Ideology and Congress

A Political Economic History of Roll Call Voting

Howard Rosenthal

• Publisher: Routledge
• Year: 2017
• ISBN:
  9781351513784
• Edition: 2
• Topic: Politics & International Relations
• Subtopic: Politics
• Index: Politics & International Relations

In Ideology and Congress, authors Poole and Rosenthal have analyzed over 13 million individual roll call votes spanning the two centuries since Congress began recording votes in 1789. By tracing the voting patterns of Congress throughout the country's history, the authors find that, despite a wide array of issues facing legislators, over 81 percent of their voting decisions can be attributed to a consistent ideological position ranging from ultraconservatism to ultraliberalism. In their classic 1997 volume, Congress: A Political Economic History of Roll Call Voting, roll call voting became the framework for a novel interpretation of important episodes in American political and economic history. Congress demonstrated that roll call voting has a very simple structure and that, for most of American history, roll call voting patterns have maintained a core stability based on two great issues: the extent of government regulation of, and intervention in, the economy; and race. In this new, paperback volume, the authors include nineteen years of additional data, bringing in the period from 1986 through 2004.

1

Introduction:
The Liberal/Conservative Structure

Roll Call Voting and the Liberal/Conservative Continuum

“All politics are local,” said Tip O’Neill, Speaker of the House from 1977 to 1987. If only because the Congress of the United States must amalgamate the diverse preferences of constituencies,¹ the task of finding a simple structure to explain how members of Congress vote when the roll is called might well be a hopeless one. Perhaps only detailed accounts of action on particular bills can fully capture the legislative process. But any science of politics must, on the contrary, seek to find simple structures that organize this apparent complexity. We have developed a parsimonious model that accounts for the vast majority of millions of individual roll call decisions during the 217 years of roll call voting in the House of Representatives and the Senate. The spatial model of voting, outlined in chapter 2, is the technical foundation of our model. For short, we refer to it as the ideological model.

Aggregating the local loyalties of members of Congress into legislation is a matter of solving an institutional labyrinth replete with committees, subcommittees, and conference committees.² The outcomes reflect not only the preferences of the legislators themselves, but the pressures and appeals of countless staff, lobbyists, and constituents. Moreover, activity in Congress will be responsive to the veto power of the president. One might expect chaotic rather than orderly behavior.³

When, on the other hand, local concerns give way to a disciplined two-party system, day-to-day roll call voting is devoid of interest. A stylized description of Great Britain, for example, would indicate that national elections create a parliamentary majority.⁴ Until the next elections, the winning party proposes legislation. The legislation is routinely approved by all members of the majority and opposed by all members of the minority.⁵

Such a model obviously doesn’t work on our shores. President Reagan was able to enact his economic program in 1981, including a large tax cut and cuts in domestic spending, in spite of a divided government, with the Democrats being in the majority in the House of Representatives. Defections of “Boll Weevil” Democrats from the South eroded the majority. Even in the disciplined world of Tom “The Hammer” DeLay, the McCain-Feingold campaign reform act passed the House of Representatives when forty-one Republicans defected to join the Democrats.⁶

It was once tempting to resuscitate a model of pure party conflict by arguing that the United States really had a three-party system, with the Democrats split into northern and southern factions. In the three-party model, majorities would be formed by shifting alliances between two of the three parties. But this model doesn’t work cleanly either. When the 1964 Civil Rights Act was passed, northern and southern Democrats took opposite sides. But, the Republican Party also split, its more conservative members joining the southern Democrats. Similarly, throughout the 1830s and 1840s, slavery roll calls cut across parties, dividing North from South rather than Whigs from Democrats.

More generally, roll calls typically split one or both of the parties. Why are there splits? Perhaps it is because American legislators are parochial — “all politics are local” — and they are overwhelmingly responsive to the needs of their constituents and not responsive to national interests.⁷ If so, for many issues, we would need an issue-specific economic model that specifies and measures the constituency interests. When a roll call involves a strong element of geographic distribution of resources, our simple structure may in fact fail to account for voting behavior. More typically, as we shall show in chapter 6, roll call voting is accounted for by the structure and little is gained by attempting to enrich this accounting by introducing measures of economic interests of constituencies.

