UNDERSTANDING THE AGGRESSIVE WORKPLACE: DEVELOPMENT OF THE WORKPLACE AGGRESSION TOLERANCE QUESTIONNAIRE

Participants were asked to indicate the appropriateness of each action in the workplace. Responses were recorded on a 5-point Likert scale ranging from "very inappropriate" to "very appropriate." An evaluation of appropriateness was selected because it would reveal the extent to which attitudes in the workplace were conducive to peoples' acceptance of aggressive acts. The scenario indicates that the workplace aggression would be directed at the manager, not the person answering the items. People could have a much different view of appropriateness if they were to be the target of the aggressive behavior. We were interested in perceptions of the general use of aggression in the workplace, not for aggression specifically targeting the respondent.

Item Evaluation and Reliability

Each WATQ item was examined to evaluate its contribution to the consistency of the scale. An item was eliminated if it reduced the Cronbach's alpha to below .90. In addition, item-total correlations for the items were reviewed. The minimum criterion for item retention in the scale was .40. No item-total scale correlation fell below .40 and the scale demonstrated an acceptable inter-item correlation of .42. It was unnecessary to delete any items from the scale. The overall reliability for the WATQ was .95 (standardized Cronbach's alpha).

Factor Analysis

A principal components factor analysis using oblique rotation was performed on the WATQ. If the WATQ contained sub-scales they probably would be related to one another, hence the use of the oblique rotation. (Varimax rotation was performed with essentially the same results.) Following the advice of Burgoon and Hale (1987), a set of four criteria were used to evaluate the factor structure: (1) eigen values had to exceed 1.0; (2) a factor had to have at least 3 items meeting the .50/.30 strength/purity standard for factor loadings; (3) the scree test had to show each additional factor was making a reasonable improvement in the variance accounted for; and (4) any item in a given factor had to have a primary factor loading of .50 or better.

Five factors having eigenvalues of 1 or better emerged and accounted for 67.8% of the total variance. The five factors accounted for 44.5%, 10.0%, 5.4%, 4.3%, and 3.6% of the variance, respectively. Because there was a great deal of cross loading between factors, only seven items could meet the .50/.30 strength/purity standard for factor loadings. The factors were not clearly interpretable. Moreover, the scree test criteria suggested a single factor solution should be retained. The analyses did not seem to justify converting the factors into sub-scales. Forcing a single factor solution resulted in factor loadings of .45 or higher which are acceptable (Stuckless & Goranson, 1992). Table 2 presents the factor loadings for the single factor solution as well as the means and standard deviations for each item. Overall, the analyses suggest that the WATQ is fundamentally uni-dimensional. All 28 items were retained for use in Study 2 because they met the item evaluation and reliability criteria.