~ Pergamon Behav. Res. Ther. Vol. 33, No. 8, pp. 991-993, 1995 Copyright © 1995 Elsevier Science Ltd

Printed i~ Great Britain. All rights reserved 0005-7967/95 $9.50 + 0.00

CONFIRMATORY FACTOR ANALYSIS OF THE AGGRESSION QUESTIONNAIRE

JULIE AITKEN HARRIS

Department of Psychology, Social Science Centre, University of Western Ontario, London, Ontario, Canada N6A 5C2

(Received 8 September 1994)

SuHnary--A confirmatory factor analysis of the factor structure of The Aggression Questionnaire created by Buss and Perry (1992) [Journal of Personality and Social Psychology, 63, 452-459] was conducted to assess whether the scale's purported 4 factors emerged. The results generally supported the 4-factor model. However, the hostility factor may be improved if 2 items pertaining to suspicion are removed from the scale. These items had relatively low loadings on that factor and decreased the hostility scale's internal reliability.

INTRODUCTION

Aggression is an important personality characteristic for both researchers and clinicians. Hues- mann, Eron, Lefkowitz and Walder (1984) reported that aggression is a stable personality trait. Based on a 22-yr longitudinal study of 600 Ss, Huesmann et al. found that early assessments of aggression predicted later antisocial behaviors. The present study examines a recently published self-report measure of aggression.

The Aggression Questionnaire was created by Buss and Perry (1992) and represents an updated and psychometrically improved version of The Buss-Durkee Hostility Inventory (Buss & Durkee, 1957). In creating The Aggression Questionnaire, Buss and Perry initially wrote items to assess 6 dimensions of aggression, including physical aggression, verbal aggression, anger, indirect aggres- sion, resentment, and suspicion. Following an exploratory factor analysis using principal axis factoring and oblimin rotation, 4 factors emerged: physical aggression, verbal aggression, anger, and hostility. The hostility factor included the items that were originally written for the resentment and suspicion scales. Using a second sample (N = 448 college students), Buss and Perry (1992) found the 4-factor model was supported by confirmatory factor analysis. The final questionnaire consists of 29 items. The S rates each item on a 5-point scale to indicate the degree to which the item is characteristic of him/her.

One of the psychometric improvements of The Aggression Questionnaire (Buss & Perry, 1992), when compared to The Buss-Durkee Hostility Inventory (Buss & Durkee, 1957), is the use of factor analysis in the construction of the scale. Factor analysis is a widely used method of obtaining the structure underlying personality measures (Eysenck & Rachman, 1966). Inventories based on a factor analytic method of construction generally have high internal consistency measures (Gorsuch, 1974), by allowing the test constructor to examine the degree to which individual items contribute to the factors (De Vellis, 1991).

The aim of the present study was to determine whether the factor structure of The Aggression Questionnaire (Buss & Perry, 1992) could be replicated (cross-validated) in an independent sample of S using confirmatory factor analysis. Since The Aggression Questionnaire is a rather recent personality measure, the present author felt that it was necessary to attempt to cross validate the structure of the questionnaire on an additional independent sample.

991

992 Jul ie A i t k e n Harris

METHOD

Subjects Three hundred and six (155 male, 151 female) university students completed The Aggression

Questionnaire as part of a battery of scales used in another study (see Harris, 1993). This sample was considered to be similar in composition to the Buss and Perry (1992) sample of college students.

PROCEDURE AND RESULTS

Confirmatory factor analysis with maximum likelihood, using LISREL VII (J6reskog & S6rbom, 1989), was employed to assess how well the data fit the 4-factor solution reported by Buss and Perry (1992). Thus, each scale represented a latent variable and each item was an indicator variable (Long, 1983). The 4-factor solution was found to have a moderately good fit to the data. Although the chi-square was significant [Z2(371) = 901.26, P < 0.001], the chi-square per degree-of-freedom was non-significant [Z2(1)=2.43, P <0.10]. The goodness-of-fit index was acceptably high (GFI = 0.825), as was the adjusted goodness-of-fit index ( A G F I - 0.794). The root mean square residual also suggested that the model provided a good fit to the data (RMSR = 0.077).

Table 1 lists the items of The Aggression Questionnaire and their estimated standardized factor loadings and standardized residuals for their designated factors. From the values listed in Table 1, it appears that 2 hostility items (numbers 6 and 8) have relatively lower loadings on their designated factor, when compared across all other items. The content of these items suggest that these items are measures of suspiciousness.

To assess the contribution of the 2 hostility items with low factor loadings to the original hostility scale, the internal consistency (alpha coefficients) were computed for the original 4 scales, as purported by Buss and Perry (1992), and for a revised hostility scale (the hostility scale after removing the 2 suspicion items). Table 2 lists the alpha coefficients for the scales as well as the predicted alpha coefficients, using the Spearman-Brown formula, for each scale if the scale contained 9 items. The predicted alpha coefficients were calculated so that comparisons could be

Table 1. Standardized factor loadings and residuals for items of The Aggression Questionnaire

Standardized Standardized Factor item factor loading residual

Physical aggression I. Once in a while I can't control the urge to strike another person. 2. Given enough provocation, I may hit another person. 3. I f somebody hits me, I hit back. 4. I get into fights a little more than the average person. 5. I f 1 have to resort to violence to protect my rights, I will. 6. There are people who pushed me so far that we came to blows. 7. I can think of no good reason for ever hitting a person? 8. I have threatened people I know. 9. I have become so mad that I have broken things.

