City of Binghamton, Estados Unidos
Scholars commonly use a measure of the discrepancy between party vote shares and seat shares (observed disproportionality) as a proxy for the effects of electoral institutions. We illustrate the problems with doing so and, instead, recommend that scholars use more direct measures of institutional characteristics. Conceptually, we demonstrate that observed disproportionality cannot accurately capture institutional effects. Empirically, we show that (1) the variance of disproportionality is much higher when electoral rules are restrictive than when they are permissive, (2) the conclusions we draw about the effects of observed disproportionality differ substantially across samples of elections, and (3) replacing measures of observed disproportionality with more direct measures of electoral system characteristics such as district magnitude produces different and more reliable results.