We study precision and bias in the US electoral college system, studying the electoral college as a statistical estimator. We take an historical view, documenting long-lived patterns that have persisted beyond the last century. We examine the last stable party system prior to the Civil War (1832-1854), as well as the party system in during reconstruction (1876 and 1892) and in the most recent period (1988-2016). The particular parties, the political alignment of states, and indeed even the roster of states varied greatly across these periods. This is useful in disentangling whether the bias and precision properties of the EC are fundamental, or happenstantial--depending, for example, on the particular political alignment of states. Hispanics, who tend to reside in large states, have been particularly underrepresented in EC votes for at least several decades. We show that winning the electoral college but losing the popular vote is not an unlikely anomaly.
Presented in Session 8. Economy, Labor Force, Education, & Inequality