583713

research-article2015

ASMXXX10.1177/1073191115583713Political Research QuarterlyMoeller et al.

Article

Validation of the Novaco Anger Scale– Provocation Inventory (Danish) With Nonclinical, Clinical, and Offender Samples

Assessment 1­–13 © The Author(s) 2015 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/1073191115583713 asm.sagepub.com

Stine Bjerrum Moeller1, Raymond W. Novaco2, Vivian Heinola-Nielsen3, and Helle Hougaard4

Abstract Anger has high prevalence in clinical and forensic settings, and it is associated with aggressive behavior and ward atmosphere on psychiatric units. Dysregulated anger is a clinical problem in Danish mental health care systems, but no anger assessment instruments have been validated in Danish. Because the Novaco Anger Scale and Provocation Inventory (NAS-PI) has been extensively validated with different clinical populations and lends itself to clinical case formulation, it was selected for translation and evaluation in the present multistudy project. Psychometric properties of the NAS-PI were investigated with samples of 477 nonclinical, 250 clinical, 167 male prisoner, and 64 male forensic participants. Anger prevalence and its relationship with other anger measures, anxiety/depression, and aggression were examined. NAS-PI was found to have high reliability, concurrent validity, and discriminant validity, and its scores discriminated the samples. High scores in the offender group demonstrated the feasibility of obtaining self-report assessments of anger with this population. Retrospective and prospective validity of the NAS were tested with the forensic patient sample regarding physically aggressive behavior in hospital. Regression analyses showed that higher scores on NAS increase the risk of having acted aggressively in the past and of acting aggressively in the future. Keywords Novaco Anger Scale and Provocation Inventory, anger assessment, cross-cultural scale validation, aggression, hospital violence The assessment of anger has wide-ranging relevance for many fields of psychology, both basic and applied. It is especially relevant for clinical, forensic, and health psychology, where interventions for the regulation of anger are formulated and implemented. As treatment for anger problems should be grounded in assessment of deficits in anger control, validation of the assessment procedures is an important clinical research agenda. The biopsychosocial nature of anger brings to the fore issues of cross-cultural differences in anger experience, anger expression, and symbolization, which bear on the content and psychometric properties of anger assessment instruments. Thus, when established anger measures are translated and transported into another cultural context, rigorous empirical analyses should be conducted. As anger is a subjectively experienced affect, self-report assessment is fundamental. There are many anger assessment instruments, and Eckhardt, Norlander, and Deffenbacher (2004) have provided a review of anger selfreport scales. One such instrument that has received extensive validation is the Novaco Anger Scale (NAS), which has a companion measure, the Provocation Inventory (PI).

The NAS and PI (Novaco, 1994) were developed for use with mentally disordered populations, and they were validated in the landmark MacArthur study on violence and mental disorder (Monahan et al., 2001). Subsequently, they were revised and formally published with a separate standardization study (Novaco, 2003). Independent studies have confirmed the validity of the NAS with a variety of populations, including as a predictor of violence (e.g., Baker, Van Hasselt, & Sellers, 2008; Doyle & Dolan, 2006a; Hornsveld, Muris, & Kraaimaat, 2011; Novaco & Taylor, 2004; Swogger, Walsh, Homaifar, Caine, & Conner, 2012; Ullrich, Keers, & Coid, 2013). The present study concerns 1

Psychiatric Center Capital Region, Psychiatric Research Unit, North of Zealand, Denmark 2 University of California, Irvine, CA, USA 3 Psychiatric Center Capital Region, Sct. Hans, Denmark 4 Psychiatric Center Capital Region, Ballerup, Denmark Corresponding Author: Stine Bjerrum Moeller, Psychiatric Center Capital Region, Psychiatric Research Unit, Dyrehavevej 48, Hilleroed, North of Zealand, Denmark. Email: [email protected]

Downloaded from asm.sagepub.com at CMU Libraries - library.cmich.edu on September 22, 2015

2

Assessment 

the Danish translation of the NAS and PI, testing their psychometric properties with nonclinical, prisoner, general psychiatric, and forensic samples.

Anger, Violence, and Clinical Care Needs The prevalence of anger as a salient problem across clinical and forensic populations is now widely recognized, as illustrated by studies with psychiatric outpatients (Posternak & Zimmerman, 2002), psychiatric inpatients (Novaco, 1997), male and female adult criminal offenders (Suter, Byrne, Byrne, Howells, & Day, 2002), and political prisoners (Schutzwohl & Maercker, 2000). Anger dysregulation is part of a broad range of clinical disorders (DiGiuseppe & Tafrate, 2007; Novaco, 2010), and its relevance for clinical and forensic populations centrally concerns it being a dynamic risk factor for violence. Studies with multiple control variables show anger to be related to the violent behavior of psychiatric patients before, during, and after hospitalization (Doyle & Dolan, 2006a, 2006b; McNiel, Eisner, & Binder, 2003; Monahan et al., 2001; Novaco & Taylor, 2004; Skeem et al., 2006; Swogger et al., 2012) and to physical aggression within institutions by incarcerated adults (Wang & Diamond, 1999). Anger is not only an important clinical need, it also bears on the therapeutic milieu and on the well-being of clinical and custodial staff. To assess anger for treatment, to formulate anger treatment, and to evaluate treatment programs for problematic anger, it is essential to have reliable and valid psychometric instruments. However, no such anger assessment measures are available in Danish. As a consequence, studies of anger as a clinical problem have been limited in Denmark, where the current clinical practice for anger assessment, to the extent that it exists at all, is not based on a differentiated assessment of the anger construct.

Problems of Anger and Aggression in Danish Populations As in other countries, anger and aggressive behavior are clinically significant ward atmosphere variables in psychiatric hospitals in Denmark, being more prevalent on locked units and generally detracting from the satisfaction of both patients and staff (Middelboe, Schjødt, Byrsting, & Gjerris, 2001; Schjødt, Middelboe, Mortensen, & Gjerris, 2003). The prevalence of physical attacks on human service staff by clients in Denmark is evidenced in a study by Rasmussen, Hogh, and Andersen (2013) involving psychiatry, eldercare, special schools, and prisons involving 5,497 respondents. In psychiatry services, previous 12-month exposure rates were 11.0% for being punched with a fist, 26.8% for being kicked, and 33.8% for having a hard object thrown at them. The rates on those attack types were substantially higher in special schools. “Threats of beatings” had a prevalence of 47.9% in psychiatry services. Being threatened in an insulting manner

(which is indicative of anger) was reported to be “frequent” (weekly or daily) by 13.8% of psychiatry services staff, and 29.5% in special schools, 40.6% in prisons, and 29.8% in eldercare. Perhaps the first study of patient violence in a Danish psychiatric hospital was conducted by Benjaminsen, Gøtzsche-Larsen, Norrie, Harder, and Luxhøi (1996). Over their 1-year study period involving 1,130 patients on five wards (four were acute admission wards) at a Danish university hospital, they found that 70 patients (6.2%) committed physically violent acts. Such violence occurred at a 1-year rate of 1.4 per occupied bed, with serious violence occurring at a rate of 0.35 per bed. The violence rate was 13 times higher (4.75 violent acts per bed) on the two wards that were locked, compared with the open wards. Pertinent to anger, among the 67 patients with paranoid schizophrenia diagnosis, for whom anger is salient (cf. Novaco, 2010), the rate of violence was 43.3%, and the vast majority (70.1%) of the violent acts were rated as having “provoking antecedents.” As discussed by Benjaminsen et al. (1996), the overall rate of patient violence was comparable to that found in various studies conducted in British, U.S., and Canadian psychiatric hospitals. More recently, Pedersen, Ramussen, and Elsass (2012) reported that 37.0% of the patients at a Danish forensic hospital had one or more aggressive incidents during hospitalization. Among a sample of patients with schizophrenia diagnosis in Danish forensic and nonforensic services (Bo, Forth, Kongerslev, Haahr, Pedersen, et al., 2013), 52.8% exhibited premeditated aggression and 47.2 % exhibited impulsive aggression. The latter type of aggression is generally understood as being anger reactive. Clearly, assaultive behavior is as prevalent in Danish psychiatric facilities as it is in other countries. However, anger was not assessed in those studies of inpatient violence in Danish hospitals, despite grounds for seeing the aggressive behavior as being associated with anger-related factors. Prompted by criticism from a European Committee for Prevention of Torture and Inhumane or Degrading Treatment in 2008, the Minister of Health in Denmark commissioned a comparative study of coercive measures in psychiatric hospitals, which found that Danish psychiatric hospitals had a high use of mechanical restraints and physical restraint, compared with six other European countries (Bak & Aggernaes, 2012). A Danish Norwegian study (Bak, Zoffmann, Sestoft, Almvik, & Brandt-Christensen, 2014) on reducing mechanical restraint, missed the potential role of patients’ anger as an activator of behaviors that result in episodes of mechanical restraint, and thus missed anger treatment as a potential remedy. Similarly, studies in Danish hospitals of measurement instruments concerning psychiatric patients’ aggressive behavior and social dysfunction (e.g., Bech, 1994; Wistedt et al., 1990) have not attended to anger. This neglect or oversight is perhaps due to the absence of a valid Danish language psychometric instrument for anger.

