Parties, Preferences, and the House Whip Process Courtney L. Behringer C. Lawrence Evans Elizabeth R. Materese Department of Government College of William and Mary Williamsburg, VA 23187-8795 Prepared for presentation at the annual meeting of the Southern Political Science Association, Atlanta, GA, January 5-7, 2006. Parties, Preference, and the House Whip Process1 This paper addresses perhaps the central conceptual and empirical dispute in contemporary congressional studies; the role played by parties and leaders in the House legislative process. Existing theories are mostly rooted in the spatial model of legislative choice, and generally assume that the preferences of members are fixed and determined exogenously to the legislative process. Our argument is that much of what constitutes “influence” on Capitol Hill occurs during the process through which legislators formulate their policy positions, and thus arguments based on fixed or exogenously determined preferences cannot adequately capture the impact of parties and leaders in Congress. In this paper, we also address a significant empirical limitation in the scholarly literature about congressional partisanship. Most often, scholars use roll-call based indicators to measure the policy preferences of individual legislators. These scores do capture essential features of the structure of the roll call record. But all vote-based indicators are themselves affected by the full range of forces that shape roll call decisions (party, constituency pressures, group lobbying, member views about what constitutes good public policy, and so on). As a result, it is difficult to use these measures to disentangle the effects of member preferences, parties, and other influences on legislative behavior. Instead, we rely on private polls (called “whip counts”) taken by House party leaders prior to major floor votes. The result, we believe, is a more accurate and nuanced portrayal of member decision making on the House floor. We find that parties matter in the House legislative process, but also that the mechanisms of leadership influence are not fully captured by existing scholarship. There are four sections to this paper. The first places our research within the broader literature about the congressional parties and reviews the nature of our evidence. We report on a 1 Preparation of this paper was assisted by financial support from the National Science Foundation (Award SES0417759) and the Roy R. Charles Center of the College of William and Mary. The data used here were primarily gathered from the Congressional Papers of Thomas S. Foley, Manuscripts, Archives and Special Collections, Holland Library, Washington State University; and The Robert H. Michel Papers, Dirksen Congressional Center, Pekin, Illinois. Ed Weber, Holly Tate, Laila Miletic-Vejzovic, and Frank Mackaman were very helpful in facilitating the archival portions of the research. Former Speakers Thomas S. Foley, D-Wash., and James Wright, D-Tex., kindly agreed to be interviewed about partisan coalition building during the era under focus in this paper. The argumentation has benefited from exchanges with Joe Cooper, Larry Dodd, Rick Hall, and Frances Lee. Most important, we thank the two dozen William and Mary students who have been members of “Team Whip Count” and helped translate the archival evidence into usable quantitative data. They are listed by name at http://clevan.people.wm.edu/whip.php. 1 broader project that makes use of archival records from more than 700 whip counts conducted by congressional party leaders since World War II. In this paper, we focus on two-dozen issues from 1977-80 upon which both parties asked nearly identical questions of their members during the whip process. Section 2 examines the way members respond to party whip counts and the linkages that exist between poll responses and roll call behavior. Even though the 1977-80 period occurred before the current era of high partisan polarization in Congress, we uncover compelling evidence of party effects in the process through which members form position in the days and weeks prior to major floor votes. Section 3 is an aggregate analysis of the coalitionbuilding efficacy of Democratic and Republican leaders for the legislative items under focus. Particular emphasis is placed on leadership strategies and effectiveness on the National Energy Policy Act of 1978, perhaps the most significant domestic policy measures considered on Capitol Hill during the late 1970s. A key argument is that research about the congressional parties has been excessively focused on the role of the majority and needs to give greater consideration to the partisan minority. In Section 4, we conclude by briefly summarizing our main contributions and the implications for research about the congressional parties. 