Searching for the impact of specific economic interests may be, more fundamentally, fruitless because the legislative process is dynamic, with a vast set of issues considered as time progresses. In a dynamic setting, rational actors may find it in their interest to coalesce and log roll (trade votes).

The linkage of commercial issues was nicely captured by Representative Hewitt (D-NY) during the debate on the Interstate Commerce Act in 1884:

Men of business in New York despair of wise legislation upon these great commercial questions from this House. They have seen this House resist the resumption of specie payments. They have seen this House thrust the silver bill down the reluctant throats of an unwilling community; and now they behold this House and this side of it forcing reactionary measures upon the commerce of the country which will paralyze the business of the port which is the throat of the commerce of this country.⁸

In other words, Hewitt saw railroad freight regulation as linked to previous votes on the gold standard and on a direct subsidy to the silver interests in Nevada. Each vote can thus reflect coalition behavior as well as the apparent substance of the vote. “Anticommercial” interests are likely to stick together on a large set of bills. When such coalitions are stable, a parsimonious model of simple structure may encapsulate the coalitions and account better for voting patterns than attempts to deal with the substance of the roll call in isolation. We in fact find that a model of flexible coalitions typically is far superior to models in the literature that use economic interests.

What simple structure permits flexible coalitions? Briefly, one in which legislators can be described by a continuum of positions. Although the continuum is an abstraction, it is convenient to use the word ideology as a shorthand code for these positions. Henceforth in this book, we use ideology as a shorthand in the sense intended by Converse (1964) in his seminal paper on belief systems. That is, voting is ideological when positions are predictable across a wide set of issues. Someone who favors higher minimum wages is also likely to favor lower defense spending, affirmative action programs, higher capital gains taxes, and so on. We can think of the continuum of ideological positions as ranging from Left to Right or from very liberal to moderate to very conservative.

In contemporary America, this continuum is a perceived reality, a part of the common knowledge of not only the players on K Street and the Beltway but also many ordinary citizens. Consider these six senators: Edward Kennedy (D-MA), Robert Byrd (D-WV), Sam Nunn (D-GA), Alphonse D’Amato (RNY), Strom Thurmond (R-SC), and Jesse Helms (R-NC). There would be widespread agreement among American politics buffs that the order given above is the appropriate liberal/conservative ordering. Our method of estimating the continuum allows us to provide interval level measurements of position, not only for the contemporary period but also for all Congresses, beginning with the first, which convened in 1789.⁹ Moreover, we show that this structure is a predominant feature of nearly all roll call voting.

We can represent most roll calls as splits along the continuum — everyone to one side of a critical point will vote one way and everyone to the other side will vote the opposite way. Which side wins depends on where the critical point is located. If it is to the left of the median of legislator locations, the conservatives get a majority. Conversely, if it is to the right, the liberals win. As the critical point shifts, coalitions shift. Coalitions are therefore flexible, but they must conform to splits along the continuum.

The fact that most roll calls are splits implies that we can represent most votes as mappings from the issues onto the continuum — examples would be the level of the minimum wage; the extent to which assault weapons should be banned; and whether prayer, silent or vocal, should be permitted in schools. Consequently, nearly everything becomes a straight liberal/conservative issue.

Nonetheless, several caveats should be noted.