Downloaded from asm.sagepub.com at CMU Libraries - library.cmich.edu on September 22, 2015

3

Moeller et al. The relevance of anger assessment for Danish populations is broader than just for psychiatric patients. Thastum, Ravn, Sommer, and Trillingsgaard (2009) found anger to be highly correlated (.61) with disruptive behavior in a large nonclinical sample of Danish schoolchildren and to be significantly higher in their clinical sample of children, especially for mood disorders and conduct disorders. Moller and Siguroardottir’s (2009) large sample study of drivers in Denmark identified anger and angry driving behaviors as one of three measurement factors of driving style, and it was significantly related to accidents, criminal record, drunk driving, and drug driving. Multivariate studies of Danish citizens by Petersen et al. (Petersen, 2010; Petersen, Sznycer, Cosmides, & Tooby, 2012) have found anger to be significantly related to public opinions about welfare and criminal justice. However, none of these studies measured anger with a validated instrument. In forensic psychiatry in Denmark, the focus is on risk assessment, typically by use of the Historical Clinical Risk Management-20 (e.g., Pedersen et al., 2012) and Psychopathy Checklist–Revised (e.g., Andersen, Sestoft, Lillebaek, Mortensen, & Kramp, 1999; Bo, Forth, Kongerslev, Haahr, Heldt, et al., 2013). The prison and probation service has implemented anger management programs; however, inmates are selected for treatment based only on their index crime—not on their clinical needs or the assessment of anger—and clinical evaluation of efficacy has typically been absent. General psychiatry and psychiatric wards in Denmark have no scientific clinical practice with regard to anger, despite the fact that the political system in Denmark is very concerned with the use of restraint. Given this established relevance of anger for both adults and children in Denmark, in clinical and nonclinical populations, the present study sought to translate the NAS and evaluate its validity with diverse samples. Previous validation studies of NAS translations include cultures comparable to the Danish, these being Dutch (Hornsveld et al., 2011), Icelandic (Gudjonsson & Sigurdsson, 2007), and Swedish (Lindqvist, Dåderman, & Hellström, 2005), demonstrating that the NAS measures a cross-cultural anger construct.

NAS-PI Components and Previous Translations Because the NAS has been extensively validated with many clinical populations, has had successful cross-cultural translation, has demonstrated validity in predicting violence by psychiatric patients, and is an anger assessment instrument that lends itself to clinical case formulation,1 it was selected for Danish translation and evaluation in the present multistudy project. The NAS and PI (Novaco, 1994, 2003) are measures of anger disposition. The NAS is composed of cognitive, arousal, and behavioral domains, plus a separate anger regulation subscale. The PI assesses anger intensity

across a range of situations; it has no measurement-level subscales. The NAS and PI are packaged as companion instruments. Fuller description of these measures is given in the Method section. The NAS-PI (1998 version) was translated into Swedish and evaluated with a group of 95 male violent offenders (Lindqvist et al., 2005). Both NAS and PI total scores were found to have high internal consistency, and NAS concurrent validity was established with correlations of .86 with the Aggression Questionnaire (Buss & Perry, 1992) and of .79 with the State-Trait Anger Expression Inventory–2 (STAXI-2; Spielberger, 1999). Its construct validity was supported by a significant correlation of .29 with previous convictions. In a study with undergraduates, factor analyses were supportive of the NAS measurement structure (Lindqvist, Dåderman, & Hellström, 2003). Hornsveld et al. (2011) translated the NAS-PI (1994 version) to Dutch and examined their psychometric properties with psychiatric inpatients, forensic outpatients, and students. Although fit indices for the three-subscale structure for the NAS were unsatisfactory, the instruments’ internal consistency, test– retest reliability, and criterion validity were robust.

Current Study Objectives The main goal of the present study was to develop NAS-PI2003-D and to test its psychometric properties. Internal reliability was evaluated in nonclinical, clinical, and forensic samples; test–retest was evaluated in the clinical sample. A confirmatory factor analysis (CFA) approach was used to evaluate the structure of the NAS (Cognitive, Arousal, Behavioral, and Regulation subscales) and the PI, expecting to confirm the solution of four factors for the NAS and a single factor for the PI with acceptable fit indices. Analyses were run for the combined samples. Relevant to clinical treatment decisions, we investigated whether the Danish version of the NAS-PI could discriminate between clinical and nonclinical populations. The concurrent validity and discriminate validity of the Danish NAS were evaluated in the clinical sample with psychometric criteria. Gender is an important factor for understanding anger from a clinical standpoint. Various reviews (e.g., Archer, 2004; Stoney & Engebretson, 1994) have concluded that there is an absence of gender differences in anger, but most studies have not involved clinical populations. However, large sample studies of psychiatric patients have highlighted important gender differences. In the Posternak and Zimmerman (2002) study of outpatients, gender was predictive of anger for those with Axis II disorders (males higher). Sadeh and McNeil (2013) found facets of the NAS to be predictive of suicide attempts as a function of gender and sexual victimization history in the year following hospital discharge—suicide attempts higher for men as a function of NAS Behavioral and for women as a function of

Downloaded from asm.sagepub.com at CMU Libraries - library.cmich.edu on September 22, 2015

4

Assessment 

NAS Arousal. Variation in cardiovascular reactivity associated with anger has also been demonstrated in experimental laboratory research (e.g., Lai & Linden, 1992), and Suter et al. (2002) reported much higher anger scores on both NAS and STAXI measures for female prisoners as compared with men. The present study’s examination of gender differences is exploratory, and it does not have females in its forensic samples. Pertinent to the aggressive behavior component of the anger construct and previous studies that have found anger to be predictive of violence by psychiatric patients, we tested the retrospective and prospective validity of the Danish NAS regarding forensic inpatients’ (males) physically aggressive behavior in hospital. We expect that NAS total scores will be significantly associated with independently recorded physical aggression, before and after anger assessment, controlling for length of stay in hospital.

Method Participants Participants were drawn from nonclinical, clinical, and correctional settings, each of which had variations within that type of setting. Nonclinical Sample.  The nonclinical sample (N = 477) that completed the NAS and the PI consisted of university students (N = 243), community adults who were members of a political party (N = 126), and private corporation employees (N = 108). The total sample consisted of 167 males (35%) and 306 (64%) females. The males (years of age, M = 33.3, SD = 13.4) were older than the females (years of age, M = 28.1, SD = 10.1), F(1, 470) = 22.2, p < .001. Males (M = 14.6, SD = 2.6) also had more years of education than did females (M = 14.3, SD = 1.6), F(1, 451) = 4.1, p = .044. There were missing data on gender (4 missing), age (5 missing), and length of education (24 missing). Regarding NAS-PI scores, for this sample and for each of the others, less than 5% of responses were missing, and no respondent was missing more than three items. The values for the missing items were replaced with the series mean for the item. Mixed Clinical Sample. This sample was composed of 87 males (35%) and 159 (64%) females. There were missing data for 4 patients on gender, 4 on age, and 10 on length of education. The instruments completed among those in the total clinical sample (N = 250) varied across settings. Patients at the psychiatric facilities in South of Zealand (N = 77) completed the NAS, the STAXI-2, and the Hospital Anxiety and Depression Scale (HADS). Outpatients at the psychiatric facility of North of Zealand, Hilleroed (N = 86), completed the NAS, the PI, and the STAXI-2. Patients

at two open wards at Frederikssund hospital (N = 87) completed the NAS and the PI. There were no significant differences in age between males (M = 40.1) and females (M = 38.1), nor in years of education (M = 10.7 and M = 11.2, respectively). A subsample of the clinical outpatients at Hilleroed (N = 57) was retested 2 to 4 weeks later to provide evaluation of test–retest reliability. Offenders Sample.  The offenders sample was composed of male inmates at five different prisons in Denmark and of male forensic patients at one forensic hospital, 10 of these being outpatients. The inmates (N = 167; 61% of whom were convicted of a violent crime) completed only the PI. The male forensic patients (N = 64; 92% of whom were convicted of a violent crime) completed only the NAS; 54 of these were inpatients at a secure hospital. The mean age, across offenders, was 32.6 years (SD = 11.4, range = 18-67); mean length of education was 9.5 years (SD = 2.2, range = 7-15). Seven cases had missing data on length of education.

