As challenges to race-conscious admissions policies are, once again, advancing through the federal courts, research proclaiming to identify the wideranging effects of affirmative action across a variety of educational settings is influencing this litigation through amici and expert testimony. It is crucial, then, that empirical research used to support claims by parties on either side of the affirmative action debate adhere to the fundamental precepts of causal inference. Yet the literature on causal inference is both vast and dense, and as a result, many judges, lawyers, legislators, and laypersons interested in understanding both the intended and unintended consequences of affirmative action are ill-equipped to understand the debate—especially when quantitative social scientists on both sides of the issue appear to draw conflicting (though not necessarily equally credible) inferences from the same data. The purpose of this Article is to lay bare the core requirements of credible causal inference to the uninitiated, highlighting how inattention to (and sometimes outright disregard for) these rules has muddied the debate over the effect of affirmative action in law schools and in college admissions more generally. The Article empirically examines the six primary deficiencies impacting extant research on affirmative action in law schools: (1) posttreatment bias, (2) nonresponse bias, (3) omitted variable bias, (4) interpolation bias, (5) extrapolation bias, and (6) measurement error bias. I conclude the Article by describing what a scientifically defensible examination of the effect of affirmative action in legal education with currently available data would entail. While no approach to causal inference is infallible, the careful analyst can attempt to ameliorate the impact of these biases.