1. The simple ideological structure does not lead to a predictive model for specific issues. True, in the short term one can predict with accuracy. For example, in Poole and Rosenthal (1991a), we show how the final vote on confirmation of Robert Bork as a Supreme Court justice could have been forecast from the early announcements of members of the Senate Judiciary Committee. Indeed, divisive voting on Supreme Court nominations, when it occurs, fits very nicely into the structure. But to obtain medium- and long-term forecasts, one would need to model how issues map onto the structure. This book will not help one to understand why, sometime before Bork was rejected, a perhaps equally conservative nominee, Antonin Scalia, was confirmed by a 99–0 vote. The book’s basic message is more limited: If issues do come to a vote, a mapping will tend to occur and make votes consistent with the structure.
2. Just one continuum of positions may not be enough. We may need two or more sets of positions to describe roll call voting behavior. Each underlying continuum is termed a dimension. For most of American history, the structure is indeed one-dimensional; at times a second dimension is an essential part of the picture. A second continuum was most important during two periods when the race issue was central to American politics. The first time was during the debate over slavery in the 1830s and ‘40s. (By the 1850s the slavery issue had become so intense that at first roll call voting patterns were chaotic rather than structured; later, patterns were restructured with the slavery issue becoming the primary dimension.) The second occasion was the civil rights controversy of the 1940s, ‘50s, and ‘60s. From the late 1970s onward, roll call voting again became largely a matter of positioning on a single, liberal/conservative dimension.
   In addition to the substantive issue of race, party loyalty — ranging from strong loyalty to one party in the two-party system to strong loyalty to the other — could provide a basis for a second continuum of positions. Indeed, legislators might be viewed, on every vote, as trading off the implementation of their liberal/conservative preferences against the need to be loyal to their party coalition. In 1989, for example, Senate majority leader George Mitchell (D-ME) was able to defeat President George H. W. Bush’s proposed capital gains tax cut by transforming a vote on an economic issue into a crucial test of party loyalty.
   Undoubtedly, party loyalty is involved in our finding that there is a slightly better accounting of roll call votes from using two dimensions, even in periods when the race issue is largely inactive. One’s loyalty to the party, however, is hardly totally independent of one’s liberal/conservative position. Indeed, for most of American history, parties define clusters on the first dimension, which, at some risk of oversimplification, basically represents conflict over economic redistribution. Nonetheless, the clusters are just clusters rather than permanently jelled voting blocs. Some degree of intra-party diversity has always been tolerated in Congress. In the contemporary Congress, there will be some issues, such as the 1981 tax bill, where moderate Democrats vote with Republicans and others, such as the 1991 Civil Rights bill, where moderate Republicans vote with Democrats. On those occasions when whips and leaders enforce party discipline, the roll call split will occur at the point on the first dimension that most clearly divides Democrats from Republicans.
3. Our method for finding the dimensions is blind both to the party affiliation of the legislator and the substance of the roll call vote. The simple structure we find is an abstraction. The fact that a simple abstraction accounts for the data suggests that, although there is some flexibility in forming coalitions, coalition formation is constrained. Parties are obviously an important constraining influence.¹⁰ We observe not only clustering of legislators by party but also clustering of roll calls by substance. Although these clusters enable us to interpret the results, the basic finding is that a simple abstract model accounts for the data.
4. Voting may appear as splits along a continuum even on bills that represent packages dealing with a multitude of policy areas. On these bills, substantial vote trading or vote buying may have taken place. For example, President Reagan obtained the defection of the “Boll Weevils” in exchange for subsidies to Louisiana sugar producers that ought to have been anathema to the free-market credo of his administration. Yet when all the deals are done, roll call voting respects the continuum. If votes are in fact bought on an issue, the buyers will seek legislators with a low price. These should be legislators who are indifferent or nearly indifferent on the issue, that is, legislators who would be close to the point that would separate Yea from Nay voters if there was no vote buying.¹¹ So vote buying is likely to move the separating point rather than to create a chaotic pattern of voting.¹² Similarly, even on issues where a specific constituency interest could cause a legislator to deviate from his usual voting patterns, the legislator must be sure that the deviation is correctly perceived by constituents. Otherwise, the legislator’s reputation may be better served by voting with people that the legislator usually votes with.
5. Voting may not appear as splits along the continuum if legislators are behaving strategically with respect to the agenda represented by a sequence of amendments to a bill. For example, conservatives might strategically vote with the liberals and against the moderates. Suppose, for example, that a bill looked too liberal to be likely to win passage and an attempt was made to moderate the bill in introducing a “saving” amendment. If the amendment were passed, it, rather than the original bill, would be voted on against the status quo. Conservatives who would like to see the status quo preserved might strategically vote against the amendment even though they would prefer a more moderate bill to a very liberal one. But, as we explain in the next chapter, as long as legislators know each other’s preferences and the agenda, “both ends against the middle” and other deviant voting patterns should not occur. In our example, the reason is that the liberals will not be fooled by the tactics of the conservatives. They, too, will be strategic and vote for the “saving” amendment e...