Procedure Data were gathered anonymously from these sites between August 2007 and May 2012, as part of a larger data collection with psychological instruments. All participants were volunteers. Written information about the study was provided, and anonymity was assured. The study was approved by the Danish Data Control system. For the forensic and psychiatric patients, it was emphasized that participants would not be identified in subsequent reports and that personal data would not be used in the psychiatric or the forensic system. The data collection procedure for the nonclinical samples varied by setting. The university students were volunteers who were given the NAS and PI at three large auditorium lectures and provided with a stamped, addressed return envelope. The volunteering students completed the questionnaire in their spare time and returned it by postal mail. About 50% of the students attending the lectures returned the questionnaire. Participants from the community subsample were recruited by the third author at two political booth camps, attended by 189 people. Volunteers filled out the NAS and PI in their spare time during the camps, and 128 (67.7%) returned the questionnaire. The private corporation employees were recruited by the first author during their lunch hour. They received oral and written information about the study and were asked to complete the NAS and PI forms within 2 days, returning the questionnaires to a postal box in the lunchroom. The return rate for these participants could not be calculated. For the prison samples, the PI was administered as part of a larger assessment battery. All participants were volunteers.

Downloaded from asm.sagepub.com at CMU Libraries - library.cmich.edu on September 22, 2015

5

Moeller et al. In one prison, they were recruited by prison staff; in two prisons, all inmates were informed about the study in writing, and then recruited by inmate spokesmen; in another prison, the recruiting was done directly by the first author; and in the last prison, the recruiting was done solely by inmate spokesmen. Test administration was conducted in small groups by the first author, except in one prison, where it was self-administered and returned to the spokesmen, who then gave the questionnaires to the first author. Since prison staff were involved in participant recruitment, and because incarcerated prisoners’ self-report of anger may be distorted by socially desirable responding, efforts were made by the first author to reduce such possible selection and response biases. In that regard, she participated in the recruitment of participants, provided reassurance of the anonymity of test responses, stressed that reassurance, and provided information about the study’s purpose. Calculating the return rate was not possible in the prison setting. The forensic inpatients were approached on their hospital wards by the first author and individually tested by her on their wards. Of 88 available inpatients, 54 (61.1%) participated. The 10 forensic outpatients were approached by their case manager and were tested in their home environment. Aggressive behavior data for inpatients were recorded from the hospital’s journal system. For the mixed clinical sample, the questionnaire set was distributed by clinical staff. In the outpatient setting, participants completed the questionnaires on their own in a private place. In the inpatient settings, the questionnaires were individually administered by clinical staff, although some participants completed them in groups of four to eight, with staff direction. The first author participated in patient/staff meetings to provide information about the study.

Measures Novaco Anger Scale (Novaco, 2003).  The NAS is a 60-item scale constructed to measure anger disposition. Its items were generated from a theoretical framework. The Cognitive, Arousal, and Behavioral subscales are each composed of 16 items. The sum of these 48 items forms the NAS total score. There is also a separate 12-item Anger Regulation subscale. All items are rated on a 3-point scale of 1 = never true, 2 = sometimes true, and 3 = always true. The Cognitive subscale is composed of items in four content categories: justification, suspicion, rumination, and hostile attitude. The Arousal subscale’s content categories are intensity, duration, somatic tension, and irritability. The Behavioral subscale consists of items that operationalize impulsive reactions, verbal aggression, physical confrontation, and indirect expression. The Anger Regulation scale assesses the capacity to regulate anger-engendering thoughts, to selfcalm somatic arousal and agitation, and to engage in constructive behavior when faced with provocation. Alpha

scores and test–retest reliability across various settings have shown excellent reliability (Novaco, 2003; Novaco & Taylor, 2004). The validity of the NAS has been established by independent investigators (e.g., Baker et al., 2008; Hornsveld et al., 2011; Jones, Thomas-Peter, & Trout, 1999), including the prediction of violent behavior by psychiatric patients (Doyle & Dolan, 2006a, 2006b; McNiel et al., 2003; Monahan et al., 2001; Swogger et al., 2012). Provocation Inventory (Novaco, 2003). The PI is a 25-item self-report instrument measuring anger intensity. The instrument describes situations that could potentially elicit anger, and the respondent rates anger intensity on a 4-point scale. The types of provocations concern disrespect, unfairness, frustration, annoying traits of others, and irritations as content areas, but the sole psychometric index is the PI total score. Higher scores indicate greater anger. In the standardization of the NAS-PI (N = 1,546), the internal reliability alpha for the PI was .95 (Novaco, 2003); in a civil psychiatric sample (N = 1,101), the alpha was .92 (Monahan et al., 2001), and among intellectually disabled forensic patients (Novaco & Taylor, 2004) the alpha was .92. The stability and validity of the PI has been supported in a variety of samples, including prisoners (Baker et al., 2008; Jones et al., 1999; Mills, Kroner, & Forth, 1998). Successful translations of the tool have been made into Swedish (Lindqvist et al., 2003) and Dutch (Hornsveld et al., 2011). With permission from the original author (Ray Novaco, personal communication, August 2007), the first author translated the NAS-PI into Danish, and two independent, bilingual native speakers (one was an expert in the field and the other had no knowledge of psychology) back-translated the questionnaire. Back-translations were compared with that of the original by an independent psychiatrist, and differences were discussed and resolved between the editing psychiatrist and the primary researcher. In this process, the original author was contacted to ensure the comparability over cultural differences and idiomatic expressions. A number of the items underwent alternations in wording during this process (NAS: Item 13, wording was adjusted to connote physical confrontation; Item 25, wording was adjusted to connote self-control; Item 42, wording was adjusted to connote verbal aggression, Item 60, wording was adjusted to connote capacity to disengage from an anger-inducing verbal exchange). State-Trait Anger Expression Inventory (Spielberger, 1999). The STAXI-2 is a 57-item scale constructed to measure anger experiences, anger disposition, and anger expression. The scale consists of six subscales measuring State Anger, Trait Anger, and the components of Anger Expression (Anger– Out, Anger–In, and Anger Control). The Anger Control dimension has component subscales of Anger Control–In, measuring the tendency to invest energy in calming down

Downloaded from asm.sagepub.com at CMU Libraries - library.cmich.edu on September 22, 2015

6

Assessment 

and securing inner control, and Anger Control–Out, measuring the tendency to invest energy in monitoring and preventing the outward expression of anger. The STAXI-2 is considered a strong anger assessment instrument with solid psychometric properties in varied settings. Spielberger (1999) reported data from 1,600 normal adults and 274 hospitalized psychiatric patients, for which the internal reliability scores had the following ranges: Trait Anger, from .84 to .87; State Anger, from .92 to .94; Anger–Out, from .74 to .80; Anger In, from .74 to .82; Anger Control–Out from .84 to .87; and Anger Control–In from .91 to .93. Regarding validity, the STAXI differentiated between healthy and clinical participants. Previous studies have found strong correlations between STAXI and NAS-PI subscales in English, in Swedish, and in Dutch, in hospital, prison, and community settings (e.g., Hornsveld et al., 2011; Lindqvist et al., 2003; Mills et al, 1998; Novaco & Taylor, 2004). With permission from the original author (Charles Spielberger, personal communication, March 2012), the STAXI-2 was translated and back-translated by a bilingual translator. The first author reviewed the back-translation. Hospital Anxiety and Depression Scale (Zigmond & Snaith, 1983). This instrument is a 14-item self-report questionnaire measuring anxiety and depression. The respondent provides ratings for his or her most recent week. Seven items measure anxiety, and seven items measure depression. Higher scores indicate higher levels of anxiety and depression. Its reliability and validity have been established in clinical settings and in the general population (Bjelland, Dahl, Haug, & Neckelmann, 2002). The questionnaire was available in the public domain and was translated and backtranslated by a bilingual translator. Staff Observation Aggression Scale–Revised (Nijman et al., 1999).  This is a form used to register aggressive incidents on wards. Staff members use the checklist to note the presence of specific characteristics of incidents that he or she has witnessed. Information collected includes provocation, means, target, consequences, and actions to stop the aggression. The Staff Observation Aggression Scale–Revised (SOAS-R) form gathers information about aggressive incidents, including the number of verbal incidents, the number of incidents involving an object without direct human contact (e.g., slamming doors or throwing objects not directly toward another person), the number of incidents involving direct contact with another person (e.g., throwing an object toward another person or grabbing another person), and last, the number of incidents involving direct contact judged to be more severe and potentially dangerous (e.g., an attempt to strangle or the direct use of an object). The SOAS-R has shown good interrater reliability in countries comparable to Denmark (Nijman et al., 1999; Nijman, Palmstierna, Almvik, & Stolker, 2005) and had already

