During U.S. presidential and congressional elections, a great deal of attention is often paid to political partisanship (see Avalon 2012; Sides 2008). How did political independents vote? Did partisans vote how they were "supposed to"? Did partisanship exert a larger (or smaller) impact on vote choice during this election than the previous one? Within the context of a given election, the answers to these questions can lead to interesting insights on electoral behavior. Analyzing the impact of partisanship on voting over the course of many elections, however, has the potential to shed light on the general relevance of parties to voter decision making. Although partisanship is one of the most enduring political orientations (Campbell et al. 1960), its impact on the vote has fluctuated considerably over time (Bafumi and Shapiro 2009; Bartels 2000). During some elections, partisanship exerts a powerful influence on the vote, while during others, partisanship has a much less pronounced effect. Although partisanship is virtually always relevant to voting behavior during U.S. presidential (and congressional) elections, the importance of parties and partisanship has varied over time (Barrels 2000). The 2008 presidential election and Barack Obama's candidacy sparked discussion about the potential for an era of "post-partisan" politics, but many commentators have suggested that 2008 was "very much a partisan, not a 'post-partisan' election" and that party loyalties were "as strong as ever" (Sides 2008). In this article, I measure the extent to which partisanship influenced voting behavior in U.S. presidential and congressional elections from 1952 to 2008. Such a long-term analysis provides an opportunity to compare the influence of partisanship from one election to the next and to compare the effects of partisanship on voting in contemporary elections to the effects of partisanship in other eras in American politics.
One issue confronting those who are interested in examining the effect of partisanship on voting behavior across multiple elections is how to measure the extent of partisan voting. In a 2000 article, Bartels develops a "summary measure of partisan voting," which takes into account the distribution of partisanship in the electorate and the extent to which partisanship influences voting behavior. The aim of this measure is to document changes in the impact of partisanship on voting behavior that have occurred over time. In this article, I replicate and update Bartels' (2000) analysis of the impact of partisanship on voting behavior in U.S. presidential and congressional elections, which extended from 1952 to 1996, with the goal of understanding how the impact of partisanship on voting behavior has changed since the 1996 elections. I use American National Election Studies (ANES) data from 1952 to 2008. Following Bartels, I use a series of simple probit models to examine the effect of partisanship across U.S. elections. I show that Bartels' findings nicely replicate and that between 1952 and 2008, the high point of partisan voting in presidential elections occurred in 2004. The level of partisan voting in 2004 was 4% higher than in 1996--the apex of partisan voting in Bartels' original analysis. The level of partisan voting in the 2008 presidential election was quite high, although it was slightly lower than in 2004, 2000, and 1996. I also examine the extent of partisan voting in U.S. congressional elections. I find that the level of partisan voting in congressional elections has remained high but reached a peak in 2006.
Measuring Partisan Voting
Before describing the approach I use to measure the level of partisan voting in a given election, it is important to provide a bit of background on parties and partisanship in the United States. In 1960, the authors of The American Voter (Campbell et al. 1960) convincingly argued that partisanship was a long-term, affective attachment to a political party--one that developed early on in peoples' lives. Importantly, The American Voter also showed us that partisan attachments play a central role in American voting behavior. Interestingly, during the 1970s, there was an increase in the number of political independents in the electorate, and many political observers viewed the increase in independents as a signal that parties and partisan loyalties were becoming less and less relevant to American politics (Burnham 1989; Smith 1988; Wattenberg 1996), an argument that came to be known as the "party decline thesis." Skeptical of this idea, Bartels used data from the ANES to measure and track levels of partisan voting over time. Barrels found that the impact of partisanship on voting behavior in presidential elections steadily increased from 1972 to 1996 and that the level of partisan voting was about 77% higher in the 1996 presidential election than in the 1972 election (the low point of partisanship in Bartels' analysis). He also found evidence of a resurgence in partisan voting in congressional elections, although levels of partisan voting never reached the levels they were at during the 1950s. This led Bartels to conclude that, contrary to the thesis of party decline, "the American political system has slipped with remarkably little fanfare into an era of increasingly vibrant partisanship in the electorate" (2000, 44). Given that the endpoint of Bartels' analysis is 1996, one question that remains is how the relationship between partisanship and voting behavior changed over the past decade or so. Has the...