Psychological Reports, 1990, 66, 1251-1258. O Psychological Reports 1990

RELIABILITY A N D VALIDITY OF T H E ABUSIVE VIOLENCE SCALE ' CHARLES HENDRIX Oklnhoma State University

AND

WALTER SCHUMM k n s a s State University

Summary.-Because of the stressful nature of the war in Vietnam, many Americans continue to be affected by their service. One aspect of war service, which may be significantly linked to the formation of Posttraumatic Stress Disorder, is the degree of personal experience with abusive violence. To develop a reliable and valid measure of this construct, construction of an Abusive Violence Scale was undertaken. In the current study, the Abusive Violence Scale showed internal consistency (Cronbach or = .81) and validity, as indicated by significant Pearson correlations with two measures of combat exposure and measures of the PTSD dimensions of intrusiveness and avoidance. Validity was further supported by factor analysis that yielded only one factor, abusive violence.

Of the nearly 3.7 million Americans who served in the Vietnam Theatre of Operations by 1975 (Brende & Parson, 1985), many continue to be negatively affected by their service in a variety of ways (Figley & Sprenkle, 1978; Figley, 1986; Rosenheck & Thomson, 1986; Silver & Iacono, 1986). For many veterans, one result of having served in Vietnam is expressed in Posttraumatic Stress Disorder (PTSD) (APA, 1987). The primary dimensions of this disorder are the intrusiveness of memories of a traumatic event and the avoidance of reminders of that event. The stressful nature of the war in Vietnam, as evidenced by the amount of combat exposure, is the factor most strongly linked to veterans' developing symptoms (MacPherson, 1985). Moos (1986) suggests that some soldiers reacted to the stressful nature of this war with rage or anger against others, including innocent victims. Laufer and Gallops (1985, p. 840) label such a reaction to war stress as abusive violence or "the arbitrary use of vioIence against persons even when not necessitated by self defense." Although there is much research linking the amount of combat exposure to the subsequent development of the disorder, there is little empirical research into the long-term effects of exposure to acts of abusive violence. This may be related to the lack of an instrument on which to assess reactions to exposure to abusive violence.

Instrument Development The object of this report is to describe the development and preliminary 'This is Contribution No. 90-233-J from the Kansas Agriculture Experiment Station. Address correspondence to Charles Hendrix, Ph.D., Department of Family Relations and Child Development, Oklahoma State University, Stillwater, Oklahoma 74078.

1252

C. HENDRIX & W. SCHUMM

evaluation of a new measuring instrument designed to assess exposure to abusive violence in a combat environment. The authors' personal experiences and discussions with Vietnam veterans suggested that there were three general categories of exposure to abusive violence: no exposure at all, exposure in the sense of having witnessed violence without participating in it, and exposure by having perpetrated abusive violence against enemy personnel or civilians. However, among those who only witnessed abusive violence, there were several possible perpetrators, including enemy forces, allied forces (e.g., South Vietnamese soldiers, Korean soldiers, etc.), and U.S. forces. Within the last three groups the authors hypothesized that personal involvement would be greatest when U.S. forces were involved and least when enemy forces were perpetrators, especially in the context of t h s study which focused on the long-term effects of involvement in abusive violence rather than short-term effects. In the short run, distress at finding the aftermath of abusive violence perpetrated by the enemy might outweigh the effects of the guilt at having witnessed (condoned) or participated in abusive violence. However, in the long run, the authors hypothesized that the effects of guilt for actions over which the individual had at least control would decrease more slowly than the effects of anger or remorse over thngs that were beyond the individual's control. In other words, the soldier is hkely to recognize the futility of continued distress about enemy abusive violence, over which the soldier had no control, but the soldier may find it harder to forget or forgive himself for illegal or immoral actions over which he did have control, even if the circumstances seemed to support such actions at the time. Therefore, five items were developed to measure increasing levels of involvement in abusive violence during the Vietnam war. The specific items in the scale were developed by the authors on the basis of their clinical and military experiences on active duty and in the U.S. Army Reserve, * in which they were able to interview informally a number of Vietnam veterans about their exposure to, and experience with, abusive violence in Vietnam. The first item was a screening question, used to allow respondents who had not been involved in abusive violence to skip the next four items and to proceed directly to the remainder of the survey. The Laufer and Gallops (1985) definition of abusive violence was included in the introduction to these questions to give subjects a basis for comparison with their own experience. In Appendix A (p. 1258) are presented the introduction and the five items.

'The first author is a marriage and Family therapist specializing in the treatment of traumatized families. The second author served two years on active duty in the U.S. Army, 1972-1974, primarily with the Firsc Infancry Division at Fort Riley, Kansas, and subsequently served nearly 16 years in the active reserves, having been selected for promotion to the rank of Lieutenant Colonel with an effective date of 15 December 1989.

