UC Santa Barbara

This dissertation consists of three papers. The first is conceptual paper concerned with creating tenets for QuantCrit research titled the Anti-Racist Tenets for Analysis (ARTA), discussing each tenet’s meaning and greater application, as well as providing research reflection questions for each tenet. Researchers may use these reflections along with the general meaning of each tenet to ensure their own research is conducted in a transparent and critical way, considering race and the social impact race has both on variables and equity within their studies. The second paper is an empirical application of the established anti-racist tenets for analysis to quantitative research. Four different ways of including race in quantitative analysis are examined. The first is a traditional multivariate approach with one racial category (White) excluded as a reference group. The second is a moderation approach where race is interacted with a number of other independent variables to represent the pervasiveness of race to lived experiences. The third is a path analysis structural equation model where the dependent variable is regressed on the independent variables which are all in turn regressed upon race, and the fourth is a grouping analysis where each racial group’s analysis is run independently. Results found that traditional multivariate and moderation approaches had the same results, suggesting moderation did not add significantly to the model. The path analysis was found to be more exemplary of the pervasiveness of race and intersectional identities than the first two models, and to be sensitive enough to be useful when race is not a central point of study. However, the grouping model was the only truly anti-racist method, as it eliminated the point of comparison between races as well as exemplifying more clearly the lived experiences of each race as well as intersectional identities. The third study tackles equity between factor analysis groups, by showing how to gender run a partial invariance model for two groups, in this instance separated by gender. While factor analysis is a useful tool, invariance can be difficult to find, and this can limit the practicality of this method for many applied researchers. This paper demonstrates how to run a partial invariance model from an exploratory factor analysis on through determining which parameters are held invariant and which are allowed to be freely estimated. This opens up this method to applied researchers who want to apply this method to multiple groups.