been translated and electronically implemented in clinical practice at the study’s forensic hospital. Its data were therefore available in the forensic patient files prior to data collection. Verbal incidents were excluded. The total number of physical incidents retrospective to hospital admission was recorded, but if the patient’s hospital stay was less than 14 days, he was omitted from the analyses. For prospective analyses, total number of physical incidents was registered up to the patient’s release from hospital. The follow-up period length had a maximum of 5 months following anger assessment. SOAS-R data were available for the 54 hospitalized forensic patients, but 4 had insufficient length of stay, leaving 50 cases with SOAS-R data. During the period from being admitted to the hospital and the test date, 44% of the patients had had at least one aggressive incident, and 36% of the patients had at least one aggressive incident from the test date to the end of the observation period.

Results Descriptive Statistics and Reliability Analyses Descriptive statistics and internal consistency coefficients (alpha) for the subscales of the Danish translation of the NAS-PI, partitioned by sample type are displayed in Table 1. For the clinical and nonclinical subsamples, means and standard deviations are separately displayed for males and females, and the analysis of variance (ANOVA) test results for the exploratory gender comparisons are also in Table 1. Regarding gender differences, the ANOVA tests, apart from a few exceptions, were nonsignificant. For those that were significant, Cohen’s d coefficients were computed on the magnitude of the effect size. Cohen’s ds are interpreted as 0.20 = small effect, 0.50 = medium effect, and 0.80 = large effect (Cohen, 1992). For the clinical patients, males had significantly higher scores on NAS Behavioral, F(1, 244) = 4.9, p = .027, d = 0.28. Males were also higher on State Anger, F(1, 157) = 4.4, p = .038, d = 0.32, and Trait Anger, F(1, 157) = 4.1, p = .045, d = 0.33. In the nonclinical sample, the obtained gender differences were reversed, with females having significantly higher NAS Arousal, F(1, 471) = 15.4, p < .001, d = 0.36, and PI scores, F(1, 471) = 6.5, p = .011, d = 0.24. The effect sizes are all small in the comparisons involving gender. Internal reliability was excellent overall for total score indices. For the clinical patients, alpha was greater than .90 for the NAS and PI total scores, as well as for the STAXI Trait Anger and State Anger scales. The NAS subscales also had high alphas (.82-.89), but the NAS Regulation scale (.71) was just above the acceptable range. For the nonclinical sample, the internal reliabilities for NAS and PI total scores were also high, and they were excellent for the offender samples (see Table 1).

Downloaded from asm.sagepub.com at CMU Libraries - library.cmich.edu on September 22, 2015

7

Moeller et al. Table 1.  Descriptive Statistics, Gender Comparisons, and Internal Reliability of Anger Scores for Clinical, Offender, and Nonclinical Samples. Total   Clinical patients NAS  Cognitive  Arousal  Behavioral  Regulationa NAS total PI total STAXI-2  Trait  State   Anger Expression–Out   Anger Expression–In   Anger Control–Out   Anger Control–In Male forensic patients NAS  Cognitive  Arousal  Behavioral  Regulationa NAS total Male inmates PI total Nonclinical sample NAS  Cognitive  Arousal  Behavioral  Regulationa NAS total PI total

Male

Female

SD

M

SD

M

SD

31.0 31.6 27.1 25.3 89.6 59.3

5.4 6.0 6.8 3.6 16.5 12.7

31.5 31.6 28.3 25.8 91.5 57.3

6.1 6.0 7.4 4.1 18.2 12.1

30.6 31.6 26.4 25.0 88.6 60.2

4.9 6.0 6.3 3.6 15.6 12.9

1.72 0.00 4.93** 2.73 1.70 2.00

.82 .85 .89 .71 .94 .92

20.7 20.8 15.2 19.1 22.1 20.0

7.3 9.4 5.1 4.7 5.9 6.2

22.1 22.7 16.0 18.8 22.8 20.5

7.8 10.4 5.9 4.6 6.2 5.8

19.7 19.6 14.6 19.1 21.8 19.7

6.9 8.5 4.5 4.6 5.8 6.5

4.10* 4.38** 3.23 0.18 1.15 0.73

.91 .96 .63 .75 .90 .65

34.1 32.2 30.9 26.8 97.1

5.8 5.8 5.8 4.4 16.8

  .80 .79 .86 .77 .93

65.4

14.0

.92

26.9 24.9 23.2 28.2 74.8 51.9

3.7 4.1 3.7 3.0 10.0 10.9

26.6 25.9 23.3 27.9 75.8 53.5

3.6 4.2 3.8 3.0 10.0 10.3

26.5 26.4 23.4 27.7 76.3 54.4

3.6 4.2 3.9 2.9 10.0 9.9

F

α

M

1.05 15.44** 0.60 2.95 2.41 6.49**

.70 .76 .74 .63 .88 .89

Note. STAXI-2 = State-Trait Anger Expression Inventory; NAS = Novaco Anger Scale; PI = Provocation Inventory. NAS-PI scores of male (N = 87) and female (N = 159) clinical patients, male forensic patients (N = 64), male inmates (N = 167), and male (N = 167) and female (N = 306) nonclinical sample. STAXI-2 scores of male (N = 59) and female (N = 100) and missing (N = 4) clinical patients. Effect sizes are given in the text. a. NAS Regulation is not included in the NAS total. F tests are ANOVAs for gender differences. *p < .05. **p < .01.

The test–retest evaluation, conducted with a subsample (N = 57) of the clinical outpatients, found strong reliability coefficients (one-way random model) for the NAS-PI and STAXI instruments: NAS total (.77), PI total (.73), STAXI State Anger (.82), and STAXI Trait Anger (.71). Overall, the NAS-PI reliability coefficients are comparable to those given in the instrument’s manual (Novaco, 2003).

Discrimination of Clinical and Nonclinical Samples Comparisons of NAS and PI means in Table 1 of the the clinical patients, the forensic patients, and the nonclinical participants

were performed by ANOVA tests. As these tests bear on the validity of the NAS and PI in discriminating nonclinical from clinical populations, Cohen’s d coefficients were computed on the magnitude of the effect size. The clinical patients and forensic patients, respectively, had significantly higher scores than did the nonclinical participants on NAS total and each of its subscales: NAS total, F(1, 725) = 197.5, p < .001, d = 1.01, and F(1, 539) = 213.6, p < .001, d = 1.54; NAS Cognitive, F(1, 725) = 163.8, p < .001, d = 0.96, and F(1, 539) = 198.5, p < .001, d = 1.55; NAS Arousal, F(1, 725) = 225.5, d = 1.10, and F(1, 539) = 115.2, p < .001, d = 1.24; and NAS Behavioral, F(1, 725) = 93.5, p < .001, d = 0.69, and F(1, 539) = 179.0, p < .001, d = 1.55. This was

Downloaded from asm.sagepub.com at CMU Libraries - library.cmich.edu on September 22, 2015

8

Assessment 

Table 2.  Correlations Between NAS-PI and STAXI-2 and HADS for the Clinical Sample (N = 163). STAXI-2

HADS (N = 77)



