This paper asks how doctoral students understand the use of race variables in statistical modeling. More specifically, it examines how doctoral students at two universities are trained to define, operationalize, and analyze race variables. The authors interviewed students and instructors in addition to conducting a document analysis of their texts and syllabus. They interviewed six advanced doctoral students from each university who study education and have taken an advanced statistics course, defined as regression analysis, multi-level modeling, or other advanced courses. These students were randomly selected from the pool of doctoral students who have completed their course work. They interviewed three instructors from each university who teach advanced statistical methods. Again, the process for selection was random. They collected the syllabus and texts used by all interviewees in their advanced statistics courses. Using responses from the interview questions and short task the authors uncovered student voices on the prevalence and relevance of investigator perspective in race-related quantitative research personally and in terms of their training. In general, participants misinterpreted the output, employing causal language and incorrectly attributing characteristics to racial group membership. Participants also noted receiving very little, if any mention of the complexity of race as a social construct in their statistical training courses. Neither the syllabus nor the text used in these courses spent adequate space and time exploring the issues associated with using racial variables in statistical modeling.