The study investigates a tension between affirmative action scholars who hypothesize that affirmative action bans would increase underrepresented minority students' STEM college completions and scholars who have empirically estimated declines. It also questions results that that find no existence of a STEM penalty (compared to non-STEM) caused by affirmative action bans and finds that a STEM penalty is likely, but varies from a very modest effect (using the Department of Homeland Security taxonomy) to a more substantial impact (using the National Science Foundation taxonomy). This study finds that many of the tensions between affirmative action scholars studying STEM are likely the product of model specification. It illustrates how the following have impacted the estimates: the period studied, the inclusion and exclusion of certain ban states, the measure of institutional selectivity, and most importantly, the taxonomy of STEM. Furthermore, this study builds upon prior studies by analyzing private schools and the shorter-term and longer-term impacts of the bans.