The Rise of Programmatic Partisanship in the United States, 1856-2020

• In Event: New Perspectives on Parties and Partisanship

Abstract

Looking at the 2016 Democratic and Republican Party platforms in the United States, clear differences on issues stand out. These distinctions reflect a programmatic party system--that is, one in which policy positions serve as a key basis of electoral competition. Parties can use various tools to compete with one another, and they do not all involve issue positions. A common alternative to programmatic partisanship is a system based on clientelism, or the exchange of goods, services, or other material benefits (e.g., jobs) for political support. Parties might also rely on charismatic candidates to woo voters. Of course, no system--past or present--relies entirely on one type of appeal. Still, the degree to which parties develop alternative issue positions to compete with one another has become a subject of interest to scholars, particularly in the field of comparative politics.

This work has shown a strong association between programmaticism and development. My book manuscript investigates this association, and the broader question of why countries transition toward issue-based electoral competition, by examining the American party system from 1856 (the first year of competition between Democrats and Republicans) to the present. To date, most work on programmaticism has been cross-national, focusing especially on new democracies and developing nations in the contemporary era. We can gain new leverage over the causes of programmaticism by examining a single case over time, following American parties as the nation grew from a relatively new, embattled democracy with an industrializing economy and largely clientelistic parties to an advanced post-industrial democracy with one of the most programmatic party systems in the world.

The chapter to be presented here develops and validates the first measure of programmatic partisanship in the United States from 1856 to the present. This is one of my project’s central contributions. A cross-national measure developed by the Democratic Accountability and Linkages Project, while invaluable, covers a limited time period (2008-9). Standard measures of polarization (e.g., the difference between the ideologies of the median Democrat and Republican in Congress) often serve as proxies for programmaticism, but work by Frances Lee (2016) cautions against this practice, as she shows that large differences between the DW-NOMINATE scores of Democrats and Republicans in Congress did not reflect programmatic distinction in the Gilded Age. The Comparative Manifestos Project tracks party positions over time, but these data only go back to 1945.

I take a novel two-stage approach to measuring programmaticism, leveraging the power of machine-learning tools to identify systematic patterns in large volumes of text. First, I estimate a structural topic model (STM) on party platforms over time at the sentence level, using word co-occurrence to identify topics in the corpus (i.e., pool of textual data) and calculate the degree to which each document (i.e., platform sentence) relates to each topic. I then estimate party ideologies in each year overall and within issue areas using Wordfish (Slapin and Proksch 2008), a scaling technique shown to produce reliable ideology estimates on German platforms (Ibid.; Grimmer and Stewart 2013), and take the difference between the parties’ scores as a measure of programmaticism.

This measure underscores Lee’s (2016) findings with respect to the Gilded Age as well as John Gerring's (1998) classic work on party ideology, and allows us to place their analyses within a broader context. Consistent with Gerring’s argument that parties have always displayed some ideological distinction, I find that there are differences between the parties’ platforms on the whole, as well as within various issue areas, over this entire period. However, these differences have not been constant; rather, their magnitude has varied greatly over time. There is an increase in programmaticism beginning around the turn of the nineteenth century, but it is far less dramatic than the sharp rise that began in the 1960s. And in between, during the mid-twentieth century, the trend flattens.

These findings offer important and novel insights into programmaticism’s trajectory over time, and also raise questions. Why has programmaticism risen to historic heights over the contemporary period? And why wasn’t the earlier, if less steep, rise sustained? This question is particularly puzzling, given what we know about the relationship between programmaticism and development. The United States essentially goes from “late bloomer” to “overachiever”, relative to expectations based on its level of economic development. These questions will be addressed in later chapters of the book.

Author

• Katherine Krimmel, Barnard College, Columbia University