Introduction
The Bem Sex Role Inventory (BSRI; Bem, 1974) explores and measures the concept of androgyny using two factors: masculinity and femininity. The instrument has three subscales to measure the construct of androgyny: masculinity, femininity and social desirability (Bem, 1974). The original scale had 20 items each for these three subscales for a total of 60 items. Bem (1979) reduced the number of scale items to 10 for each subscale, maintaining two factors. Other researchers found four factors and debated factor analysis accuracy (see Pedhazur & Tetenbaum, 1979; Locksley & Colton, 1979; Bem, 1979). Additionally Spence & Helmrich, 1979 debated if the BSRI was the most accurate instrument to measure androgyny (i.e. PAQ versus BSRI). …show more content…
The original form consisted of 60 items divided into three subscales that measured a person’s masculinity, femininity, and a social desirability scale. The social desirability scale items were judged and statistically shown not to load on either masculinity or femininity (Bem, 1974, 1977). BSRI critics focused on dichotomous factor structure validity proposed by Bem to measure androgyny (seen as a high score on both the masculinity and femininity subscales) with little to no focus on social desirability ratings. (Spence & Helmrich, 1979; Locksley & Colten, 1979). Pedhazur and Tetenbaum (1979) analyzed the social desirability scale neutrality and suggested that some items were not neutral. Additionally, findings suggested that the neutral constructs ambiguous definition gave the social desirability scale a “dubious nature” that could preclude a so-called neutral item from actual neutrality and possibly, with more concrete definition, could lead to item loading significantly onto the femininity or masculinity constructs (Pedhazur and Tetenbaum (1979) p.1007). In response to criticism, Bem shortened the form to 10 items per measure and added a fourth outcome dimension (Bem, …show more content…
Despite initial and continuing two-factor solution analysis by Bem (1974, 1977) other researchers proposed a four-factor solution. This is an important component of social desirability scale research as this debate led to the mixture of subscale traits and the four-factor solution development (Pedhazur & Tetenbaum, 1979). For example, Martin and Ramanaiah (1988) suggested that the short form was better than the long form for both the two- and four-factor solutions. Additionally, Brems and Johnson (1990) extracted a four-factor solution labeling the new scales Interpersonal Sensitivity and Interpersonal Potency. Although it appears that the social desirability scale was not used in the factor analysis, Brems & Johnson (1990) do provide a table that includes how many times neutral items (as well as the masculine and feminine items) were extracted from previous factor analysis. Research expanded outside of the two and four factor model suggesting a model based on seven first-order factors that were similar in loadings to male and female, as well as two second-order factors (Blanchard-Fields, Suhere-Roussel & Hertzog, 1994). Attempting to bridge the gap between conflicting two- and four-factor findings, Campbell, Gillaspy and Thompson (1997) conducted CFAs using different populations. The findings suggested that the short form was the more useful form for factor
The Bem Sex Role Inventory (BSRI; Bem, 1974) explores and measures the concept of androgyny using two factors: masculinity and femininity. The instrument has three subscales to measure the construct of androgyny: masculinity, femininity and social desirability (Bem, 1974). The original scale had 20 items each for these three subscales for a total of 60 items. Bem (1979) reduced the number of scale items to 10 for each subscale, maintaining two factors. Other researchers found four factors and debated factor analysis accuracy (see Pedhazur & Tetenbaum, 1979; Locksley & Colton, 1979; Bem, 1979). Additionally Spence & Helmrich, 1979 debated if the BSRI was the most accurate instrument to measure androgyny (i.e. PAQ versus BSRI). …show more content…
The original form consisted of 60 items divided into three subscales that measured a person’s masculinity, femininity, and a social desirability scale. The social desirability scale items were judged and statistically shown not to load on either masculinity or femininity (Bem, 1974, 1977). BSRI critics focused on dichotomous factor structure validity proposed by Bem to measure androgyny (seen as a high score on both the masculinity and femininity subscales) with little to no focus on social desirability ratings. (Spence & Helmrich, 1979; Locksley & Colten, 1979). Pedhazur and Tetenbaum (1979) analyzed the social desirability scale neutrality and suggested that some items were not neutral. Additionally, findings suggested that the neutral constructs ambiguous definition gave the social desirability scale a “dubious nature” that could preclude a so-called neutral item from actual neutrality and possibly, with more concrete definition, could lead to item loading significantly onto the femininity or masculinity constructs (Pedhazur and Tetenbaum (1979) p.1007). In response to criticism, Bem shortened the form to 10 items per measure and added a fourth outcome dimension (Bem, …show more content…
Despite initial and continuing two-factor solution analysis by Bem (1974, 1977) other researchers proposed a four-factor solution. This is an important component of social desirability scale research as this debate led to the mixture of subscale traits and the four-factor solution development (Pedhazur & Tetenbaum, 1979). For example, Martin and Ramanaiah (1988) suggested that the short form was better than the long form for both the two- and four-factor solutions. Additionally, Brems and Johnson (1990) extracted a four-factor solution labeling the new scales Interpersonal Sensitivity and Interpersonal Potency. Although it appears that the social desirability scale was not used in the factor analysis, Brems & Johnson (1990) do provide a table that includes how many times neutral items (as well as the masculine and feminine items) were extracted from previous factor analysis. Research expanded outside of the two and four factor model suggesting a model based on seven first-order factors that were similar in loadings to male and female, as well as two second-order factors (Blanchard-Fields, Suhere-Roussel & Hertzog, 1994). Attempting to bridge the gap between conflicting two- and four-factor findings, Campbell, Gillaspy and Thompson (1997) conducted CFAs using different populations. The findings suggested that the short form was the more useful form for factor