to hurt him

Shove him

Hit him

Other (specify)

To illustrate, if a participant selected “rudely say something to him” in response to the antagonist who just unapologetically spilled beer on his shirt, then he would be presented with the antagonist insulting him back, and finally he would select a response to the antagonist’s insult (e.g., “threaten to hurt him”).

We presented via online surveys a randomly selected five of the 15 scenarios in a random order to 226 male university students and 302 men from the community (Qualtrics Panel). Sample sizes for each of the 15 vignettes ranged from 71 to 78 for the student sample and 91 to 105 for the community sample.

For the students, there was a total of 154 other responses, ranging from one to 12 per vignette response opportunity (e.g., for the second response opportunity on Vignette 3 there were 12 other responses). We coded the other responses into one of the following four categories: non-violent, threat of violence, use of physical violence, or other (none of the above). Inter-rater reliability for coding other responses was high (>80% agreement) for 26 of the vignette response opportunities, but low in the remaining four vignette response opportunities (50.0% to 66.7% agreement). For the community participants, there was a total of 333 other responses, ranging from four to 28 per vignette response opportunity. Inter-rater reliability for coding other responses was consistently greater than 80% agreement.

We classified responses to each vignette as either violent (threaten, shove, hit, or other threat or use of violence on either round) or non-violent (ignore, talk, insult, or other non-violent on both rounds). We dropped five of the vignettes because their baserate of violent responses was too low among the community and student participants (10% or lower). We also computed a total score for the vignette scale for participants who received at least three vignettes. Specifically, we gave 1 point for each vignette if a violent response was selected for either round of the vignette, and then averaged the points across vignettes. Total scores can range from 0 to 1.

Violent Behaviour Scale (VBS)

The VBS (Nunes et al., 2015) consists of eight self-report items regarding various violent behaviors and involvement with the criminal justice system for violent behavior since age 16 (e.g., “From when you were 16 years old to today, how many times have you started a physical fight with someone?”). Participants rate each item on a 10-point scale ranging from 0 (never) to 9 (9 times or more) and the total score is the mean of all the items. Scores can range from 0 to 9, with higher scores indicating more violent behavior. Though the construct validity of scores on the VBS has not yet been tested, similar self-report measures of antisocial behavior are generally associated with official criminal justice indicators of antisocial behavior (Kroner et al., 2007; Thornberry & Krohn, 2000).

Physical Aggression Scale of the Aggression Questionnaire (PA-AQ; Buss & Perry, 1992)

The PA-AQ consists of nine self-report items, which are rated on a 5-point Likert scale from 1 (extremely uncharacteristic of me) to 5 (extremely characteristic of me). The total score is the mean of all the items and can range from 1 to 5, with higher total scores indicating greater physical aggressiveness. Overall, previous research indicates that the PA-AQ has acceptable reliability and validity. Higher scores on the PA-AQ are associated with independent indicators of physical aggressiveness, such as peer ratings of physical aggressiveness (r = .45, r = .78), history of violent offending (d = .36, d = .46), and higher rates of violent recidivism (d = 0.65; Buss & Perry, 1992; Diamond & Magaletta, 2006; O’Connor et al., 2001; Olver et al., 2014). Previous studies report good internal consistency for the PA-AQ in correctional (.73, .92) and non-correctional (.84 - .86) samples (Bryant & Smith, 2001; Buss & Perry, 1992; O’Connor et al., 2001; Olver et al., 2014; Pettersen et al., 2018).

Quality control items

Five items were designed to identify participants who were not attending to or understanding the study material. They instructed participants to select a specific response (e.g., Click “strongly agree”) and were distributed throughout the survey.

Procedure

This study was approved by our university’s research ethics board. The order in which the measures were completed was counterbalanced across participants.

Results and discussion

The baserate of violent responses (on either the first or second round) for each of the retained 10 vignettes for the student and community samples is presented in Table 1. The any category refers to whether participants selected a violent response to any of the 10 vignettes versus never selected any violent responses to any of the vignettes. In the community sample the baserates for violent responses in the vignettes ranged from a minimum of 14% to a maximum of 34%, suggesting a reasonable range of level of provocation and no floor or ceiling effects. A similar pattern was observed for the student sample, but the range of violent response baserates was larger.

Table 1.

Baserate of violent responses by vignette.

	Study 1		Study 2	Study 3	Study 4
Vignette	Students^a	Community^b	Offenders	Community	Offenders
1	26.03%	21.43%	37.50%	28.27%	32.69%
2	11.84%	14.42%	31.25%	18.71%	11.54%
3	13.33%	14.29%	43.75%	27.27%	36.54%
4	30.26%	29.17%	37.50%	36.70%	38.46%
5	32.00%	27.66%	26.67%	31.56%	36.54%
6	2.67%	15.79%	37.50%	24.03%	21.15%
7	21.92%	18.56%	37.50%	14.50%	11.54%
8	32.43%	22.22%	43.75%	40.77%	30.77%
9	50.00%	33.98%	43.75%	51.29%	44.23%
10	50.00%	29.29%	53.33%	42.06%	34.62
Any^c	51.77%	39.74%	56.25%	73.89%	67.31%

^aSample size ranged from 71 to 78 per vignette.

^bSample size ranged from 91 to 105 per vignette.

^cAny refers to whether participants selected a violent response to any of the 10 vignettes versus never selected any violent responses to any of the vignettes.

Violent responses to the vignettes were moderately to strongly associated with more self-reported past violent behavior and physical aggressiveness. As shown in Figures 1(a) and 2(a), participants who selected a violent response to the vignettes consistently reported more past violent behavior than participants who selected a non-violent response; effect sizes were medium to large (Cohen’s d ranged from 0.55 to 1.47 for the student sample and 0.77 to 1.39 for the community sample). More detailed information relevant to these comparisons is reported in Tables S1 and S3 in the supplemental material (95% confidence interval for each effect size, group means, standard deviations, and sample sizes). Cohen’s d indicates the number of standard deviations by which the means of two groups differ (e.g., d = 0.50 indicates that the group means are half a standard deviation away from each other). According to general guidelines for interpreting effect sizes, d of 0.20 is small, 0.50 is medium, and 0.80 is large (Cohen, 1988). d can also be interpreted in terms of the percentage of participants in one group with scores greater than the mean score of the other group (57.9% for d = 0.20, 69.1% for d = 0.50, 78.8% for d = 0.80) and the percentage of overlap between the groups (92.0% for d = 0.20, 80.3% for d = 0.50, 68.9% for d = 0.80; Magnusson, 2020). As shown in Figures 1(b) and 2(b) (and Tables S2 and S4 in the supplemental material), participants who selected a violent response to the vignettes consistently reported more physical aggressiveness than participants who selected a non-violent response; effect sizes were medium to large (d ranged from 0.48 to 1.55 for the student sample and 0.79 to 1.66 for the community sample). Consistent with these results for the individual vignettes, those who responded with any violence across the vignettes reported much more past violent behavior (d was 0.68 for the student sample and 0.96 for the community sample) and physical aggressiveness (d was 0.98 for the student sample and 1.26 for the community sample) than those who never responded with violence.

Figure 1.

