METHODOLOGICAL ISSUES IN ANDROGYNY RESEARCH

PRICE COLEMAN LOCKE, Purdue University

Abstract

This research addressed several methodological issues in the area of androgyny research, including (1) a potential lack of parismony and (2) an unnecessarily restrictive or uniformed consideration of the data. One area of theoretical focus was that of the relationship between sex role identity and field dependence-independence. Another area of theoretical focus was that of the relationship between instrumentality (a facet of masculinity) and extraversion. The subjects, 156 male and 238 female students from introductory psychology courses, were administered the Personal Attributes Questionnaire (PAQ), the Hidden Figures Test (HFT), and the Eysenck Personality Inventory (EPI). The PAQ assesses instrumentality (M) and expressiveness (F); the HFT assesses disembedding ability (a construct related to field dependence); the EPI assesses extraversion (E) and neuroticism (N). Although a sex difference was found in performance on the HFT, no significant association was found with either M or F. As predicted, a positive association was found between E and M. Also there was a positive association between E and F, and a negative association between N and M. In order to assess methodological issues, several methods of analysis were performed on each set of data. Both partialled and unpartialled effects were calculated as estimates of effect size. Analyses were run on both the dichotomized M and F data ("high" and "low" categories) and the original many-valued data. The traditional sex type categorization effects were investigated via two-way analysis of variance, and contrasted with M, F, and M x F effects generated via three-way analyses of variance. As expected, analysis of the original M and F data provided more powerful tests of hypotheses than did analysis of the dichotomous M and F data, and in general partialled effects provided lower estimates of effect size than did unpartialled effects (but the partialled effects were viewed as being more accurate estimates). On the basis of the two types of ANOVA carried out, it was concluded that use of the traditional fourfold typology could lead easily to unparsimonious, misleading results which are difficult to interpret. Other illustrations of this problem with the traditional typology were given, and recommendations were made for future research in this area.

Degree

Ph.D.

Subject Area

Psychology

COinS