Measuring Congressional Partisanship and Its Consequences

Abstract

We develop a method for measuring a legislator’s partisanship using their Twitter rhetoric. To do so, we classify over 2.1 million tweets sent during two congressional terms (2015 through 2018) to determine how often members use explicitly partisan language. Since lawmakers are strategic in how they communicate with the public, we argue our approach captures a member’s partisan intensity, the time and effort they devote to supporting their party. After validating our measure, we examine how partisanship affects commonly studied legislative behaviors. We show it predicts, independent of ideology, a lawmaker’s party-unity voting and expressed bipartisanship. Additionally, we find that presidential support is principally driven by partisanship, not ideology. Our findings offer two contributions. First, we show that a member’s partisanship, based on how they talk about Democrats and Republicans online, is associated with their legislative behavior. Second, we measure a concept that is difficult to operationalize.