State

Trait

Anger Expression–Out

Anger Expression–In

Anger Control–Out

Anger Control–In

NAS  Cognitive  Arousal  Behavioral  Regulation Total PI

.49 .53 .53 −.25 .56 .38

.73 .78 .82 −.33 .85 .42

.52 .52 .65 −.24 .62 .36

.52 .62 .35 −.27 .53 .54

−.41 −.47 −.63 .62 −.56 −.46

−.23 −.30 −.44 .52 −.36 −.21

Depression

Anxiety

.12 .27 .26 −.31 .24

.42 .61 .44 −.26 .54  

Note. STAXI-2 = State-Trait Anger Expression Inventory; HADS = Hospital Anxiety and Depression Scale; NAS = Novaco Anger Scale; PI = Provocation Inventory. Cohen (1992), r: .20 = small, .30 = medium, .50 = large.

also the case for the PI total, F(1, 648) = 35.0, p < .001, d = 0.50, and F(1, 642) = 133.7, p < .001, d = 0.97 for the clinical and forensic patients compared with the nonclinical participants. Thus, for each of these comparisons, the effect size is large and is in the expected direction. For NAS Regulation score contrasts, the clinical patients and forensic patients, respectively, had significantly lower scores (i.e., less anger regulation) than did the nonclinical participants, F(1, 725) = 108.4, p < .001, d = 0.78, and F(1, 539) = 6.1, p = .014, d = 0.29. Thus, the clinical patients and the forensic patients reported the expected lower anger regulation than the nonclinical sample. The effect size is medium/large in the comparison involving the clinical patients and small for that involving the forensic patients. Comparing the NAS and PI means of the clinical and forensic patients, the scores of the clinical patients were significantly lower on all NAS measures, except the Arousal subscale: NAS total, F(1, 312) = 10.5, p = .001, d = 0.45; NAS Cognitive, F(1, 312) = 16.3, p < .000, d = 0.55; NAS Arousal, F(1, 312) = 0.5, p = .477, d = 0.10; NAS Behavioral, F(1, 312) = 16.3, p < .000, d = 0.60; NAS Regulation, F(1, 312) = 8.7, p = .004, d = 0.37. The clinical patient PI mean is also significantly lower than that for the forensic patients, F(1, 312) = 17.7, p < .000, d = 0.46. Thus, the forensic patients had higher anger disposition, but, unexpectedly, they also reported higher anger regulation than the clinical patients. The effect sizes for the significant differences between the forensic patients and clinical patients are small to medium.

Concurrent Validity The concurrent validity of the Danish NAS and PI was addressed by their correlations (Pearson) with the STAXI-2 in the clinical sample. Results are given in Table 2, showing the expected high correlations with the STAXI-2 across all NAS-PI indices, with the highest correlation being between NAS total and Trait Anger. Regarding the other

most conceptually comparable subscales, NAS Behavioral is strongly associated with STAXI Anger Expression–Out and, inversely, with Anger Control–Out. The NAS Behavioral correlation with STAXI Anger–Out is substantially higher than it is with STAXI Anger–In; whereas the NAS Arousal correlation with STAXI Anger–In is higher than it is with STAXI Anger–Out. Also, NAS Regulation has a strong positive association with STAXI Anger Control–Out and Anger Control–In.

Discriminant Validity Evidence for discriminant validity is also contained in Table 2. The correlation of NAS total and its subscales with STAXI Trait Anger are substantially higher than their correlations with HADS Depression and HADS Anxiety. The least discriminating comparison is the difference between the NAS Arousal and STAXI Trait Anger correlation compared with the NAS Arousal and HADS Anxiety correlation (r = .78 vs. r = .61), yet this contrast is significant in differential magnitude of correlation, z = 2.40, p < .01, despite the likely shared variance between the NAS and HADS scales in assessing somatic arousal. The correlation of NAS Behavioral is substantially higher with STAXI-2 Anger Expression–Out than with Anger Expression–In (r = .64 vs. r = .35), z = 3.51, p < .01.

Confirmatory Factor Analyses The factor structure of the Danish translation of the NAS-PI was tested by running a CFA using STATA 13 on the combined sample. The models were composed of NAS Cognitive, NAS Arousal, NAS Behavioral, and NAS Regulation, and separately PI total, as specified in the NAS-PI manual (Novaco, 2003). Factors were allowed to intercorrelate. The goodness-of-fit was determined on the basis of several indices: chi-square, comparative fit index (CFI), and the root mean square error of approximation (RMSEA). When a

Downloaded from asm.sagepub.com at CMU Libraries - library.cmich.edu on September 22, 2015

9

Moeller et al. Table 3.  Poisson Regression of Anger on Patients’ Physical Aggressive Behavior in Hospital (Forensic Sample). Variable

B

Retrospective aggression  Intercept −8.46   NAS total .039 Prospective aggression  Intercept −8.70   NAS total .036

SE

Wald

p

Exp(B)

CI

1.52 0.015

30.95 6.94

.000 .008

0.000 1.039

[1.079, 0.004] [1.010, 1.069]

1.07 0.010

66.26 12.88

.000 .000

0.000 1.037

[2.048, 0.001] [1.017, 1.058]

Note. CI = confidence interval; NAS = Novaco Anger Scale. n = 50. The dependent variable is the number of physical aggressive acts as measured by the Staff Observatory Aggressive Scale−Revised. The test variable is the NAS total score. The time since admission is used as an offset varibale to account for the opportunity to behave aggressively.

good-fit model has been achieved, the chi-square test should be nonsignificant, or the ratio of the chi-square divided by the degrees of freedom should be less than 2. CFI values larger than .90 to .95 and a RMSEA value below .06 to .07 are indicative of a reasonable model fit (Ullman, 2007). The model fit tests for the NAS were χ2(1704) = 6362.48, p < .0001, CFI = .688, RMSEA = .059, indicating a model fit on only one of the indices. The model fit tests for the PI were χ2(275) = 1374.17, p < .0001, CFI = .844, RMSEA = .070, indicating approximate model fit on only one index.

Predictive Validity The predictive validity of the Danish NAS was tested using a Poisson regression model (Venables & Ripley, 2003) with NAS total as a continuous predictor variable (as it is the summary anger disposition index) and the number of aggressive incidents as the dependent variable. Those data were available for the 54 cases with inpatient status. Four cases were omitted because they had stayed less than 14 days on the ward prior to testing or had stayed less than 14 days in the follow-up. The sample size was based on practical considerations, rather than power planning. Age and length of education were examined as potential covariates; however, because both had nonsignificant, zero-order correlations with both retrospective and prospective aggression, they were not entered into the regressions. To account for the different times under risk, the logarithm of the time since admission was used as an offset variable. Possible heterogeneous aggressive tendencies within the sample (i.e., overdispersion) were accounted for by means of the robust Huber–White sandwich estimator for the covariance matrix and standard errors of the parameter estimates (Freedman, 2006). The analyses are presented in Table 3, and we report the results as risk ratios per 1 and 10 unit(s) increase in NAS, along with their 95% confidence intervals (CIs). With regard to the retrospective aggression, higher scores on NAS are associated with more aggressive incidents (risk ratio = 1.039, 95% CI [1.01, 1.07], p = .008). In other words, a 1-unit increase in NAS is accompanied by a

4% increase in aggressive incidents, or 10 units NAS increase by 47% increase in aggressive incidents. Regarding future aggression, higher NAS scores also go along with more aggressive incidents (risk ratio = 1.037, 95% CI [1.02, 1.06], p = .000). A 1-unit increase in NAS is accompanied by a 4% increase in aggressive incidents, or 10-unit NAS increase by 44% increase in aggressive incidents.

