Thanks are due to Brian Posler and Allison Rinden, both of whom did some of the grunt work associated with assembling the data used in this paper. The National Science Foundation is not reponsible for any of the conclusions reached in this research. Paper prepared for the Annual Meeting of the American Political Science Associaton, Washington, D.C., August 28-31, 1997.
Explanations for the decline of partisanship in the early part of the 20th century are at odds. The received wisdom holds that a set of electoral reforms led Congressmen to break their partisan ties, engaging in more familiar modes of personalistic behavior. This view has recently been challenged noting that the bulk of the reforms passed in the Populist and Progressive periods eliminated factional strife within parties and led to increased partisanship. This paper looks at a wide variety of reforms introduced in a 40 year period. While a set of early ballot reforms did result in increased levels of state delegation partisanship, subsequent reforms, combined with the passage of time, undermined partisan strength in the U.S. House of Representatives.
Introduction The ebb and flow of party strength remains a puzzle for Political Scientists. This is especially true for partisanship within the U.S. Congress. At the outset of the twentieth century levels of party cohesion were extremely high, but steadily declined in the subsequent two decades. Reaching a low water mark in the 1960s, party cohesion resurged in the 1980s, with measurable levels of party voting approaching that observed in the heyday of "boss rule." These swings in Congress have been well documented (Brady, Cooper and Hurley, 1979; Hurley and Wilson, 1989; Patterson and Caldeira, 1988). While the resurgence in the 1980s has been explained by many (in particular, see Rohde, 1992), our explanations remain incomplete.
Key to understanding partisan strength in the U.S. Congress is explaining the precipitous decline of partisanship at the turn of this century. Numerous explanations have been offered, and one which has gained widespread currency points to fundamental changes in electoral rules. In particular, ballot reform rules that were adopted to overcome local party factionalism in the 1890's and Progressive reforms designed to both expand and constrict electoral participation, were thought to undermine partisan linkages between the electorate and those holding office. In turn, it is argued that the electoral linkage was transformed such that office holders became more closely tied to their constituents. The net effect was a slow erosion of party controls over the behavior of office holders.
This paper first details the current explanations for partisan decline around the turn of the 20th century. While there is a standard story focused on the role of rules, recently this view has been challenged (Reynolds, 1995, 1997). In the second section several hypotheses are offered that are derived from these competing approaches. The third section details a set of reforms that dominated the political landscape during the period 1876 to 1932. Finally, a set of models estimate the impact of these reforms on levels of partisanship among Congressional state delegations throughout this period.
Explanations for Partisan Decline in the "Golden Age of Party”
The dominant explanation for the decline of partisanship at the turn of the twentieth century focuses on changes in the electoral rules pressed by Populists and Progressives.1 By changing the "rules of the game" reformers broke down the ability of party bosses to control elections. Weakened partisan bonds meant that elected officials were now more closely attached to constituents, and constituent interests often diverged from purely partisan politics. Consequently elected officials were less likely to toe a party line and observable measures of partisan attachment were likely to decline. A number of studies have shown that changes in electoral rules, whether it be the introduction of the Australian ballot (Rusk, 1970) or the direct primary (Galderisi and Ginsberg, 1986) were correlated with a decline in partisan attachment. Katz and Sala (1996) go so far as to argue that the introduction of the secret ballot transformed the institution of Congress, cementing into place a norm in which re-elected members had a property right claim to committee appointments.
It has been proposed that each change in the rules had an impact on eroding the control of party bosses. As Galderisi and Ginsberg (1986) argue, the direct primary was one of the more interesting of the reforms introduced during the Progressive period.
"The primary can be seen as an antiparty reform on three separate counts. First, by weakening party leaders' capacity to control nominating processes, primary elections undermine the organizational coherence of established parties. Second, primaries tend to direct the attention of voters and political activists toward the nominating contests of the party most likely to win the general election, and away from the interparty race. ... Last, and most interesting, primary elections have the effect of inhibiting the formation of new parties." (p. 116)
The argument for the latter point is that contested primaries allow dissidents to oppose within established parties rather than challenge from the outside. Consequently, attention turns away from building a new third party in order to opt to work from within.