Differences between student participants in Study 1 who selected a violent response on the vignettes versus those who did not select a violent response. (a) Past violent behavior (N = 226). Values at data points are Cohen’s d effect sizes indicating the magnitude of the difference on the Violent Behaviour Scale (VBS) between participants who selected a violent versus a non-violent response for each vignette. (b) Aggressiveness (N = 226). Values at data points are Cohen’s d effect sizes indicating the magnitude of the difference on the Physical Aggression Scale of the Aggression Questionnaire (PA-AQ) between participants who selected a violent versus a non-violent response for each vignette.

Figure 2.

Differences between community participants in Study 1 who selected a violent response on the vignettes versus those who did not select a violent response. (a) Past violent behavior (N = 302). Values at data points are Cohen’s d effect sizes indicating the magnitude of the difference on the Violent Behaviour Scale (VBS) between participants who selected a violent versus a non-violent response for each vignette. (b) Aggressiveness (N = 302). Values at data points are Cohen’s d effect sizes indicating the magnitude of the difference on the Physical Aggression Scale of the Aggression Questionnaire (PA-AQ) between participants who selected a violent versus a non-violent response for each vignette.

As shown in Table 3, the vignette total score was strongly correlated with more self-reported violence (VBS) and more physical aggressiveness (PA-AQ) in the student and community samples (descriptive and internal consistency statistics are presented in Table 2). According to general guidelines for interpreting effect sizes, r of .10 is small, .30 is medium, and .50 is large (Cohen, 1988).

Table 2.

Descriptive statistics and internal consistency for measures.

	M	SD	n	Range	α
Study 1
Students
Vignette total score^a	0.27	0.31	190	0.00–1.00	–^b
Violent Behaviour Scale	0.78	0.93	226	0.00–4.50	.71
Physical Aggression (PA-AQ)	2.42	0.79	226	1.00–5.00	.84
Community participants
Vignette total score^a	0.23	0.32	247	0.00–1.00	–
Violent Behaviour Scale	0.92	1.36	302	0.00–7.25	.86
Physical Aggression (PA-AQ)	2.43	0.87	302	1.00–4.56	.86
Study 2
Offenders
Vignette total score^a	0.39	0.44	16	0.00–1.00	.97
Violent Behaviour Scale	4.74	2.61	16	0.75–9.00	.90
Physical Aggression (PA-AQ)	3.17	1.02	16	1.78–4.56	.84
Violent convictions	10.25	5.51	16	1.00–19.00	–
SAQ	36.19	10.51	16	23.00–54.00	.89
SIR-R1 Scale	−5.22	7.90	9	−15.00 to 10.00	–^c
Study 3
Community participants
VBVQ total score^a	0.31	.028	471	0.00–1.00	.82
Violent Behaviour Scale	0.97	1.26	471	0.00–6.13	.81
Physical Aggression (PA-AQ)	2.55	0.80	471	1.00–4.89	.80
Study 4
Offenders
VBVQ total score^a	0.30	0.28	52	0.00–0.90	.82
Violent Behaviour Scale	3.84	2.39	52	0.00–8.25	.90
Physical Aggression (PA-AQ)	2.82	0.94	52	1.11–4.78	.83
Violent convictions	4.84	5.58	45	0–19	–
SAQ	31.54	11.70	50	6–54	.92
SIR-R1 Scale	1.70	8.80	47	−18 to 23	–^c

Note. AQ = Aggression Questionnaire; SAQ = Self-Appraisal Questionnaire; SIR-R1 = Statistical Information on Recidivism – Revised 1; VBVQ = Violent Behavior Vignette Questionnaire.

^a1 point was given for each vignette if a violent response was selected for either round of the vignette, and these points were averaged across vignettes such that total scores can range from 0 to 1; in Study 1 total scores were computed only for participants who received at least 3 vignettes because participants were presented with only a subset of the vignettes.

^bCronbach’s alpha was not computed for vignette scores in Study 1 because participants were presented with different subsets of vignettes.

^cCronbach’s alpha was not computed for SIR-R1 Scale because we did not have item-level data.

Table 3.

Pearson correlations [and 95% confidence intervals] between vignette scale total score^a and violence variables.

	Study 1		Study 2	Study 3	Study 4
	Students	Community	Offenders	Community	Offenders
Violent Behaviour Scale	.45* [.31, .58]	.42* [.30, .54]	.83* [.70, .92]	.47* [.38, .55]	.51* [.28, .71]
Physical Aggression (PA-AQ)	.52* [.42, .62]	.53* [.44, .62]	.73* [.39, .93]	.58* [.51, .64]	.74* [.59, .84]
Violent convictions	–	–	−.04 [−.60, .44]	–	−.02 [−.33, .32]
SAQ	–	–	.78* [.52, .95]	–	.60* [.42, .75]
SIR-R1 Scale	–	–	−.48 [−.88, .26]	–	−.20 [−.46, .04]

Note. AQ = Aggression Questionnaire; SAQ = Self-Appraisal Questionnaire; SIR-R1 = Statistical Information on Recidivism – Revised 1 (higher SIR-R1 scores indicate lower risk of recidivism). 95% confidence intervals were computed with bootstrap method (1000 samples). Spearman rho correlations yielded virtually identical values as the Pearson correlations reported here.

^a1 point was given for each vignette if a violent response was selected for either round of that vignette, and these points were averaged across vignettes such that total scores can range from 0 to 1; in Study 1 total scores were computed only for participants who received at least 3 vignettes because participants were presented with only a subset of the vignettes.

*p < .05.

Study 2

In Study 2 we followed the general template of the preceding study, but with a sample of adult male offenders incarcerated in a maximum-security prison. Participants were presented the violence vignette measure and the other self-report measures of violent behavior and physical aggressiveness used in the preceding study, as well as a validated self-report measure of risk of re-offending. We also gathered information from their official files about their violent criminal history and actuarial risk for re-offending. As in the preceding study, the purpose of this study was to explore whether our vignette measure is associated with other measures of violent behavior and physical aggressiveness.

Method

Participants

Participants included in the main analyses were 16 adult male offenders recruited from a maximum-security federal penitentiary in Ontario, Canada. Eligibility criteria were male gender; 18 years old or older; able to speak and read English; and at least one current (index) non-sexual violent offense against a man or non-violent offense. All participants reported they could understand written English and correctly answered all of the quality control items. Participants’ median age was 24–29 years old and most were single (56.3%). Approximately one third (37.5%) of participants were First Nation, 31.3% were White, 18.8% were Black, 6.3% were Inuit, and 6.3% were Metis. Median education level was grade 12. Mean aggregate sentence was 1551.44 days (SD = 1727.39). Mean number of criminal offense convictions was 32.56 (SD = 17.69).

Measures

We used the same measures as in Study 1, except that that we only presented the 10 vignettes that had shown an 11% or higher violent response baserate in the community sample (the same 10 vignettes we retained for analyses in Study 1), we revised the vignette response options to make them identical across vignettes, and we included other indicators of past and propensity for violence. Note that our decision to drop the five lower baserate vignettes was motivated by psychometric concerns (i.e., baserate) as well as practical considerations; 15 scenarios would have made the procedure too lengthy, and presenting each participant with a random subset of the scenarios (as we did in Study 1) would have diluted what we anticipated would already be a small sample of offenders.