Discussion Anger is highly prevalent in clinical populations, signifying clinical needs, as well as risk for violent behavior by psychiatric patients. In treatment settings, it is also an important ward atmosphere variable bearing on clinical care in hospital (Eklund & Hansson, 1997; Middelboe et al., 2001; Røssberg & Friis, 2003), and it is associated with quality of life for patients in therapeutic community settings (Fassino, Amianto, Gastaldo, & Leombruni, 2009). Among nonpsychiatric community adults, those with high anger dispositions report more physical aggression and more substance use (Tafrate, Kassinove, & Dundin, 2002). With anger having such relevance, the present study set out to translate an established anger psychometric scale, the NAS-PI, and evaluate its validity for Danish populations, nonclinical, clinical, and forensic. The present evaluation of the psychometric properties of the Danish NAS-PI was done in a multistudy, multisite project. It is also the first research on any anger instrument in Denmark. The results across the set of studies provide support for the reliability (internal and test–retest), concurrent validity, discriminant validity, and predictive validity of the Danish NAS-PI. However, the tests of the NAS and PI factor structure were generally not supportive, as acceptable model fit was found only for the RMSEA index. Those results for the CFA are comparable to that obtained by Hornsveld et al. (2011) for the Dutch translation of the NAS-PI. That nonfitting of the items to the Cognitive, Arousal, Behavioral, and Regulation structure of the NAS may be due either to language translation, cultural differences, or instrument limitations. Regarding the translation process, we were diligent in attending to nuances in item meaning and adjusting item

Downloaded from asm.sagepub.com at CMU Libraries - library.cmich.edu on September 22, 2015

10

Assessment 

wording through back-translation in correspondence with the instrument’s original author. As the NAS-PI construction was intended to facilitate clinical case formulation and identify targets in treatment planning, this may have detracted from producing a psychometrically robust factor structure. Barrett (2007) criticized approximate fit indices such as RMSEA and argued that the value of a model, whether it fits or not, hinges on external criteria, which CFA does not supply, and that, in the absence of fit, the issue is whether the model has empirical or pragmatic value. The practicality of the NAS and PI structure is its interface with a cognitive–behavioral treatment approach, as noted earlier. However, the nonsupportive CFA findings call in question the usefulness of the NAS subscales in cross-cultural applications. The correlations with the STAXI are supportive of concurrent validity. The evidence for discriminant validity was reflected in the NAS correlations with the STAXI being significantly higher than with HADS Depression, whereas with regard to HADS Anxiety, the comparative results were mixed. The higher correlation between the NASs and HADS Anxiety than between the NASs and HADS Depression is perhaps indicative of shared variance with anxiety in negative affectivity and autonomic system arousal (e.g., DiGiuseppe & Tafrate, 2007; Kreibig, 2010). The present study finding that, among the NAS subscale scores, the Arousal subscale had the highest correlation with HADS Anxiety is consistent with findings reported in the NAS manual (Novaco, 2003). Discriminative validity for the Danish NAS and PI in differentiating the clinical and nonclinical samples was substantiated. Also, the forensic patients had significantly higher PI total, NAS total, and NAS subscale scores than did the clinical patients. That the forensic patients reported relatively high Regulation scores—those in the present study are comparable to those reported in the NAS manual (Novaco, 2003)—may be because the Regulation scale does not differentiate between what one is “trying to do” and what one is successfully doing in actuality or because forensic patients may be trying to represent their anger coping skills as being better than they are (social desirability or reactivity2). Alternatively, forensic patients who are experiencing anger more frequently, more intensely, and for longer duration and whose behavioral manifestations of anger have caused them problematic consequences may be more cognizant of their anger regulation efforts and/or more actively trying to self-regulate. Gender differences were generally absent, except that in the clinical sample, females had significantly lower scores on NAS Behavioral and on STAXI-2 State and Trait anger. In the nonclinical sample, females had significantly higher scores on NAS Arousal and PI total. We had no prediction regarding gender, because previous findings with nonforensic populations have been mixed. Unfortunately, the present study’s offender samples did not include females. In studies

with forensic populations, higher anger among females compared with males has been found on the NAS (Chilvers & Thomas, 2011; Leenaars, 2005; Suter et al., 2002) and on other anger measures (Archer & Haigh, 1997; Novaco, 1997). Among offenders, Baker et al. (2008) reported that females had significantly higher NAS Arousal and PI scores than did males. As there were no female prisoners or female forensic patients in the present study, the gender difference issue among offender populations remains to be evaluated with the Danish-translated anger measures. In view of the findings of Sadeh and McNeil (2013) regarding gender, sexual victimization, and NAS subscales bearing on suicide attempts, there is relevance for self-harm as well as harm to others. The Danish NAS and PI scores are generally comparable to those reported in published studies involving the instrument translations into Swedish (Lindqvist et al., 2005) and Dutch (Hornsveld et al., 2011) with nonclinical, clinical, or offender/forensic samples. The Danish NAS and PI scores for the nonclinical sample are also comparable in magnitude to those of a British sample (Jones, Thomas-Peter, & Gangstad, 2003), but significantly lower (p < .01) than that study’s clinical sample. The overall comparability of the Danish anger scores to those studies in three other countries support the cross-cultural applicability of NAS-PI. Furthermore, the high level of anger reported by the participants in the forensic hospital and prison samples give credence to the merit of inquiring about anger among such populations and to including anger as a dynamic variable (cf. Douglas & Skeem, 2005) when conducting research on aggression with offender populations. This has been lacking in Denmark, partly due to the absence of assessment tools. Predictive validity was evaluated by retrospective and prospective analyses of physically aggressive behavior in hospital, testing NAS total as the predictor and controlling for length of stay in hospital. The results were significant in both the retrospective and prospective analyses, using the number of aggressive incidents as criteria. NAS total was predictive of violent behavior. As that predictive value of the NAS for hospital patient violence has been found by other investigators retrospectively prehospital and in hospital (McNiel et al., 2003; Novaco & Taylor, 2004), prospectively in hospital (Doyle & Dolan, 2006a), and prospectively in the community (Doyle & Dolan, 2006b; Monahan et al., 2001, Swogger et al., 2012; Ullrich et al., 2013), this has larger importance because anger is a viable treatment target. The latter point has been established by numerous meta-analyses. Regarding offender populations, whether in correctional institutions or in hospitals, the present study results have relevance for selection of candidates for “anger management” interventions in Denmark and for the evaluation of such programs. For those purposes, our findings support the NAS-PI as a suitable instrument to correct shortcomings in procedures in Danish facilities, where individuals are enlisted into

Downloaded from asm.sagepub.com at CMU Libraries - library.cmich.edu on September 22, 2015

11

Moeller et al. such programs without anger having been assessed. That the NAS and PI scores for the offender samples are in the clinical range is indicative of those respondents being anger treatment candidates. The present study has several limitations. First, the nonclinical sample has limited representativeness, as it was composed of a convenience aggregate of university students, political party members, and private company employees. It had more females than males, a modest age range, and relatively high education. However, the clinical sample is reflective of such populations in Denmark, with varied diagnostic status and a mixture of inpatients and outpatients from different locations. The offender sample was recruited from different prisons to provide for representativeness, but participants were all volunteers and were all males, as was the forensic patient sample. Also, NAS data were not obtained from the inmates, and PI data were not obtained from the forensic patients, both due to practical constraints involving resources. The forensic sample was small with only 54 inpatients and 10 outpatients. An additional limitation was the absence of a check on deception. It is reassuring that the study participants who had histories of offending were not reticent about reporting their anger, in contrast to very low anger scores being reported by incarcerated felons in some studies (e.g., Loza & LozaFanous, 1999). However, we did not incorporate measures of social desirability, deception, or impression management as validity checks. The relatively high NAS Regulation scores for the forensic sample may be indicative of impression management or self-deception, as we have discussed. Reflecting on the Danish national health survey reporting that violence exposure increases health care costs, particularly for women (Helweg-Larsen, Sørensen, Brønnum-Hansen, & Kruse, 2011), our study findings regarding anger and aggressive behavior point to dysregulated anger among patients having a link to health care costs. That a cross-culturally adapted, psychometrically sound anger instrument is now available in Danish will allow for more nuanced research on aggression in care facilities in Denmark and generate data points on anger for cross-cultural comparison. In adding to research on anger, we also hope to stimulate psychotherapy outcome studies aimed at decreasing anger dysregulation. Author’s note We are thankful for the statistical assistance by Matthias Gondan and the staff´s assistance in patient recruitment and data collection at the hospitals in Frederikssund and South of Zealand. We also would like to thank Kuno Herman Lund at the Department of Prison and Probation Service in Denmark for allowing the data collection within the 5 prisons, Tine Wøbbe for including the first author in her group of clinical psychologists during the period of datacollection at Sct. Hans Hospital, forensic department, and finally Mette Lynge Ungstrup for helping to collect data in the non-clinical sample.

Declaration of Conflicting Interests The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Stine Bjerrum Moeller receives royalities from Hogrefe Publisher, but none were received from the use of the instrument in this study.

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 Ph.D. scholarship from the University of Copenhagen, Department of Psychology.

Notes 1.