Verbal aggression 1. I tell my friends openly when I disagree with them. 2. I often find myself disagreeing with people. 3. When people annoy me, I may tell them what I think of them. 4. I can ' t help getting into arguments when people disagree with me. 5. My friends say that I'm somewhat argumentative.

Anger I. I flare up quickly but get over it quickly. 2. When frustrated, I let my irritation show, 3. I sometimes feel like a powder keg ready to explode. 4. I am an even-tempered person? 5. Some of my friends think I 'm a hothead. 6. Sometimes I fly off the handle for no good reason. 7. I have trouble controlling my temper.

Hostility I. I am sometimes eaten up with jealousy. 2. At times I feel I have gotten a raw deal out of life. 3. Other people always seem to get the breaks. 4. I wonder why sometimes I feel so bitter about things. 5. I know that "fr iends" talk about me behind my back. 6. I am suspicious of overly friendly strangers. 7. I sometimes feel that people are laughing at me behind my back. 8. When people are especially nice, I wonder what they want.

0.521 0.853 0.671 0.741 0.617 0.787 0.654 0.757 0.627 0.779 0.670 0.742 0.472 0.882 0.632 0.775 0.660 0.752

0.451 0.893 0.677 0.736 0.540 0.842 0.650 0.760 0.758 0.652

0.576 0.818 0.474 0.881 0.620 0.784 0.643 0.766 0.680 0.733 0.727 0.686 0.804 0.594

0.614 0.790 0.675 0.738 0.622 0.783 0.708 0.706 0.610 0.793 0.357 0.934 0.629 0.777 0.378 0.926

aNegatively keyed items.

Confirmatory factor analysis of The Aggression Questionnaire 993

Table 2. Alpha coefficients for The Aggression Questionnaire scales

Questionnaire scale ~t Number of items Predicted ~b

Physical aggression 0.845 9 0.845 Verbal aggression 0.752 5 0.845 Hostility 0.799 8 0.817 Anger 0.830 7 0.863 Revised hostility" 0.811 6 0.866

~After removing the 2 suspicion items bpredicted based on a 9-item scale

made across the scales and 9 items was selected since the longest scale, physical aggression, contains 9 items.

The results in Table 2 suggest that the revised hostility scale (a = 0.811, 6 items, predicted 9 item at = 0.866) is slightly more homogeneous than the original hostility scale (ct = 0.799, 8 items, predicted 9 item ~ = 0.817), even though 2 items were deleted. Therefore, the above results suggest that the hostility scale may be a more homogeneous scale if the 2 items dealing with suspicion were deleted.

CONCLUSION

The purpose of this paper was to test whether the factor structure of The Aggression Questionnaire (Buss & Perry, 1992) could be replicated on an independent sample using confirmatory factor analysis with maximum likelihood estimates. On examination of the factor loadings and standardized residuals, 2 items dealing with suspicion from the original hostility scale were found to have relatively low factor loadings. Removing the 2 suspicion items from the hostility scale resulted in a slightly higher internal consistency value than was obtained from the original scale, even after estimating alpha values with equal scale length. Based on the results of this study, it is suggested that the hostility scale may be improved psychometrically if the 2 suspicion items were removed from the scale. Following this suggestion, The Aggression Questionnaire may be improved by either remaining with a 4-factor model, with revisions to the hostility factor, as mentioned above, or possibly by developing a separate suspicion factor, as was originally proposed by Buss and Perry (1992), using the 2 suspicion items as a starting point.

Acknowledgements--I would like to thank P.A. Vernon and D. N. Jackson for their comments on this paper.

R E F E R E N C E S

Buss, A. H. & Durkee, A. (1957). An inventory for assessing different kings of hostility. Journal of Consulting Psychology, 21, 343-349.

Buss, A. H. & Perry, M. (1992). The Aggression Questionnaire. Journal of Personality and Social Psychology, 63, 452~,59. De Vellis, R. F. (1991). Scale development. London, UK: Sage. Eysenck, H. J. & Rachman, S. (1966). Dimensions of personality. In Semenoff, B. (Ed.), Personality assessment

(pp. 345-357). Baltimore, MD: Penguin Books. Gorsuch, R. L. (1974). Factor analysis. Philadelphia, PA: (W. B.) Saunders. Harris, J. A. (1993). Salivary testosterone and aggression and pro-social behavior. Unpublished masters thesis, The

University of Western Ontario, London, Ontario. Huesmann, L. R., Eron, L. D., Lefkowitz, M. M. & Walder, L. O. (1984). Stability of aggression over time and generations.

Developmental Psychology, 20, 1120-1134. JSreskog, K. G. & S6rbom, D. (1989). LISREL 7 user's reference guide. Mooresville, IN: Scientific Software. Long, J. S. (1983). Confirmatory factor analysis. Beverly Hills, CA: Sage.