We presented all 10 vignettes in a random order. There was a total of 44 other responses, ranging from zero to five per vignette response opportunity. As in Study 1, we coded the other responses into one of the following four categories: non-violent, threat of violence, use of physical violence, or other (none of the above). Inter-rater reliability for coding other responses was perfect (100% agreement) for 19 of the vignette response opportunities, but low in the remaining one vignette response opportunity (33.3% agreement; due to an oversight one coder did not code two of the three responses).

As in Study 1, we classified responses to each vignette as either violent (threaten, shove, hit, or other threat or use of violence on either round) or non-violent (ignore, talk, insult, or other non-violent on both rounds), and computed a total score by averaging the number of vignettes with a violent response.

Self-Appraisal Questionnaire (SAQ)

The SAQ (Loza, 2005) is a self-report risk assessment instrument that has been found to predict violent recidivism equally or more accurately than the most widely used validated clinician-administered violence risk assessment instruments (average r = .37, k = 8; Campbell et al., 2009). Items are summed to compute a total score, and higher scores indicate higher risk. The mean total score for the current sample was 36.19 (Table 2), which was higher (86^th percentile) than the mean total score for the normative sample of 938 Canadian male offenders (M = 23.10, SD = 10.88) reported by Loza (2005). Total scores can be classified as low risk (0–10), low-moderate risk (11–22), high-moderate risk (23–42), and high risk (43–67). The current sample’s mean score falls in the high-moderate risk category, which corresponds to a 21% probability of violent recidivism within 5 years (Loza, 2005).

Statistical Information on Recidivism - Revised 1 (SIR-R1) Scale

The SIR-R1 Scale is an actuarial risk assessment instrument routinely used with non-Aboriginal men by the Correctional Service of Canada (CSC; see Barnum & Gobeil, 2012). Evaluators complete the SIR-R1 from official file information. Items reflect criminal and correctional history as well as age and marital status. Though designed to assess risk for general criminal recidivism, it is also a good predictor of violent recidivism (average r = –.22, k = 17, Campbell et al., 2009; r = –.36, N = 12845, Barnum & Gobeil, 2012). Items are summed to compute a total score, and higher scores indicate lower risk. Total scores can be classified as very good (6 to 27), good (1 to 5), fair (–4 to 0), fair/poor (–8 to –5), and poor (–30 to –9). The current sample’s mean score falls in the fair/poor risk category; 7% of male non-Aboriginal offenders (N = 12845) in the fair/poor risk category violently recidivated within 3 years (Barnum & Gobeil, 2012).

Procedure

This study was approved by our university’s research ethics board. Measures were presented in a counterbalanced order via an Eprime-2 programmed laptop computer, except that the paper-and-pencil SAQ was always completed after the measures on the computer. SIR-R1 Scale scores were obtained from official records. We coded the total number of non-sexual violent offense convictions (e.g., uttering threats, assault, robbery, murder) on participants’ official criminal records.

Results and discussion

The baserate of violent responses (on either the first or second round) for each of the 10 vignettes for the offender sample is presented in Table 1. Baserates for violent responses ranged from a minimum of 27% to a maximum of 53%. As in Study 1, this suggests a reasonable range of level of provocation and no floor or ceiling effects.

Violent responses to the vignettes were strongly associated with more self-reported past violent behavior and physical aggressiveness. As shown in Figure 3(a) (and Table S5 in the supplemental material), participants who selected a violent response to the vignettes consistently reported more past violent behavior than participants who selected a non-violent response; effect sizes were all very large (Cohen’s d ranged from 1.15 to 2.65). As shown in Figure 3(b) (and Table S6 in the supplemental material), participants who selected a violent response to the vignettes consistently reported more physical aggressiveness than participants who selected a non-violent response; effect sizes were all very large (Cohen’s d ranged from 1.04 to 2.13). Consistent with these results for the individual vignettes, those who responded with any violence across the vignettes reported much more past violent behavior (Cohen’s d = 2.43) and physical aggressiveness (Cohen’s d = 1.94) than those who never responded with violence.

Figure 3.

Differences between offender participants in Study 2 who selected a violent response on the vignettes versus those who did not select a violent response. (a) Past violent behavior (N = 16). (b) Aggressiveness (N = 16). (c) Number of violent convictions (N = 15 to 16). (d) Estimated risk of violent reoffending (N = 15 to 16). (e) SIR-R1 scores (higher SIR-R1 scores indicate lower risk of recidivism; N = 9).

Violent responses to the vignettes were not, however, clearly associated with official criminal records of violent offenses. As shown in Figure 3(c) (and Table S7 in the supplemental material), participants who selected a violent response to the vignettes did not consistently have more convictions for violent offenses than participants who selected a non-violent response; effect sizes were mostly small and varied in direction (Cohen’s d ranged from -0.58 to 0.14). Those who responded with any violence across the vignettes had slightly more convictions for violent offenses (Cohen’s d = 0.21) than those who never responded with violence.

In contrast, violent responses to the vignettes were more clearly associated with risk of violent re-offending. As shown in Figure 3(d) (and Table S8 in the supplemental material), participants who selected a violent response to the vignettes consistently had higher scores on the SAQ than participants who selected a non-violent response; effect sizes were all very large (Cohen’s d ranged from 1.06 to 2.78). As shown in Figure 3(e) (and Table S9 in the supplemental material), participants who selected a violent response to the vignettes consistently had lower scores on the SIR-R1 Scale (reflecting higher risk) than participants who selected a non-violent response; effect sizes were medium to large (Cohen’s d ranged from –0.39 to –2.00). Those who responded with any violence across the vignettes were also much higher risk on the SAQ (Cohen’s d = 1.77) and SIR-R1 Scale (Cohen’s d = –1.16) than those who never responded with violence.

As in Study 1, we examined the correlation between the vignette scale total score and the indicators of violent behavior (descriptive and internal consistency statistics are presented in Table 2). As shown in Table 3, the vignette total score was strongly associated with more self-reported violence (VBS), more physical aggressiveness (PA-AQ), and higher risk of violent recidivism (SAQ and SIR-R1 Scale); however, it showed a small association with fewer convictions for violent offenses.

A potential limitation of this and the preceding study was that the response options were ordered in an obvious progression from least to most aggressive. This may have exaggerated or otherwise distorted the results because the response option order may have made too salient our assumptions about their rank on a continuum of non-violence to violence. The offender sample size was also clearly lacking, but the results were very consistent with the larger student and community participant samples. Nevertheless, it would be important to attempt to conceptually replicate these findings with a refined version of the vignette measure and larger samples.

Study 3

The results of Studies 1 and 2 confirmed that our vignette approach generally corresponded to indicators of violent behavior. We next refined the vignettes and response options and tested them with a larger sample of community participants. Specifically, we added details to the scripts for both the initial scenario and what would follow participants’ selection of each response option. We also added more response options and put them in a fixed random order. Finally, we asked participants about their sexual orientation so that we could include only heterosexual men in the analyses; we expected that some of the vignettes may not be as relevant or provocative for gay men (e.g., another man expressing romantic interest in your girlfriend or wife).