The NAS, with its Cognitive, Arousal, and Behavioral domains and the theoretically generated thematic content areas within them, was constructed to interface with the cognitive-restructuring, arousal reduction, and behavioral coping skills components of anger treatment (cf. Novaco, 1994, 2003). Scores on NAS item thematic categories (e.g., “justification” and “rumination” in the Cognitive domain; “intensity” and “duration” in the Arousal domain; “impulsivity” and “physical confrontation” in the Behavioral domain) facilitate case formulation by identifying treatment targets. The PI, with its inventory of situations in which anger is activated, can be used as a structured interview to facilitate understanding of how the person experiences anger. 2. Anger assessment, particularly in forensic contexts, is susceptible to reactivity as a threat to validity, as respondents may report their anger in anticipation of what it means to an audience. It is not uncommon for those in custodial settings to underreport or “mask” their anger on psychometric instruments (cf. Novaco, 2003).

References Andersen, H. S., Sestoft, D., Lillebaek, T., Mortensen, E. L., & Kramp, P. (1999). Psychopathy and psychopathological profiles in prisoners on remand. Acta Psychiatrica Scandinavica, 99, 33-39. Archer, J. (2004). Sex differences in aggression in real-world settings: A meta-analytic review. Review of General Psychology, 8, 291-322. doi:10.1037/1089-2680.8.4.291 Archer, J., & Haigh, A. (1997). Beliefs about aggression among male and female prisoners. Aggressive Behavior, 23, 405415. doi:10.1002/(SICI)1098-2337(1997)23:63.0.CO;2-F Bak, J., & Aggernaes, H. (2012). Coercion within Danish psychiatry compared with 10 other European countries. Nordic Journal of Psychiatry, 66, 297-302. Bak, J., Zoffmann, V., Sestoft, D. M., Almvik, R., & BrandtChristensen, M. (2014). Mechanical restraint in psychiatry: Preventive factors in theory and practice. A Danish-Norwegian Association study. Perspectives in Psychiatric Care, 50, 155-166. Baker, M. T., Van Hasselt, V. B., & Sellers, A. H. (2008). Validation of the Novaco Anger Scale in an incarcerated offender population. Criminal Justice and Behavior, 35, 741754. doi:10.1177/0093854808316275

Downloaded from asm.sagepub.com at CMU Libraries - library.cmich.edu on September 22, 2015

12

Assessment 

Barrett, P. (2007). Structural equation modelling: Adjudging model fit. Personality and Individual Differences, 42, 815824. doi:10.1016/j.paid.2006.09.018 Bech, P. (1994). Measurement by observations of aggressive behaviour and activities in clinical situations. Criminal Behaviour and Mental Health, 4, 290-302. Benjaminsen, S., Gøtzsche-Larsen, K., Norrie, B., Harder, L., & Luxhøi, A. (1996). Patient violence in a psychiatric hospital in Denmark: Rate of violence and relation to diagnosis. Nordic Journal of Psychiatry, 50, 233-242. doi:10.3109/08039489609081413 Bjelland, I., Dahl, A. A., Haug, T. T., & Neckelmann, D. (2002). The validity of the Hospital Anxiety and Depression Scale: An updated literature review. Journal of Psychosomatic Research, 52(2), 69-77. doi:10.1016/S0022-3999(01)00296-3 Bo, S., Forth, A., Kongerslev, M., Haahr, U. H., Heldt, U., Pedersen, L., & Simonsen, E. (2013). Subtypes of aggression in patients with schizophrenia: The role of psychopathy. Journal of Forensic Psychology & Psychiatry, 24, 496-513. Bo, S., Forth, A., Kongerslev, M., Haahr, U. H., Pedersen, L., & Simonsen, E. (2013). Subtypes of aggression in patients with schizophrenia: The role of personality disorders. Criminal Behaviour and Mental Health, 23, 124-137. Buss, A. H., & Perry, M. (1992). The aggression questionnaire. Journal of Personality and Social Psychology, 63, 452-459. doi:10.1037/0022-3514.63.3.452 Chilvers, J., & Thomas, C. (2011). Do male and female forensic patients with learning disabilities differ on subscales of the Novaco Anger Scale and Provocation Inventory (NAS-PI)? Journal of Learning Disability and Offending Behaviour, 2, 84-97. doi:10-1108/20420921111152469 Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155-159. DiGiuseppe, R., & Tafrate, R. C. (2007). Understanding anger disorders. New York, NY: Oxford University Press. Douglas, K., & Skeem, J. (2005). Violence risk assessment: Getting specific about being dynamic. Psychology, Public Policy, and Law, 11, 347-383. doi:10.1037/1076-8971.11.3.347 Doyle, M., & Dolan, M. (2006a). Evaluating the validity of anger regulation problems, interpersonal style, and disturbed mental state for predicting inpatient violence. Behavioral Sciences & the Law, 24, 783-798. doi:10.1002/bsl.739 Doyle, M., & Dolan, M. (2006b). Predicting community violence from patients discharged from mental health services. British Journal of Psychiatry, 189, 520-526. doi:10.1192/bjp. bp.105.021204 Eckhardt, C., Norlander, B., & Deffenbacher, J. (2004). The assessment of anger and hostility: A critical review. Aggression and Violent Behavior, 9, 17-43. doi:10.1016/ S1359-1789(02)00116-7 Eklund, M., & Hansson, L. (1997). Relationships between characteristics of the ward atmosphere and treatment outcome in a psychiatric day-care unit based on occupational therapy. Acta Psychiatrica Scandinavica, 95, 329-335. doi:10.1111/j.1600-0447.1997.tb09640.x Fassino, S., Amianto, F., Gastaldo, L., & Leombruni, P. (2009). Anger and functioning amongst inpatients with schizophrenia or schizoaffective disorder living in a therapeutic community. Psychiatry and Clinical Neurosciences, 63, 186-194. doi:10.1111/j.1440-1819.2009.01940.x

Freedman, D. A. (2006). On the so-called “Huber sandwich estimator” and “robust standard errors.” The American Statistician, 60, 299-302. Gudjonsson, G. H., & Sigurdsson, J. F. (2007). Motivation for offending and personality: A study among young offenders on probation. Personality and Individual Differences, 42, 12431253. doi:10.1016/j.paid.2006.10.003 Helweg-Larsen, K., Sørensen, J., Brønnum-Hansen, H., & Kruse, M. (2011). Risk factors for violence exposure and attributable healthcare costs: Results from the Danish national health interview surveys. Scandinavian Journal of Public Health, 39(1), 10-16. doi:10.1177/1403494810380774 Hornsveld, R. H. J., Muris, P., & Kraaimaat, F. W. (2011). The Novaco Anger Scale–Provocation Inventory (1994 version) in Dutch forensic psychiatric patients. Psychological Assessment, 23, 937-944. doi:10.1037/a0024018 Jones, J. P., Thomas-Peter, B. A., & Gangstad, B. (2003). An investigation of the factor structure of the Novaco Anger Scale. Behavioural and Cognitive Psychotherapy, 31, 429437. doi:10.1017/S1352465803004041 Jones, J. P., Thomas-Peter, B. A., & Trout, A. (1999). Normative data for the Novaco Anger Scale from a non-clinical sample and implications for clinical use. British Journal of Clinical Psychology, 38, 417-424. doi:10.1348/014466599163024 Kreibig, S. D. (2010). Autonomic nervous system activity in emotion: A review. Biological Psychology, 84, 394-421. doi:10.1016/j.biopsycho.2010.03.010 Lai, J. Y., & Linden, W. (1992). Gender, anger expression style, opportunity for anger release determine cardiovascular reaction to and recovery from anger provocation. Psychosomatic Medicine, 54, 297-310. Leenaars, P. E. M. (2005). Differences between violent male and violent female forensic psychiatric outpatients: Consequences for treatment. Psychology, Crime & Law, 11, 445-455. doi:10.1080/10683160500256412 Lindqvist, J. K., Dåderman, A. M., & Hellström, Å. (2003). Swedish adaptations of the Novaco Anger Scale-1998, the Provocation Inventory, and the State-Trait Anger Expression Inventory-2. Social Behavior and Personality, 31, 773-788. Lindqvist, J. K., Dåderman, A. M., & Hellström, Å. (2005). Internal reliability and construct validity of the Novaco Anger Scale-1998-S in a sample of violent prison inmates in Sweden. Psychology, Crime & Law, 11, 223-237. doi:10.1080/10683160500036863 Loza, W., & Loza-Fanous, A. (1999). Anger and prediction of violent and non-violent offenders’ recidivism. Journal of Interpersonal Violence, 14, 1014-1029. McNiel, D. E., Eisner, J. P., & Binder, R. L. (2003). The relationship between aggressive attributional style and violence by psychiatric patients. Journal of Consulting and Clinical Psychology, 71, 399-403. doi:10.1037/0022006X.71.2.399 Middelboe, T., Schjødt, T., Byrsting, K., & Gjerris, A. (2001). Ward atmosphere in acute psychiatric in-patient care: Patients’ perceptions, ideals and satisfaction. Acta Psychiatrica Scandinavica, 103, 212-219. doi:10.1034/j.16000447.2001.00102.x Mills, J. M., Kroner, D. G., & Forth, A. E. (1998). Novaco Anger Scale: Reliability and validity within an adult criminal sample. Assessment, 5, 237-248.

