(M= 2.30, SD = 1.17) reporting slightly greater religiosity than women (M= 2.03, SD = 1.10).
Gender accounted for approximately 1% of variability in these data. Parallel results were found
when gender comparisons were made with transgender persons excluded from the analyses;
again, the overall model was significant, but no gender effects were significant with alpha
adjusted to .01, and only psychological well-being yielded a significant gender effect with alpha
at .05, again with women scoring higher than men. Thus, overall, gender differences on the
dependent variables were non-significant or negligible. As such, hypotheses were tested with the
entire sample, and without gender as a covariate.
Test for Order Effects
To test for order effects across the two orders of the survey, a MANOVA was conducted
with survey order as the independent variable and the variables of interest (i.e., psychological
distress, psychological well-being, experiences of prejudice, expectations of stigma, internalized
homophobia, concealment of sexual orientation, spirituality, and religiosity) as dependent
variables. Box's test of equality of covariance matrices and Levene's test of equality of error
variances were not significant indicating that the data met assumptions of homogeneity of
covariance matrices and variance. The overall model was significant (F [1, 384] = 2.25,p <.05,
r2 = .05) suggesting a significant but small order difference in the set of dependent variables.
Again given the number of comparisons being conducted, a more conservative alpha of .01 was
used for follow-up univariate analyses. These analyses indicated that there were no significant
order effects at thep = .01 level. At the less conservative = .05 level, only religiosity yielded a
significant order effect (F [1, 384] = 5.42, p < .05, r/p = .01), with the order effect accounting for
approximately 1% of variance in the data. Thus, overall, order effects on the dependent variables
were non-significant or negligible.