Method

Participants

Participants included in the main analyses were 471 adult male community participants in Canada and the United States recruited from a Qualtrics panel. This sample excluded a total of 200 additional participants who consented but did not report they were male (n = 43), did not report they could understand written English (n = 53), did not correctly answer all of the quality control items (n = 150), did not report heterosexual orientation (n = 101), did not answer more than 20% of the items on the VBS (n = 103) or PA-AQ (n = 103), or were missing data on the vignettes (n = 2)—these exclusion categories were not mutually exclusive. Median age was 35 years old (range 18–81 years old) and the most common marital status was married (46.3%, n = 218). Average time to complete the survey was 12.92 minutes (SD = 15.96).

Measures

The measures were the same as in the preceding studies, with the exception that the vignette measure was refined as detailed below.

Violent Behavior Vignette Questionnaire (VBVQ). The VBVQ is our refined version of the vignette measure. Like the earlier version, it consists of 10 vignettes and there are two rounds of responses for each vignette. The vignettes reflected the same general themes described above in Study 1 (the VBVQ is presented in full in Appendix S2 in the supplemental material). After each round, participants were asked “What do you do?” and presented with the following nine response options in the following order:

Report it to someone – report what he’s doing (or what he did) to someone like the manager, police, or security, and ask them to deal with the problem.

Talk it out – without insulting him, say something to let him know you’re not okay with what he’s doing (or what he did) and to try to work things out peacefully.

Insult him – say something insulting to him, like swearing at him or calling him a name.

Joke about it – joke about it to yourself or other people around you.

Shove him.

Hit, punch, kick, or tackle him.

Threaten to hurt him.

Ignore it – don’t do or say anything about it.

Leave – go somewhere else to get away from the person or situation.

We presented all 10 vignettes in a random order via an online survey. As in the previous studies, we classified responses to each vignette as either violent (threaten to hurt him; shove him; or hit, punch, kick, or tackle him selected on either round) or non-violent (ignore it, leave, report it, talk it out, joke about it, or insult him selected on both rounds). Total score is the average number of vignettes with a violent response, and can range from 0 to 1.

Procedure

This study was approved by our university’s research ethics board. The order in which the measures were completed was counterbalanced across participants.

Results and discussion

The baserate of violent responses (on either the first or second round) for each of the 10 VBVQ vignettes is presented in Table 1. Baserates for violent responses ranged from a minimum of 15% to a maximum of 51%. As in Studies 1 and 2, this suggests a reasonable range of level of provocation and no floor or ceiling effects.

Violent responses to the vignettes were moderately to strongly associated with more self-reported past violent behavior and physical aggressiveness. As shown in Figure 4(a) (and Table S10 in the supplemental material), participants who selected a violent response to the vignettes consistently reported more past violent behavior than participants who selected a non-violent response; effect sizes were medium to large (Cohen’s d ranged from 0.58 to 0.98). As shown in Figure 4(b) (and Table S11 in the supplemental material), participants who selected a violent response to the vignettes consistently reported more physical aggressiveness than participants who selected a non-violent response; effect sizes were large (Cohen’s d ranged from 0.73 to 0.94). Consistent with these results for the individual vignettes, those who responded with any violence across the vignettes reported much more past violent behavior (Cohen’s d = 0.80) and physical aggressiveness (Cohen’s d = 1.15) than those who never responded with violence.

Figure 4.

Differences between community participants in Study 3 who selected a violent response on the vignettes versus those who did not select a violent response. (a) Past violent behavior (N = 462 to 471). (b) Aggressiveness (N = 462 to 471).

Study 4

The results of Study 3 showed generally strong correspondence between the VBVQ and indicators of violent behavior in a large sample of men from the community. In Study 4, we examined the degree of correspondence in a sample of adult male offenders incarcerated in medium- and maximum-security prisons.

Method

Participants

Participants included in the main analyses were 52 adult male offenders recruited from medium- and maximum-security federal penitentiaries in Ontario, Canada. Eligibility criteria were male gender; 18 years old or older; able to speak and read English; and at least one current (index) non-sexual violent offense against a man or non-violent offense. This sample excluded three additional participants who incorrectly answered any of the quality control items. Mean age was 31.04 years old (SD = 8.59). Approximately one third (34.6%) of participants were White, 34.6% were Black, 7.7% were First Nation, 5.8% were Caribbean, 5.8% were Asian, 3.8% were Metis, 3.8% were Eastern European, 1.9% were Arab, and 1.9% were multi-racial/ethnic. Median education level was grade 10. Mean aggregate sentence was 1276.40 days (SD = 841.20). Most participants had at least one violent offense conviction on record (82.22%). The mean SAQ total score for the current sample was higher (82.4th percentile) than the mean total score for the normative sample of Canadian male offenders (Loza, 2005). The current sample’s SAQ mean score fell in the high-moderate risk category, which corresponds to a 21% probability of violent recidivism within 5 years (Loza, 2005). The current sample’s mean SIR-R1 score falls in the good risk category; 4% of male non-Aboriginal offenders (N = 12845) in the good risk category violently recidivated within 3 years (Barnum & Gobeil, 2012).

Measures and procedure

This study was approved by our university’s research ethics board. The procedure and measures were the same as in Study 3, except that that we also included the other indicators of past and propensity for violence described in Study 2 and the measures were presented via Eprime2 on a laptop computer.

Results and discussion

The baserate of violent responses (on either the first or second round) for each of the 10 VBVQ vignettes is presented in Table 1. Baserates for violent responses ranged from a minimum of 12% to a maximum of 44%. As in the preceding studies, this suggests a reasonable range of level of provocation and no floor or ceiling effects.

Violent responses to the vignettes were consistently associated with more self-reported past violent behavior and physical aggressiveness. As shown in Figure 5(a) (and Table S12 in the supplemental material), participants who selected a violent response to the vignettes consistently reported more past violent behavior than participants who selected a non-violent response; effect sizes were small to large (Cohen’s d ranged from 0.21 to 1.13). As shown in Figure 5(b) (and Table S13 in the supplemental material), participants who selected a violent response to the vignettes consistently reported more physical aggressiveness than participants who selected a non-violent response; effect sizes were medium to large (Cohen’s d ranged from 0.56 to 1.60). Consistent with these results for the individual vignettes, those who responded with any violence across the vignettes reported much more past violent behavior (Cohen’s d = 0.93) and physical aggressiveness (Cohen’s d = 1.75) than those who never responded with violence.

Figure 5.

Differences between offender participants in Study 4 who selected a violent response on the vignettes versus those who did not select a violent response. (a) Past violent behavior (N = 52). (b) Aggressiveness (N = 52). (c) Number of violent convictions (N = 45). (d) Estimated risk of violent reoffending (N = 50). (e) SIR-R1 scores (higher SIR-R1 scores indicate lower risk of recidivism; N = 47).

Violent responses to the vignettes were not, however, clearly associated with official criminal records of violent offenses. As shown in Figure 5(c) (and Table S14 in the supplemental material), participants who selected a violent response to the vignettes did not consistently have more convictions for violent offenses than participants who selected a non-violent response; effect sizes were mostly small and varied in direction (Cohen’s d ranged from -0.37 to 0.29). Those who responded with any violence across the vignettes had slightly fewer convictions for violent offenses (Cohen’s d = -0.27) than those who never responded with violence.