Downloaded from asm.sagepub.com at CMU Libraries - library.cmich.edu on September 22, 2015

13

Moeller et al. Moller, M., & Siguroardottir, S. B. (2009). The relationship between leisure time and driving style in two groups of male drivers. Transportation Research Part F, 12, 462-469. Monahan, J., Steadman, H. J., Silver, E., Appelbaum, P. S., Robbins, P. C., Mulvey, E. P., . . . Banks, S. (2001). Rethinking risk assessment: The MacArthur study of mental disorder and violence. New York, NY: Oxford University Press. Nijman, H. L. I., Muris, P., Merckelbach, H. L. G. J., Palmstierna, T., Wistedt, B., Vos, A. M., . . . Allertz, W. (1999). The staff observation aggression scale–revised (SOAS-R). Aggressive Behavior, 25, 197-209. doi:10.1002/(SICI)10982337(1999)25:33.0.CO;2-C Nijman, H. L. I., Palmstierna, T., Almvik, R., & Stolker, J. J. (2005). Fifteen years of research with the staff observation aggression scale: A review. Acta Psychiatrica Scandinavica, 111, 12-21. doi:10.1111/j.1600-0447.2004.00417.x Novaco, R. W. (1994). Anger as a risk factor for violence among the mentally disordered. In J. Monahan & H. J. Steadman (Eds.), Violence and mental disorder: Developments in risk assessment (pp. 21-61). Chicago, IL: University of Chicago Press. Novaco, R. W. (1997). Remediating anger and aggression with violent offenders. Legal and Criminological Psychology, 2(1), 77-88. doi:10.1111/j.2044-8333.1997.tb00334.x Novaco, R. W. (2003). The Novaco Anger Scale and Provocation Inventory. Los Angeles, CA: Western Psychological Services. Novaco, R. W. (2010). Anger and psychopathology. In M. Potegal, G. Stemmler & C. Spielberger (Eds.), International handbook of anger (pp. 465-497). New York, NY: Springer. Novaco, R. W., & Taylor, J. L. (2004). Assessment of anger and aggression in male offenders with developmental disabilities. Psychological Assessment, 16, 42-50. doi:10.1037/10403590.16.1.42 Pedersen, L., Ramussen, K., & Elsass, P. (2012). HCR-20 violence risk assessments as a guide for treating and managing violence risk in a forensic psychiatric setting. Psychology, Crime & Law, 18, 733-743. doi:10.1080/1068316X.2010.548814 Petersen, M. B. (2010). Distinct emotions, distinct domains: Anger, anxiety and perceptions of intentionality. Journal of Politics, 72, 357-365. doi:10.1017/S00223816099079X Petersen, M. B., Sznycer, D., Cosmides, L., & Tooby, J. (2012). Who deserves help? Evolutionary psychology, social emotions, and public opinions about welfare. Political Psychology, 33, 395-418. doi:10.1111/j.1467-9221.2012.00883.x Posternak, M. A., & Zimmerman, M. (2002). Anger and aggression in psychiatric outpatients. Journal of Clinical Psychiatry, 63, 665-672. doi:10.4088/JCP.v63n0803 Rasmussen, C. A., Hogh, A., & Andersen, L. P. (2013). Threats and physical violence in the workplace: A comparative study of four areas of human service work. Journal of Interpersonal Violence, 28, 2749-2769. Røssberg, J. I., & Friis, S. (2003). Do the Spontaneity and Anger and Aggression subscales of the Ward Atmosphere Scale form homogeneous dimensions? A cross-sectional study of 54 wards for psychotic patients. Acta Psychiatrica Scandinavica, 107, 118-123. Sadeh, N., & McNeil, D. E. (2013). Facets of anger, childhood sexual victimization, and gender as predictors of suicide attempts by psychiatric patients after hospital discharge. Journal of Abnormal Psychology, 122, 879-890. doi:10.1037/a0032769

Schjødt, T., Middelboe, T., Mortensen, E. L., & Gjerris, A. (2003). Ward atmosphere in acute psychiatric inpatient care: Differences and similarities between patient and staff perceptions. Nordic Journal of Psychiatry, 57, 215-220. doi:10.1080/08039480310001382 Schutzwohl, M., & Maercker, A. (2000). Anger in former East German political prisoners: Relationship to posttraumatic stress reactions and social support. Journal of Nervous and Mental Disease, 188, 483-489. Skeem, J. L., Schubert, C., Odgers, C., Mulvey, E. P., Gardner, W., & Lidz, C. (2006). Psychiatric symptoms and community violence among high-risk patients: A test of the relationship at the weekly level. Journal of Consulting and Clinical Psychology, 74, 967-979. doi:10.1037/0022-006X.74.5.967 Spielberger, C. D. (1999). STAXI-2 State-Trait Anger Expression Inventory–2 (Professional manual). Tampa. FL: Psychological Assessment Resources. Stoney, C. M., & Engebretson, T. O. (1994). Anger and hostility: Potential mediators of the gender difference in coronary heart disease. In A. W. Siegman & T. W. Smith (Eds.), Anger, hostility, and the heart (pp. 215-237). Hillsdale, NJ: Erlbaum. Suter, J. M., Byrne, M. K., Byrne, S., Howells, K., & Day, A. (2002). Anger in prisoners: Women are different from men. Personality and Individual Differences, 32, 1087-1100. doi:10.1016/S0191-8869(01)00105-2 Swogger, M. T., Walsh, Z., Homaifar, B. Y., Caine, E. D., & Conner, K. R. (2012). Predicting self- and other-directed violence among discharged psychiatric patients: The roles of anger and psychopathic traits. Psychological Medicine, 42, 371-379. doi:10.1017/S0033291711001243 Tafrate, R. C., Kassinove, H., & Dundin, L. (2002). Anger episodes in high- and low-trait-anger community adults. Journal of Clinical Psychology, 58, 1573-1590. doi:10.1002/jclp.10076 Thastum, M., Ravn, K., Sommer, S., & Trillingsgaard, A. (2009). Reliability, validity and normative data for the Danish Beck Youth Inventories. Scandinavian Journal of Psychology, 50(1), 47-54. doi:10.1111/j.1467-9450.2008.00690.x Ullman, J. B. (2007). Structural equation modeling. In B. G. Tabachnick & L. S. Fidell (Eds.), Using multivariate statistics (5th ed., pp. 676-781). Boston, MA: Pearson. Ullrich, S., Keers, R., & Coid, J. W. (2013). Delusions, anger, and serious violence: New findings from the MacArthur Violence Risk Assessment Study. Schizophrenia Bulletin, 40, 11741181. doi:10.1093/schbul/sbt126 Venables, W. N., & Ripley, B. D. (2003). Statistics and computing: Modern applied statistics with S (4th ed.). New York, NY: Springer. Wang, E. W., & Diamond, P. M. (1999). Empirically identifying factors related to violence risk in corrections. Behavioral Sciences & the Law, 17, 377-389. doi:10.1002/(SICI)10990798(199907/09)17:33.0.CO;2-M Wistedt, B., Rasmussen, A., Pedersen, L., Malm, U., TraskmanBendz, L., Wakelin, J., & Bech, P. (1990). The development of an observer-scale for measuring social dysfunction and aggression. Pharmacopsychiatry, 23, 249-252. Zigmond, A. S., & Snaith, R. P. (1983). The Hospital Anxiety and Depression Scale. Acta Psychiatrica Scandinavica, 67, 361370. doi:10.1111/j.1600-0447.1983.tb09716.x

Downloaded from asm.sagepub.com at CMU Libraries - library.cmich.edu on September 22, 2015

Validation of the Novaco Anger Scale-Provocation Inventory (Danish) With Nonclinical, Clinical, and Offender Samples.

Anger has high prevalence in clinical and forensic settings, and it is associated with aggressive behavior and ward atmosphere on psychiatric units. D...
315KB Sizes 1 Downloads 5 Views