A Study to Develop and Validate a Scale to Assess Heterosexuals' Attitudes Toward Lesbian, Gay, Bisexual, and Transgender People

This study reports on the development and validation of a scale to assess heterosexuals' attitudes toward lesbian, gay, bisexual, and transgender (LGBT) people. This study was conducted as part of a campus climate survey at a large public university in the Midwest. Separate measures concerning homosexual, bisexual, and transgender persons are available; however, a measure concerning LGBT people (as a group) is not available. A short, easily implemented scale assessing attitudes toward LGBT people is needed to potentially test educational interventions designed to promote acceptance of LGBT people.

Scale development entailed: (1) Developing 15 initial items (scored 1=strongly disagree, 7=strongly agree) based on literature review and advisory committee input. Scores were summed (after appropriate reverse-scoring) and averaged to create an overall score (higher scores represent more positive attitudes). Sample item: “It is perfectly okay for people to have intimate relationships with people of the same sex.” (2) Administering the scale to heterosexual U.S. students via an online survey (n=1,817, age M=23.84, SD=6.00; 63% female; 62% undergraduate; 78% White). (3) Removal of one highly skewed item. (4) Psychometric analysis of remaining items on a random selection of 50% of the sample (n=898, age M=22.72, SD=5.71; 63% female; 62% undergraduate; 71% White) using Principal Components Analysis (PCA) with an oblique rotation method. Factor loadings were considered acceptable at the .40 level. (5) A confirmatory factor analysis (CFA)utilizing structural equation modeling (SEM) was pursued to examine the latent factor structure of the scale using the remaining 50% of sample (n=887; age M=22.18, SD=5.94; 63% female; 61% undergraduate; 76% White). Reliability testing was conducted using Cronbach's alpha. (6) Examining the external validity of the final scale, we tested for group differences based on sex and exposure to LGBT educational content. We also examined the relationship between religiosity and LGBT attitudes.

The initial PCA produced 3 factors. After items were removed from the scale for double-loading, loading below .40 or for loading with less than two other items, the EFA produced a 9-item unifactorial solution accounting for 51% of the variance. Following minimal model modifications, the CFA indicated a good model fit: x2 (18, p= >.09) = 1.47, RMSEA=.024, GFI=.993, CFI=.997. Due to the presence of multivariate non-normality all results are reported are from a Bollen-Stine bootstrap analysis. The final 9-item scale demonstrated excellent reliability (a=.87).

On average, respondents held accepting attitudes toward LGBT people (M=5.30; SD=1.19). Significant differences between males and females were found (t(1,238)=8.56, p<.001) and between exposure to LGBT course content or not (t(1,297)=-7.50, p<.001). A negative correlation was found between LGBT attitudes and religiosity (r=-.35, p<.01). All of these results were in the anticipated directions.

The final 9-item scale assesses cognitive, affective, and behavioral dimensions of attitudes toward LGBT people. The scale demonstrated sound psychometric properties and can help to improve our understanding of the attitudes held toward LGBT people and be useful when evaluating educational interventions. Further validation of this measure with different groups, such as rural students and non-student populations, would be wise next steps.