Abstract
Infante and Wigley's (1986) Verbal Aggressiveness Scale (VAS) is a widely accepted and frequently used measure of trait verbal aggression. Although the scale is almost always scored as if it were unidimensional, previous factor analytic studies provide evidence that it is multidimensional with two distinct factors. The present studies (N = 194 and 177) used confirmatory factor analysis to replicate the two‐factor solution. The two‐factor model was consistent with the data, and provides a better fit to the data than the unidimensional solution. The first factor, comprised of all aggressively worded, nonreflected items, appears to measure verbal aggressiveness as intended whereas the second factor, comprised of all reverse‐scored items (benevolently worded), appears to measure a communication style related to other‐esteem confirmation and supportiveness. Given this interpretation, it is recommended that only the 10 aggressively worded items be scored. Hamilton, Buck, and Chory‐Assad, in an adversarial collaborative discussion, agree that the VAS is bidimensional, but offer an alternative conceptual model. They hold that the two factors reflect selfish individualism and prosocial cooperation.
Notes
Timothy R. Levine (Ph.D. Michigan State University, 1992) is Professor of Communication at Michigan State University. Michael J. Beatty (Ph.D. Ohio State University, 1976) is Professor and Chair of Communication at University of Missouri, St. Louis. Sean Limon (Ph.D. Michigan State University, 2000) is Assistant Professor of Communication at Illinois State University. Mark A. Hamilton (Ph.D., Michigan State University, 1987) is Associate Professor of Communication Sciences and Psychology at the University of Connecticut. Ross Buck (Ph.D., University of Pittsburgh, 1970) is Professor of Communication Sciences and Psychology at the University of Connecticut. Rebecca M. Chory‐Assad (Ph.D., Michigan State University, 2000) is an Assistant Professor of Communication Studies at West Virginia University. Correspondence to: Tim Levine, Department of Communication, Michigan State University, East Lansing, MI 48824‐1212, USA. Email: [email protected].
Given that the two dimensions were substantially correlated (−.46), and that the nonreflected aggressively worded items correlated more highly (.40) with the aggressive messages than the reflected, benevolently worded items (−.22), the negative correlation between the benevolently worded items and verbally aggressive messages is possibly spurious. To illustrate, the partial correlation between the aggressively worded items and aggressive messages controlling for the benevolently worded items is .35, whereas the partial correlation between the benevolently worded items and aggressive messages controlling for the aggressively worded items is −.04.
Although averaging the 10 aggressively worded items produced a score that is less reliable then the 20‐item total, applying Spearman‐Brown, it can be shown that the decline in reliability is less than would be expected given the reduction in the number of items. Given a 20‐item scale with α = .854 and all items of equal strength, α = .745 is expected if the scale was cut to 10 items. Calculated differently, given that the 10 aggressively worded items yielded a reliability estimate of α = .819, a reliability of α = .90 would be expected if the number of items were doubled. Similar observations were reported in Beatty et al. (Citation1999).
The scores on the 10 benevolently worded (reflected items) were reflected before they were averaged. Thus, lower scores on the reflected item subscale reflect endorsement of supportive, confirming communication.