Violent responses to the vignettes were more clearly associated with risk of violent re-offending. As shown in Figure 5(d) (and Table S15 in the supplemental material), participants who selected a violent response to the vignettes consistently had higher scores on the SAQ than participants who selected a non-violent response; effect sizes were small to large (Cohen’s d ranged from 0.36 to 1.54). As shown in Figure 5(e) (and Table S16 in the supplemental material), participants who selected a violent response to the vignettes mostly had lower scores on the SIR-R1 (reflecting higher risk) than participants who selected a non-violent response; effect sizes were small to medium (Cohen’s d values in the expected direction ranged from -0.12 to -0.44; the two Cohen’s d values in the opposite direction were small: 0.06 and 0.19). Those who responded with any violence across the vignettes were also much higher risk on the SAQ (Cohen’s d = 1.32) and SIR-R1 (Cohen’s d = -0.85) than those who never responded with violence.

As in the preceding studies, we examined the correlation between the VBVQ total score and the indicators of violent behavior (descriptive and internal consistency statistics are presented in Table 2). As shown in Table 3, the VBVQ total score was strongly associated with more self-reported violence (VBS) and more physical aggressiveness (PA-AQ). It was also strongly associated with higher risk of violent recidivism as measured by the SAQ, but only slightly associated with higher risk of violent recidivism as measured by the SIR-R1 Scale and uncorrelated with number of convictions for violent offenses.

We also compared the VBVQ total score of the offenders in the current study to the non-violent community participants in Study 3 (VBS score = 0). The offenders had moderately higher VBVQ total scores (M = 0.30, SD = 0.28, N = 52) than did the non-violent community sample (M = 0.16, SD = 0.23, N = 125; d = 0.57, 95% CI [0.24, 0.90]). Though caution is warranted given that there were a number of other variables confounded with sample, such as setting (prison vs. community), presentation (Eprime 2 programmed computer vs. Qualtrics online survey), conditions (research-assistant administered and supervised completion of procedure vs. online survey), and degree of privacy (no anonymity and access to official records vs. anonymity of responses), these results are supportive of the construct validity of VBVQ scores. Future research should compare violent offenders, non-violent offenders, and non-offenders on the VBVQ using more similar procedures for all groups.

General discussion

The goal of the current article was to develop a self-report measure designed to assess current violent behavior—the Violent Behavior Vignette Questionnaire (VBVQ)—and conduct an initial test of its construct validity with samples of men from a university, the community, and prisons. In Studies 1 and 2, violent responses on the preliminary version of the VBVQ generally corresponded to independent indicators of physical aggressiveness and violent behavior among male university students, men in the community, and incarcerated male offenders. In Studies 3 and 4, we refined the VBVQ and again tested the validity of its scores in new samples of men in the community and incarcerated male offenders. In all samples, men who selected a violent response option on the VBVQ generally had much higher levels of physical aggressiveness and violent behavior than did men who selected non-violent response options. In addition, offenders (Study 4) responded more violently on the VBVQ than did the non-violent community participants (Study 3). Our findings are consistent with past research with questionnaires and analogs of aggressive behavior (Buss, 1961; Cherek, 1992; DeWall et al., 2013; Lieberman et al., 1999; Liebowitz, 1968; O’Connor et al., 2001; Taylor, 1967; Zeichner et al., 1999), and build on these findings to provide a measure of violence that should better satisfy the standards, sensibilities, foci, priorities, and restrictions of both forensic and non-forensic researchers studying violent behavior.

There were some limitations that may affect the conclusions that can be drawn from our findings. Though the vignette responses corresponded well with most of the indicators of violent behavior, there was little correspondence with violent offense records among the offender samples. This is not supportive of the construct validity of VBVQ scores and compares negatively with most of the few studies that have tested the association between other self-report and analog measures of aggression and official records of violent offending (e.g., Cherek et al., 1997; Diamond & Magaletta, 2006; Lim et al., 2011; Olver et al., 2014; Pettersen et al., 2018; but see T. Y. Williams et al., 1996). Future research should replicate and extend our tests with more thorough coding of official records of violent offending. For example, perhaps different results would be obtained if we could have distinguished between prior violent offenses and current (or index) violent offenses (i.e., the offense or set of offenses for which the person is serving his current prison sentence; Bonta et al., 1998; Harris et al., 1993) or between violence against another man and other types of violence (e.g., against women). In our coding, we did not have access to information beyond the names of the violent offense conviction listed in the official criminal record (e.g., assault, murder); thus, some of the violent convictions we tallied may not have included violence against other men. A more informative test of validity would be to examine the association between the VBVQ and convictions for violent offenses against men. In addition, the small samples in Studies 2 and 4 reduce confidence in the stability of their results; future research with larger samples of offenders is needed.

Future research testing the construct validity of VBVQ responses should build on and address the limitations of the current studies. The most direct extension of the current studies would be to examine how well the VBVQ correlates with additional indicators of criminal violent behavior, such as actual violent recidivism or non-self-report risk assessment instruments designed specifically to assess risk for violent recidivism, such as the Revised Violence Risk Appraisal Guide (Rice et al., 2013). Another strong test would be to examine the extent to which one’s responses on the VBVQ would correspond to peer-reports on the VBVQ specifically (e.g., which response would your friend choose for each vignette?) or violent behavior more generally (e.g., ratings of the frequency of your friend’s violent behavior in the past, and his current likelihood of violent behavior); this peer-assessment approach has been used to test the validity of scores on the Aggression Questionnaire (Buss & Perry, 1992; O’Connor et al., 2001). A thorough test of construct validity would also require examining the extent to which VBVQ scores diverge from measures of presumably distinct constructs (i.e., divergent validity).

Another informative approach would be to examine the extent to which the VBVQ shows the expected relationship with variables known to correlate with or affect criminal violent behavior, such as low self-control, psychopathy, or participation in an effective violence reduction treatment program (Bonta et al., 1998; Harris et al., 1993; Henry et al., 1996; Longshore & Turner, 1998). It would also be important to test our presumption that the VBVQ assesses current violent behavior by examining the extent to which scores can change (or be changed) and whether such changes correspond to changing propensity for criminal violent behavior. Somewhat relatedly, future research should also examine the test-retest reliability of VBVQ scores. Finally, it would be important to determine the extent to which the VBVQ is a better—or, at least, a non-redundant—measure of current violent behavior than other measures researchers have typically used for this purpose, such as the Physical Aggression scale of the Aggression Questionnaire, and other vignette measures (e.g., Coccaro et al., 2017; Hilton et al., 2003; O’Connor et al., 2001). One way to test this would be to include such measures along with the VBVQ in the validation studies we have suggested above and test whether stronger or independent associations are observed with the VBVQ relative to these other measures.

Incorporating more sophisticated technology may also be a worthwhile direction for future research (Abbey & Wegner, 2015; Hilton et al., 2003; Paschall et al., 2005; Udell, 2009). Though we believe incorporating audio, visual, or even virtual reality technology could increase construct validity, there are also challenges and potential drawbacks. In contrast, our current self-report VBVQ is relatively simple and versatile, allowing for easier and broader implementation. For example, researchers can simply download our script and run the procedure with a basic computer online or offline with no need to worry about additional settings (e.g., volume) or equipment (e.g., virtual reality headset, powerful computer). In addition, the simple written form may actually make the scenarios more relatable for participants because they may be able to fill in the blanks with their own personally relevant visual and audio features of antagonists and settings more so than if they were presented with our preconceived visual and audio depictions.