ABUSIVE VIOLENCE SCALE

1253

METHOD The five-item Abusive Violence Scale was administered to 60 midwestern Vietnam veterans along with the Combat Exposure Scale (Lund, Foy, Sipprelle, & Strachan, 1984), the Impact of Events Scale (Schwarzwald, Solomon, Weisenberg, & Mikulincer, 1987), and a question asking for a personal rating of intensity of combat exposure (scored 4, heavy combat; 3, moderate; 2, light; and 1, no combat). The Combat Exposure Scale has seven items representing stressful military events that are ordered hierarchically by Guttman scaling. As a measure of reliability, Lund, et al. (1984) report a coefficient of reproducibility of .93, where .90 indicates a reliably constructed Guttman scale. As a measure of construct validity, scores on the Combat Exposure Scale yielded a Pearson correlation of .86 ( p < ,001) with scores on the combat scale used by Egendorf, Kadushin, Laufer, Rothbart, and Sloan (1981). The Impact of Events Scale is a 15-item scale describing emotional reactions to war stress that hvides reactions into two categories, intrusion of memories and avoidance of reminders of Vietnam. Schwarzwald, et al. (1987) cite Cronbach alpha coefficients of .91 (intrusion) and .84 (avoidance) as evidence of internal consistency. The scale's construct validity was demonstrated by discrimination among three distinct groups (diagnosed combat stress reactions, combat control, and noncombat control) on both subscales. Subjects

All respondents were midwestern, U.S. military veterans who had served in the Vietnam Theatre of Operations. For the 57 indicating length of service in Vietnam, the range in months was from 1 to 36 (M = 16, SD = 7). About one-third (32%, n = 19) had been wounded sometime during their war service. For the 59 indicating combat exposure, there was a fairly even distribution of self-reported levels of combat exposure, with 31% (n = 18) reporting "Heavy Combat," 20% (n = 12) "Moderate Combat," 27% (n = 16) "Light Combat," and 22% (n = 13) "No Combat." Of the 60 total respondents, four were not included in analysis because their answers on the Abusive Violence Scale were incomplete. Abusive Violence Scale scores ranged from 1 (answering "Yes" to screening question, "No" to all others) to 5 (answering "Yes" to Question 5 , indicating that the respondent had committed acts of abusive violence), with a mean of 2.4 (SD = 1.6). Half of the sample indicated never having witnessed or participated in any acts of abusive violence (n = 28). Scores were computed based on the highest level of personal involvement in acts of abusive violence. Scores on the Combat Exposure Scale ranged from 0 to 7 (M= 4.3, SD= 1.6). In this sample, the Combat Exposure Scale yielded a coefficient of reproducibility of .9 1, indicating acceptable reliability. This compared to .93 reported by Lund, et a/. (1984). The scores on the Impact of Events

1254

C. HENDRIX & W. SCHUMM

Scale ranged from 0 to 5 (M = 1.9, SD = 1.4) for intrusion and from 0 to 4.3 (M = 1.1, SD = 1.2) for avoidance. Cronbach alpha coefficients for the intrusion and avoidance subscales were .91 and .82 compared to .91 and .84 from prior work by Schwarzwald, et al. (1987).

Reliability An SPSSX (1988) Reliability analysis of the internal reliability of four questions comprising the Abusive Violence Scale (dropping the first question) yielded a Cronbach (1951) alpha estimate for internal consistency of .81. Cronbach alpha is a modified, more general version of a split-half reliability coefficient, ranging from 0 to 1.0 increasing as the average interitem correlations and the number of items increase within any given set of items. Zeller and Carmines (1980) have described the characteristics of Cronbach (1951) alpha in detail (pp. 56-59) and have noted (p. 76) that "Cronbach's alpha is undoubtedly the most popular coefficient of internal consistency." Carmines and Zeller (1979) have furthermore stated that "As a general rule, we believe that reliabilities should not be below .80 for widely used scales." The means, standard deviations, and corrected item-total correlations are given in Table 1. TABLE 1 VALUESOF CRONBACH ALPHA OF THE ABUS~VE VIOLENCESCALEIF SPECIFIC ITEMS WEREDELETED, MEANSAND STANDARDDEVIATIONS ( N = 56) Item

a

M

Witnessed enemy abuse Witnessed allied abuse Witnessed U.S. abuse Participated i n abuse No items deleted

.78 .78 .72 .78 .81

.32 .38 .23 .18

SD .47 .49 .43 .39

The Cronbach alpha for the Abusive Violence Scale (.81) suggests reasonable internal consistency over the four items. The corrected item-total correlations show that all items should be retained, since deleting any item reduces the scale's internal reliability.

Construct Validio Adequate internal consistency reliability is a necessary, but not sufficient, prerequisite to high construct validity of any measure. Therefore, the Abusive Violence Scale was hypothesized to have adequate internal consistency reliability in terms of Cronbach alpha (1951), using the .80 criterion set by Carmines and Zeller (1979). To the authors' knowledge there is no previous measure of abusive violence; so concurrent validity could not be es-

1255

ABUSIVE VIOLENCE SCALE

tablished. However, the authors hypothesized that abusive violence would be significantly more likely to be encountered, or participated in, by service personnel who were involved in more intense combat or who had been exposed most to combat. Furthermore, the authors expected that involvement in abusive violence would be associated with subsequent development of PTSD symptoms, including intrusion and avoidance. Therefore, as a preliminary test of construct vahdity the authors hypothesized that scores on the Abusive Violence Scale would be positively correlated with the self-reported intensity of combat, scores on the Combat Exposure Scale, and scores on intrusion and avoidance. Finally, the authors expected that the common factor structure of the items would be unidimensional, so that a common factor analysis of the items was planned as a further test of the validity of the measure. A contrary finding of two or more factors would suggest that more than one concept was being measured. The Abusive Violence Scale yielded Pearson correlations of .58 (p

Reliability and validity of the Abusive Violence Scale.

Because of the stressful nature of the war in Vietnam, many Americans continue to be affected by their service. One aspect of war service, which may b...
274KB Sizes 0 Downloads 0 Views