The principal evidence offered for the impact of changing electoral rules has focused on the Australian ballot. The argument typically offered holds that the secret ballot directly removed the ability of partisan bosses to control votes. Prior to the introduction of the secret ballot, parties (or even splinter groups) printed their own ballots which were handed to voters and which could be dropped into a ballot box. Because only one such voting slip could be dropped into the ballot box, this required that someone watch to prevent more than one vote being cast. Obviously, because ballots were distinctive, this also allowed partisan observers to monitor who was for and against them. The Australian ballot, which required that the state or local government print a standard ballot and required voters to mark their ballot, ensured secrecy when casting the ballot and removed partisan monitoring. If a voter was "bought" it was no longer clear whether he stayed that way when casting his vote.
Recently this view has been challenged. Reynolds (1995) points out that ballot reforms were aimed at curbing abuses created by urban political machines. Rather than behaving as tightly controlled national political parties these local machines were fractionated and factional. Reynolds' claim is that the reformist movements were led by state party leaders who viewed these reforms as an opportunity to retrench the two major parties in the political process. By requiring a centralized, standardized ballot, the major parties could squeeze out third party competition and they could curtail factions within their own party. Indeed, Reynolds provides very strong evidence that the introduction of the Australian ballot and mechanisms for voting via party list led to increased cohesion within the parties -- particularly at the national level (see Reynolds, 1997).
Reynolds may well be correct in arguing that the net effect of populist (and later progressive) reforms strengthened national political parties, rather than undermined them. However, the Australian ballot was only one of many reforms pursued by reformers. As Burnham (1970), among many others, notes the reform movements that gained steam in the cities in the 1890s pressed for a wholesale transformation of the American electoral system. At least two general sets of reforms were sought. The first tried to undermine urban political machines by implementing the Australian Ballot, direct primaries and at-large, non-partisan elections. The net effect of these major reforms (and a host of minor experiments) was, in Burnham's mind, to erode party functions. In turn, as Rusk (1970) and others claim, this weakened the grip of Congressional partisans in both Chambers.
The second set of reforms, rather than breaking down control by party elites, turned to curbing electoral abuses by citizens. These reforms included allowing women's suffrage, thereby enfranchising a group that was expected to have a leavening effect on the political process, the introduction of poll taxes and other barriers to voting, effectively disenfranchising a different group and lessening competition in the South, and finally by requiring personal registration of voters to limit the amount of fraud due to individuals casting multiple ballots.
Burnham is quite deliberate in his conclusions about the consequences of these reforms. As he argues, it was not one or the other of these reforms that lead to the decline of partisan attachment at the outset of the 20th century. Instead, he notes
"...most if not all of these fundamental changes in the 'rule of the game' were in effect devices of political establishment and control, with strongly conservative latent consequences if not overt justifications, and with an overwhelmingly antipartisan bias." (p. 74).
Burnham is only echoing the arguments offered by many Progressives at the turn of this century. For example, Ernst Mayer, in a self-published treatise advocating the primary system as a means for wresting control from partisan bosses, argued
"Closely related to the Australian ballot and the primary election reforms safeguarding our nominations and elections, is that which aims to punish the corrupt use of money and other forms of bribery in securing either nominations or elections to office. These reforms are all complementary. Neither one is complete in itself, and all cooperate for a common end, -- the establishment of more perfect institutions for the selection of our public service." (1902, p. 444).
Changing a single rule of the electoral game was not regarded as sufficient. Instead, a galaxy of electoral reforms were necessary in order to return to political order (of course, see Wiebe, 1967 for a trenchant analysis as to whose order was being restored).
