Electoral Institutions and the Evolution of Partisan Conventions, 1880-1940

• Author: Bumba Mukherjee,Anna Harvey
• Published date: 01 May 2006
• DOI: 10.1177/1532673X05281487
• Date: 01 May 2006
• Subject Matter: Articles

10.1177/1532673X05281487American Politics ResearchHarvey, Mukherjee / Partisanship Levels

Electoral Institutions and

the Evolution of Partisan

Conventions, 1880-1940

Anna Harvey

New York University, New York City

Bumba Mukherjee

Florida State University, Tallahassee

This article examines variation in partisanship levels across the United States

between 1880 and 1940 and suggests that the introduction of three electoral

laws in this time period—party registration, primaries and secret ballots—can

explain the geographic variation in partisanship levelsacross the United States

during this era. Specifically, the article argues that the introduction of party

registration increased the observability of partisan behavior; and this, in turn,

increased partisanship strength in the states in which party registration was

introduced. Conversely,primaries and secret ballots reduced the observability

of partisan actions, which consequently weakened the level of partisanship.

The authors test their theoretical predictions on aggregate levels of split-ticket

voting across the United States between 1880 and 1940. The authors find con-

siderable support for their predictions in time-series cross-section (TSCS) esti-

mates of the effects of electoral institutions on levels of partisanship across

states, support undiminished after corrections for endogeneity and selection bias.

Keywords: progressive reform; political parties; partisanship; political

party conventions; political machines; political party strength

There has long been a recognition that partisanship in the United States, in

terms of both its strength and its direction, possesses a distinct geo-

graphical component that cannot be accounted for by individual-level vari-

ables (Campbell, Converse, Miller, & Stokes, 1960, chap. 11; Elazar, 1966,

1986; Elazar & Zikmund, 1975; Erikson, Wright, & McIver, 1987, 1993;

Finkel & Scarrow, 1985; Mayhew, 1986, pp. 236-237; Norrander, 1989;

Sharkansky, 1969; Wright, Erikson, & McIver, 1985). Residents of some

368

Authors’Note: We thank Jonathan Nagler, Rebecca Morton, Alastair Smith, and other partici-

pants in workshops at YaleUniversity, University of Wisconsin–Madison, and New YorkUniver-

sityfor helpful comments and suggestions. Any errors are the sole responsibility of the authors.

American Politics Research

Volume 34 Number 3

May 2006 368-398

© 2006 Sage Publications

10.1177/1532673X05281487

http://apr.sagepub.com

hosted at

http://online.sagepub.com

states are significantly more likely to engage in partisan acts than are resi-

dents of other states, even after controlling for the characteristics of state

populations. Similarly, residents of some states are significantly more likely

to engage in acts supporting a given party than are residents of other states,

again even after controlling for the composition of state populations.

Political scientists and historians have recognized that there exists varia-

tion in the strength (i.e., level) of partisanship in the United States (Campbell

et al., 1960; Formisano, 1999; Norrander, 1989; Silbey, 1991). For instance,

in her study of partisanship levels between1976 and 1982, Norrander (1989)

suggests that “the proportion of a state’s residents who consider themselves

to be independent varies vastly across the states” (p. 516). Geographic varia-

tion in the strength of partisanship is not merely restricted to the contempo-

rary era, however.In his work on political parties, the historian Silbey (1991)

claimed that intense “two-party competition existed throughout the nation in

the late nineteenth century . . . yet, antipartyism still lived in several nooks

and corners of the nation” (p. 326). Similarly, historians Altschuler and

Blumin (1997) suggest that although election totals in the late nineteenthcen-

tury reflect deep political engagement in some places, there were other areas in

the post-civil war America where large numbers of voters were “uninterested

in, skeptical about, or even averse to political affairs” (pp. 884-885).

A cursory examination of the strength of partisanship —which, following

standard practice (see, e.g., Burnham, 1965; Rusk, 1970), is operationalized

by our measure of split-ticket voting2—between 1880 and 1940 across 45

states reported in Table 1 confirms the idea that the strength of partisanship

did possess a distinct geographic component during this time period. For

instance, in Table1, we find that the mean level of split-ticket voting between

1880 and 1940 in Illinois, North Carolina, Indiana, West Virginia, and New

Jersey is as low as 2.191, 2.986, 1.158, 1.925, and 2.631, respectively. This

indicates that the level of partisanship wasquite high in these five states dur-

ing this era. The mean level of split-ticket voting between 1880 and 1940 in

Minnesota, Tennessee, Oregon, and Texas was as high as 9.124, 10.447,

12.113, and 9.673, respectively. This suggests that the level of partisanship

was substantially lower in these four states. Difference of means tests con-

firm that the difference in the mean level of split-ticket voting between each

of the states mentioned above in this time period is statistically significant

(p= .000). The statistically significant difference in the mean level of split-

ticket voting across these states from widely different regional locations indi-

cates that there is substantial variation in the strength of partisanship during

the 1880 to 1940 time period.

Although there is little doubt that geographic variation in the level of par-

tisanship is distinct across states during the 1880-1940 era, there is remark-

Harvey, Mukherjee / Partisanship Levels 369