Another important next step would be to examine the extent and limits regarding the violence assessed by the VBVQ, and work towards developing similar measures for any unaddressed important types of violence. For example, we think our vignettes probably assess hostile or reactive violence more so than instrumental violence because they all present provocative situations (e.g., hostile, aggressive, and disrespectful antagonists; Cornell et al., 1996; Feshbach, 1964; Stanford et al., 2003). We initially tried to include some more instrumental violence vignettes in the earlier version, but we dropped them because their baserates of violent responses were too low; this is consistent with evidence that hostile violence is much more common than instrumental violence (Stanford et al., 2003). Thus, we think the VBVQ assesses typical violent behavior, but certainly not all violent behavior. It would be useful to explore the extent to which distinct motives for violence are reflected in responses to the VBVQ, and, if need be, develop additional vignette measures.

Relatedly, it would be important to test the degree of consistency in the interpretation and meaning of the VBVQ between different populations and settings. Though the VBVQ is intended for use with both general community and correctional/forensic populations and settings, a number of differences between these populations and settings could affect how participants process and respond to the VBVQ. For example, lower levels of reading and formal education typical of correctional/forensic samples relative to general community samples could conceivably reduce comprehension of the written material and thereby attenuate the provocativeness of the conflict scenarios, which could underestimate violent behavior in correctional/forensic samples. In addition, given the focus of the VBVQ on current violent behavior (“… say what you would really do if you were in that situation right now…”), scores may be more sensitive and limited to the setting in which the VBVQ is completed compared to many other measures of violent behavior. For example, the same person may respond more—or less—violently on the VBVQ while he is in prison than while he is in the community because of the different contingencies in each setting. Any such sensitivity to real-time changes in current likelihood of violence would be by design and a strength of the VBVQ, but would potentially limit the interpretation of a given person’s scores to the particular time and context in which he completed the VBVQ. Whatever effect population and setting may or may not have on VBVQ scores, there were no floor or ceiling effects observed for either the community or correctional samples in the current studies, and the VBVQ was associated with the other self-report measures of violent and aggressive behavior in both samples. Future research should test for measurement invariance of VBVQ scores between different populations and contexts.

It is noteworthy that in the current studies we did not examine social desirability. Although socially desirable responding is a common concern for self-report measures, we would not expect typical social desirability scales to be useful for testing the extent to which the VBVQ may be susceptible to biased responding. Mills and colleagues have found evidence that social desirability scales like the Balanced Inventory of Desirable Responding (BIDR; Paulhus, 1994) may inadvertently assess general antisociality in addition to, or instead of, deceptive responding. More specifically, greater “social desirability” scores have been associated with lower risk and lower rates of recidivism (Hanson & Wallace-Capretta, 2004; Mills & Kroner, 2005, 2006; Mills et al., 2003). In light of these findings, we did not think the available measures of socially desirable responding would be useful in understanding the validity of the VBVQ.

As noted in the Introduction section, we focused exclusively on male participants and violence perpetrated by men against men because men are responsible for the majority of criminal violence and because we wanted to maximize the precision and efficiency of our studies. Future research should develop similar analog measures for other populations (e.g., women) and other types of violence. Some work on this has already begun in areas such as domestic violence (Russa & Rodriguez, 2010) and sexual aggression (Abbey & Wegner, 2015; Hall & Hirschman, 1994; Hermann & Nunes, 2018). Such measures have the potential to facilitate more rapid advancement of scientific knowledge in areas that have hitherto largely been hamstrung by the difficulties of doing rigorous research on low base rate and often officially undetected violence.

For the most part our initial findings are encouraging. They support further investigation of the VBVQ as a potentially useful measure of current violent behavior that may facilitate greater symmetry and consilience between the forensic and non-forensic literatures, and allow research on the causes and reduction of violence to advance more quickly.

Supplemental Material

sj-pdf-1-prx-10.1177_0033294120939308 - Supplemental material for The Violent Behavior Vignette Questionnaire (VBVQ): A Measure of Violent Behavior for Research in Forensic and Non-Forensic Settings and Populations

Supplemental material, sj-pdf-1-prx-10.1177_0033294120939308 for The Violent Behavior Vignette Questionnaire (VBVQ): A Measure of Violent Behavior for Research in Forensic and Non-Forensic Settings and Populations by Kevin L. Nunes, Chantal A. Hermann, Sacha Maimone, Maya Atlas and Brian A. Grant in Psychological Reports

Footnotes

Acknowledgments

We are grateful to the many people at the Correctional Service of Canada who facilitated our studies in the prisons and access to data: Colette Cousineau, Jenn Denny, Lana Di Fazio, Justin Gileno, Janet Graham, Devon Gunn, Sara Johnson, Peter Marquis, Erin McCormick, Andrea Moser, Ali Phillips, Penny Scott, Terri Scott, Geris Serran, John Weekes, and many others. We also would like to acknowledge the valuable contributions made by many students who assisted with interviewing, transcribing, coding, and data entry, as well as searching for and summarizing relevant information from existing measures and treatment programs: Stephanie Biro, Carolyn Blank, Nick Chadwick, Erin DeJong, Becky Grace, Lindsay Grenon, Chloe Pedneault, Tori Semple, Emily Start, Mandie Woods, and many others. We would also like to thank the people who gave feedback on the content of the items that ultimately formed the VBVQ: Liam Ennis, Ian McPhail, Edward Nunes, James Nunes, and Cathrine Pettersen.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by a Partnership Development Grant (890–2011-0035) from the Social Sciences and Humanities Research Council of Canada, the Research Branch of the Correctional Service of Canada, and Carleton University.

ORCID iDs

Kevin L. Nunes

Sacha Maimone

Supplemental material

Supplemental material for this article is available online.

Author Biographies

Kevin L. Nunes is a professor and director of the Aggressive Cognitions and Behaviour Research Lab in the Department of Psychology at Carleton University. The main focus of his research is on the conceptualization and measurement of cognitions (e.g., attitudes) thought to be relevant to violent behavior, and the role these cognitions may play in violent behavior.

Chantal A. Hermann is a Program Effectiveness Analyst in the Program Evaluation Unit (PEU) for the Ontario Ministry of the Solicitor General. Hermann completed her undergraduate degree in Forensic Science (specialization Psychology) at Laurentian University and her masters and doctorate in psychology at Carleton University. Hermann’s areas of research primarily focus on the causes of sexual and violent offending.

Sacha Maimone is a PhD student in Psychology at Carleton University. The main focus of her research is on examining the conceptualization and measurement of cognitions, particularly attitudes, thought to be relevant for violent behavior and the role these cognitions may play in violence and aggression.

Maya Atlas is a graduate student in clinical psychology at the Psychosocial Medicine Lab at Ryerson University. Her research focuses on psychosocial oncology.

Brian A. Grant is an adjunct professor in the Department of Psychology at Carleton University.

References

Abbey

Wegner

(2015). Using experimental paradigms to examine alcohol’s role in men’s sexual aggression: Opportunities and challenges in proxy development. Violence Against Women, 21(8), 975–996. https://doi.org/10.1177/1077801215589378

Anderson

C. A.

Bushman

B. J.

(1997). External validity of “trivial” experiments: The case of laboratory aggression. Review of General Psychology, 1(1), 19–41.