Regardless of the reformist claims, perhaps there is truth in the matter that an array of reforms, which were implemented at different times, had an impact on voter's partisan attachments and consequently on the attachments of members to party-in-the-Congress. Certainly Rusk and Stucker (1978) provide strong evidence for at least the first part of this equation in southern states. They analyze the confluence of a large number of electoral reforms and links their impact to Congressional state delegations. Rather than resting with Reynold's conclusions about the impact of the Australian ballot on cohesion among partisans in Congress, I press as to whether the change in the ballot, coupled with a myriad of other reforms, had an impact on shifting levels of cohesion in the U.S. Congress.
Hypotheses The principle hypothesis is very straightforward. The cumulative effect of ballot and electoral process reforms was to break down levels of cohesion in state congressional delegations. While several of the ballot reforms in the 1890's may have boosted state party control over the nomination and election of members to the U.S. Congress, over time that control shifted from the party to the individual representatives. With the addition of Progressive reforms that controlled access to the ballot and enhanced the ability of voters to mask their partisan allegiance, representatives increasingly had to tie their fortunes to the voters and were less bound to the dictates of party interests. This meant that where the electoral rules provided new incentives to respond to voters, members of congress did so. All of this occurred at the same time that the tenure for members of congress was increasing.
Hypothesis 1. The net effect of ballot and electoral reforms was a decline in state delegation partisan attachment.
If the hypothesis is incorrect then the reforms should have no independent effect. Instead, the observed shift in aggregate partisan attachment is due to some other combination of variables. Because these reforms were adopted at different times in different states, the effects of different reforms will be negligible.
Hypothesis 2. The short term effects of ballot reforms was to strengthen state delegation partisan attachment.
Hypothesis 2 extends directly from the discussions by Reynolds (1995, 1997). Her arguments and data point out that fundamental changes in the ballot structure meant increased control by state party leaders and eliminated the factionalism that had often been present in parties. However, these reforms, while eliminating factional strife within the party, also planted the seeds for individual candidates to build their own voting coalitions. If similar findings are not uncovered here, then the measures I use can be called into question.
Hypothesis 3. The short and long term effects of Progressive electoral system reforms was to weaken state delegation partisan attachment.
Hypothesis 3 distinguishes between ballot reforms and electoral reforms. The latter aimed to eliminate electoral fraud and to "cleanse" the pool of eligible voters. This in turn decreased the ability of partisan machines to control who was placed on the ballot and led to a steady decline in partisan attachment.
Fortunately, from the standpoint of a social scientist trying to disentangle these different effects, not all reforms were adopt in all states nor were they adopted at the same time. As a consequence, by taking a lengthy time series and tracing when reforms were differentially implemented, it is possible to tease out the impact of these different reforms and address these different hypotheses.
Data The data used for this analysis cover the period 1876 (when members of the 45th Congress were first elected) to 1932 (when members of the 73rd Congress were elected). Two distinct types of data were used. The first constitute measures of partisanship within the U.S. House of Representatives. These are recovered from recorded votes provided by the ICPSR. The second type of data concerns the electoral rules adopted by each state during this period.
The key dependent variables are derived from measures of party voting in the US Congress. Two types of measures are used in order to tap levels of delegation cohesion. The first is a standard measure that often has been used. The PARTY VOTING measure is based on the percentage of the time that 50 percent of one party voted against 50 percent of the other party.2 As such this measure simultaneously (but imperfectly) captures inter-party conflict as well as the degree to which parties cohere. Usually this measure is reported in the aggregate for the Congress. In this analysis, however, the focus is with state delegations. For each state an average Party Voting score was calculated for each party. The score was derived from the percentage of times individual members voted with their party on party votes. Someone who rarely voted with the party majority when the parties were in conflict would have a low score (a score below 50 was quite rare), while someone always backing the party would have a score approaching 100. Consequently, highly partisan delegations had average scores approaching 100 and weakly partisan delegations had lower scores. Figure 1 shows the overall party voting averages, by party and year, for the House. What is apparent is that the overall rates of party voting are high for both parties. They climb during the 1890s and then begin to decline in the second decade of the 20th century.