Examining the Factor Structure of the Personality Attributes Questionnaire

Abstract

This study uses confirmatory factor analysis to assess the factor structure and unidimensionality of the Personality Attributes Questionnaire (PAQ; Spence, Helmreich, & Stapp, 1974). The results suggest that neither the original 24-item PAQ scale (Spence et al., 1974) nor the 18-item PAQ scale proposed by Ward, Thorn, Clements, Dixon, and Sanford (2006) is satisfactory. This revised scale was subjected to tests of measurement invariance and examined for its predictive utility in sex classification. Findings confirm Ward et al. The PAQ–MF (masculine or feminine traits) was identified as a statistically reasonable substitute for the PAQ–M (masculine traits) when either unidimensionality or reliability is not possible for all factors. However, the lack of unidimensionality and reliability in the metric begs the question of whether the instrument has utility in gender assessment at all.

Keywords:

• Confirmatory Factor Analysis
• Gender
• Measurement
• Personality Attributes Questionnaire

This revised manuscript has been presented at the National Communication Association's annual conference in San Diego, CA. This manuscript has not been published, and it has not been submitted simultaneously for publication elsewhere.

Notes

Note. Item titles have been abbreviated for presentational purposes. All factor loadings are significant at the p < .05 level. Original model N = 259; Ward et al. model N = 264; and final model N = 270. PAQ = Personality Attributes Questionnaire; M = masculine traits; F = feminine traits; MF = both masculine and feminine traits.

Note. Item titles have been abbreviated for presentational purposes. Female model, χ²(62, N = 128) = 80.15, p = .06 (comparative fit index [CFI] = .95, root mean square error of approximation [RMSEA] = .048, and standardized root mean square residual [SRMR] = .066; male model, χ²(62, N = 142) = 141.74, p < .05 (CFI = .816, RMSEA = .096, and SRMR = .085). PAQ = Personality Attributes Questionnaire; M = masculine traits; F = feminine traits; MF = both masculine and feminine traits.

^a n = 128.

^b n = 142.

*p < .05.

Note. M = masculine traits; F = feminine traits; MF = both masculine and feminine traits.

^a n = 128.

^b n = 142.

And, potentially, Items 5 and 10, which were reliable and unidimensional, yet demonstrated some—although minimal—measurement differences between men and women.

Additional information

Notes on contributors

Michael D. Curran

Michael D. Curran (PhD, University of Arizona, 2008) is Director of Advertising Analysis at Ace Metrix, Los Angeles, CA.

Katie M. Warber

Katie M. Warber (PhD, University of Arizona, 2007) is an Assistant Professor in the Department of Communication at Wittenberg University, Springfield, OH.