Anderson

C. A.

Lindsay

J. J.

Bushman

B. J.

(1999). Research in the psychological laboratory: Truth or triviality?. Current Directions in Psychological Science, 8(1), 3–9. https://doi.org/10.1111/1467-8721.00002

Barnum

Gobeil

(2012). Prediction of re-offence using the SIR-R1 and a proxy (Research Report No. R-281). Research Branch, Correctional Service of Canada.

Bonta

Law

Hanson

R. K.

(1998). The prediction of criminal and violent recidivism among mentally disordered offenders: A meta-analysis. Psychological Bulletin, 123(2), 123–142.

Bryant

F. B.

Smith

B. D.

(2001). Refining the architecture of aggression: A measurement model for the Buss-Perry Questionnaire. Journal of Research in Personality, 35(2), 138–167. https://doi.org/10.1006/jrpe.2000.2302

Buss

A. H.

(1961). The psychology of aggression. Wiley.

Buss

A. H.

Perry

(1992). The aggression questionnaire. Journal of Personality and Social Psychology, 63(3), 452–459.

Campbell

M. A.

French

Gendreau

(2009). The prediction of violence in adult offenders: A meta-analytic comparison of instruments and methods of assessment. Criminal Justice and Behavior, 36(6), 567–590. https://doi.org/10.1177/0093854809333610

10.

Cherek

D. R.

(1992). Manual for point subtraction aggression paradigm. A computer program to measure aggressive responding in human subjects under controlled laboratory conditions. University of Texas Health Science Center.

11.

Cherek

D. R.

Moeller

F. G.

Schnapp

Dougherty

D. M.

(1997). Studies of violent and nonviolent male parolees: I. Laboratory and psychometric measurements of aggression. Biological Psychiatry, 41(5), 514–522. https://doi.org/10.1016/S0006-3223(96)00059-5

12.

Coccaro

E. F.

Fanning

Lee

(2017). Development of a Social Emotional Information Processing assessment for adults (SEIP-Q). Aggressive Behavior, 43(1), 47–59. https://doi.org/10.1002/ab.21661

13.

Cohen

(1988). A power primer. Quantitative Methods in Psychology, 112, 155–159.

14.

Cornell

D. G.

Warren

Hawk

Stafford

Oram

Pine

(1996). Psychopathy in instrumental and reactive violent offenders. Journal of Consulting and Clinical Psychology, 64(4), 783–790. https://doi.org/10.1037/0022-006X.64.4.783

15.

Correctional Service of Canada (2007). Violence prevention program [Unpublished program materials].

16.

DeWall

C. N.

Finkel

E. J.

Lambert

N. M.

Slotter

E. B.

Bodenhausen

G. V.

Pond

R. S.

Jr. Renzetti

C. M.

Fincham

F. D.

(2013). The Voodoo Doll Task: Introducing and validating a novel method for studying aggressive inclinations. Aggressive Behavior, 39(6), 419–439. https://doi.org/10.1002/ab.21496

17.

Diamond

P. M.

Magaletta

P. R.

(2006). The Short-Form Buss-Perry Aggression Questionnaire (BPAQ-SF): A validation study with federal offenders. Assessment, 13(3), 227–240. https://doi.org/10.1177/1073191106287666

18.

Dixon

Browne

(2003). The heterogeneity of spouse abuse: A review. Aggression and Violent Behavior, 8(1), 107–130. https://doi.org/10.1016/S1359-1789(02)00104-0

19.

Ferguson

C. J.

Rueda

S. M.

(2009). Examining the validity of the modified Taylor Competitive Reaction Time Test of Aggression. Journal of Experimental Criminology, 5(2), 121–137. https://doi.org/10.1007/s11292-009-9069-5

20.

Feshbach

(1964). The function of aggression and the regulation of aggressive drive. Psychological Review, 71, 257–272. https://doi.org/10.1037/h0043041

21.

Gannon

Mihorean

(2005). Criminal victimization in Canada, 2004 (Catologue No. 85-002-XPE). Juristat, 25(7), 1–27.

22.

Giancola

P. R.

Parrott

D. J.

(2008). Further evidence for the validity of the Taylor Aggression Paradigm. Aggressive Behavior, 34(2), 214–229. https://doi.org/10.1002/ab.20235

23.

Giancola

P. R.

Zeichner

(1995). Construct validity of a competitive reaction-time aggression paradigm. Aggressive Behavior, 21(3), 199–204.

24.

Golomb

B. A.

Cortez-Perez

Jaworski

B. A.

Mednick

Dimsdale

(2007). Point Subtraction Aggression Paradigm: Validity of a brief schedule of use. Violence and Victims, 22(1), 95–103. https://doi.org/10.1891/vv-v22i1a006

25.

Hall

G. C. N.

Hirschman

(1994). The relationship between men’s sexual aggression inside and outside the laboratory. Journal of Consulting and Clinical Psychology, 62(2), 375–380. https://doi.org/10.1037/0022-006X.62.2.375

26.

Hanson

R. K.

Wallace-Capretta

(2004). Predictors of criminal recidivism among male batterers. Psychology, Crime, and Law, 10(4), 413–427. https://doi.org/10.1080/10683160310001 629283

27.

Harris

G. T.

Rice

M. E.

Quinsey

V. L.

(1993). Violent recidivism of mentally disordered offenders: The development of a statistical prediction instrument. Criminal Justice and Behavior, 20(4), 315–335.

28.

Harris

G. T.

Rice

M. E.

Quinsey

V. L.

Cormier

C. A.

(2015). Violent offenders: Appraising and managing risk. (3rd ed.). American Psychological Association.

29.

Henry

Caspi

Moffitt

T. E.

Silva

P. A.

(1996). Temperamental and familial predictors of violent and nonviolent criminal convictions: Age 3 to age 18. Developmental Psychology, 32(4), 614–623.

30.

Hermann

C. A.

Nunes

K. L.

(2018). Implicit and explicit evaluations of sexual aggression predict subsequent sexually aggressive behavior in a sample of community men. Sexual Abuse: A Journal of Research and Treatment, 30(5), 510–532. https://doi.org/10.1177/1079063216682952

31.

Hilton

N. Z.

Harris

G. T.

Rice

M. E.

(2003). Correspondence between self-report measures of interpersonal aggression. Journal of Interpersonal Violence, 18(3), 223–239. https://doi.org/10.1177/0886260502250065

32.

Kennedy

S. M.

(1990). Anger management training with adult prisoners (NN60049) [Doctoral dissertation, University of Ottawa]. ProQuest Dissertations and Theses.

33.

Kroner

D. G.

Mills

J. F.

Morgan

R. D.

(2007). Underreporting of crime-related content and the prediction of criminal recidivism among violent offenders. Psychological Services, 4(2), 85–95. https://doi.org/10.1037/1541-1559.4.2.85

34.

Lieberman

J. D.

Solomon

Greenberg

McGregor

H. A.

(1999). A hot new way to measure aggression: Hot sauce allocation. Aggressive Behavior, 25(5), 331–348.

35.

Liebowitz

(1968). Comparison of self-report and behavioral techniques of assessing aggression. Journal of Consulting and Clinical Psychology, 32(1), 21–25. https://doi.org/10.1037/h0025431

36.

Lim

Day

Casey

(2011). Social cognitive processing in violent male offenders. Psychiatry, Psychology, and Law, 18(2), 177–189. https://doi.org/10.1080/13218711003739490

37.

