homonegativity, inadvertent focus on racial differences could probably be better discussed as a product of religious affiliation and dogma, even while controlling for the effects of gender role ideologies. Nonetheless, a discussion of interaction/nonadditivity may help to better understand this relationship. Nonadditivity

      Finally, multivariate analysis took into consideration nonadditive relationships among the independent variables. Analysis of nonadditive relationships helps to better understand interactive effects among variables (Table 4-7). Rather, non-consideration of interactive models assumes identical change for one variable over the variability of another. It would predict, for example, that males and females experience the same predicted change in homonegativity as level of education increased. In theory, either males or females may be more sensitive to increases or decreases in education and this can be reflected in predicted homonegativity. Nonadditivity in this study will give consideration that males may differ from females just as blacks may differ from whites across the variation of other independent variables. Models 11 through 14 take into consideration the interaction between both race and gender on other independent variables. Models 15 and 16 take into consideration the interaction between race and gender on the gender role scale.