Longshore

Turner

(1998). Self-control and criminal opportunity: Cross-sectional test of the General Theory of Crime. Criminal Justice and Behavior, 25(1), 81–98. https://doi.org/10.1177/0093854898025001005

38.

Loza

(2005). The Self-Appraisal Questionnare (SAQ): A tool for assessing violent and non-violent recidivism. Multi-Health Systems.

39.

Magnusson

(2020, February 22). Interpreting Cohen’s d effect size: An interactive visualization (Version 2.0.0, beta). https://rpsychologist.com/d3/cohend/

40.

Mills

J. F.

Kroner

D. G.

(2005). An investigation into the relationship between socially desirable responding and offender self-report. Psychological Services, 2(1), 70–80. https://doi.org/10.1037/1541-1559.2.1.70

41.

Mills

J. F.

Kroner

D. G.

(2006). Impression management and self-report among violent offenders. Journal of Interpersonal Violence, 21(2), 178–192. https://doi.org/10.1177/0886260 505282288

42.

Mills

J. F.

Loza

Kroner

D. G.

(2003). Predictive validity despite social desirability: Evidence for the robustness of self-report among offenders. Criminal Behaviour and Mental Health : CBMH, 13(2), 140–150. https://doi.org/10.1002/cbm.536

43.

Murphy

Preston

D. L.

(1999). Reactions to hostile situations survey. In Preston

D. L.

Fazio

R. Di

Serin

R. C.

(Eds.), Persistently violent offender program assessment manual (pp. 59–65). Correctional Service Canada.

44.

New Zealand Department of Corrections. (2007). Violence prevention program [Unpublished treatment manual].

45.

Novaco

R. W.

(1975). Anger control: The development and evaluation of an experimental treatment. Lexington.

46.

Nunes

K. L.

Hermann

C. A.

Maimone

Woods

(2015). Thinking clearly about violent cognitions: Attitudes may be distinct from other cognitions. Journal of Interpersonal Violence, 30(8), 1322–1347. https://doi.org/10.1177/0886260514540329

47.

O’Connor

D. B.

Archer

F. W. C.

(2001). Measuring aggression: Self-reports, partner reports, and responses to provoking scenarios. Aggressive Behavior, 27(2), 79–101. https://doi.org/10.1002/ab.2

48.

Olver

M. E.

Kingston

D. A.

Nicholaichuk

T. P.

Wong

S. C. P.

(2014). A psychometric examination of treatment change in a multisite sample of treated Canadian federal sexual offenders. Law and Human Behavior, 38(6), 544–559. https://doi.org/10.1037/lhb0000086

49.

Paulhus, D. L. (1994). Balanced Inventory of Desirable Responding: Reference manual for the BIDR version 6. Unpublished manuscript, University of British Columbia.

50.

Papalia

Spivak

Daffern

Ogloff

J. R. P.

(2019). A meta‐analytic review of the efficacy of psychological treatments for violent offenders in correctional and forensic mental health settings. Clinical Psychology: Science and Practice, 26(2), e12282. https://doi.org/10.1111/cpsp.12282

51.

Paschall

M. J.

Fishbein

D. H.

Hubal

R. C.

Eldreth

(2005). Psychometric properties of virtual reality vignette performance measures: A novel approach for assessing adolescents’ social competency skills. Health Education Research, 20(1), 61–70. https://doi.org/10.1093/her/cyg103

52.

Perrault

Brennan

(2010). Criminal victimization in Canada, 2009 (Catalogue No. 85-002-X). Juristat, 30(2), 1–33.

53.

Pettersen

Nunes

K. L.

Cortoni

(2018). The factor structure of the Aggression Questionnaire with violent offenders. International Journal of Offender Therapy and Comparative Criminology, 62(7), 1888–1905. https://doi.org/10.1177/0306624X17691242

54.

Polaschek

D. L.

Dixon

B. G.

(2001). The violence prevention project the development and evaluation of a treatment programme for violent offenders. Psychology, Crime, & Law, 7(1), 1–23. https://doi.org/10.1080/1068316010840177.

55.

Quinsey

V. L.

Maguire

Varney

G. W.

(1983). Assertion and over-controlled hostility among mentally disordered murderers. Journal of Consulting and Clinical Psychology, 51(4), 550–556.

56.

Rice

M. E.

Harris

G. T.

Lang

(2013). Validation of a revision to the VRAG and SORAG: The Violence Risk Appraisal Guide—Revised (VRAG-R). Psychological Assessment, 25(3), 951–965. https://doi.org/10.1037/a0032878

57.

Rice

M. E.

Harris

G. T.

Lang

Cormier

(2006). Violent sex offenses: How are they best measured from official records? Law and Human Behavior, 30(4), 525–541. https://doi.org/10.1007/s10979-006-9022-3

58.

Russa

M. B.

Rodriguez

C. M.

(2010). Physical discipline, escalation, and child abuse potential: Psychometric evidence for the analog parenting task. Aggressive Behavior, 36(4), 251–260. https://doi.org/10.1002/ab.20345

59.

Stanford

M. S.

Houston

R. J.

Mathias

C. W.

Villemarette-Pittman

N. R.

Helfritz

L. E.

Conklin

S. M.

(2003). Characterizing aggressive behavior. Assessment, 10(2), 183–190. https://doi.org/10.1177/1073191103010002009

60.

Taylor

(1967). Aggressive behavior and physiological arousal as a function of provocation and the tendency to inhibit aggression. Journal of Personality, 35(2), 297–310.

61.

Theobald

Farrington

D. P.

Loeber

Pardini

D. A.

Piquero

A. R.

(2014). Scaling up from convictions to self-reported offending. Criminal Behaviour and Mental Health: CBMH, 24(4), 265–276. https://doi.org/10.1002/cbm.1928

62.

Thornberry

T. P.

Krohn

M. D.

(2000). The self-report method for measuring delinquency and crime. In Duffee

Crutchfield

R. D.

Mastrofski

Mazerolle

McDowall

Ostrom

(Eds.), Criminal justice 2000: Innovations in measurement and analysis. National Institute of Justice.

63.

Udell

C. J.

(2009). Application of videos and role plays for the assessment of social problem solving in participants of a violence reduction training program (UMI No. 3368323) [Doctoral dissertation, Central Michigan University]. ProQuest Dissertations and Theses.

64.

Williams

J. F.

Meyerson

L. J.

Eron

L. D.

Semler

I. J.

(1967). Peer-rated aggression and aggressive responses elicited in an experimental situation. Child Development, 38(1), 181–190. https://doi.org/10.2307/1127139

65.

Williams

T. Y.

Boyd

J. C.

Cascardi

M. A.

Poythress

(1996). Factor structure and convergent validity of the Aggression Questionnaire in an offender population. Psychological Assessment, 8(4), 398–403.

66.

Zeichner

Frey

F. C.

Parrott

D. J.

Butryn

(1999). Measurement of laboratory aggression: A new response choice paradigm. Psychological Reports, 85(3_suppl), 1229–1237.

67.

Zeichner

Parrott

D. J.

Frey

F. C.

(2003). Gender differences in laboratory aggression under response choice conditions. Aggressive Behavior, 29(2), 95–106. https://doi.org/10.1002/ab.10030

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.36 MB