1. Goals, Preferences, and Whip Operations The early 1990s was a transformative period in legislative scholarship. Earlier research tended to focus on member goals as the main theoretical primitives in conceptualizing about legislative behavior. Mayhew (1973) famously explored how much of what occurs in Congress is consistent with the goal of reelection. In his landmark study of committees, Fenno (1973) argued that members are motivated by three main goals; reelection, promoting good public policy, and securing influence within the chamber. Kingdon (1973, 1989) found that the roll call decisions of House members were conditioned by Fenno’s goals, subject to the complexity and uncertainty that characterizes the political process. By “goals,” these scholars primarily were referring to the broader ends of political behavior in conjunction with the audiences a legislator hoped to please.2 Member goals served as a transitional concept between the more 2 Reelection seekers aim to stay in office and the constituency is the main audience. Policy-oriented members attempt to shape policy outcomes and the audience is comprised of the legislator and others who share her interests and views (e.g., political activists). The influence goal concerns the accumulation of power and other colleagues are the primary audience. The concept of goal, then, is not easily distinguished from the broader environment of decision making (but see Fenno 1973, Chapter 2). 2 sociologically-oriented conceptualizations of Congress popular in the 1950s and 1960s and the rational choice theorizing that has since come to dominate the field. During the 1980s, scholars increasingly turned to spatial representations of legislative behavior. There are many different applications of the spatial model to Congress and we do not want to over-generalize.3 But rather than member goals, the main ingredients are the “preferences” of legislators over concrete policy alternatives. A central assumption is that preferences and alternatives can be arrayed along one or more underlying dimensions of evaluation, such as the basic liberal-conservative spectrum. Other ingredients include premises about behavior (e.g., sincere versus sophisticated voting), rules and procedures (e.g., the assignment of agenda prerogatives), and the distribution of information. Early applications of the spatial model mostly relied on anecdotes and stylized facts to illustrate empirical relevance, but by the end of the 1980s scholars were regularly confronting these models with systematic data about Congress. Much of the empirical work that ensued examined whether party leaders exert a significant and independent impact over legislative outcomes. Indeed, the way the question of party influence is now framed and the kinds of empirical tests that are most often conducted are firmly rooted in the basic logic of the spatial model. There are several competing perspectives. First consider the majoritarian perspective, which is associated in particular with the work of Krehbiel (1998) and is explicitly spatial in its foundations. The House, according to Krehbiel, is first and foremost a majority-rule institution. If the preferences of members can be arrayed as points along a single ideological continuum; members are fully informed about the main ingredients of the legislative game; and they vote for the proposal closest to their policy preferences; then the predicted outcome is the alternative most preferred by the median legislator in the full chamber. At least within the House, the middle should rule and party organizations are at most a secondary feature of the legislative process. Other scholars, in contrast, maintain that under the right conditions majority party leaders in the chamber do have the resources necessary to move policy outcomes away from the preferences of the floor median and in the direction of median viewpoints within the majority caucus. It is feasible to devise a spatial model of legislative choice that allows for learning, persuasion, and the emergence of member positions over time. But the scholars who have 3 For a good reviews, consult Krehbiel (1988) and Shepsle and Weingast (1995). 3 attempted to incorporate parties into the spatial model typically take the policy preferences of legislators as given. According to the conditional party government perspective of Aldrich and Rohde, when preferences within the majority party are homogeneous and there are significant differences in viewpoints between the parties, there may be incentives for rank-and-file members of the majority to provide their leaders with the formal powers necessary to shift outcomes from the floor toward the majority party median (see Rohde, 1991, and Aldrich and Rohde, 1998). Alternatively, according to the party cartel perspective of Cox and McCubbins, the shared electoral fate of majority party members may lead them to provide the leadership with the power to block from the agenda issues that would divide their party and undermine its “name brand” with the public (Cox and McCubbins, 1993, 2005). Alternatives and outcomes preferred by the chamber median never make it onto the floor agenda. Although all of these scholars recognize that parties and leaders may influence how legislators make up their minds on major issues, their focus is more on other features of the Congress (especially agenda control in committee and on the floor) for explaining why parties matter. And their common benchmark for establishing party influence is whether or not policies diverge from the preferences of the floor median toward viewpoints within the majority party. At least since the early 1990s, this manner of framing the question of party power has shaped the research agendas of empirically-minded scholars of Congress. The literature here is also extensive and we do not want to over-generalize. But scholars often use data from roll call votes on the House and Senate floor to estimate member ideological preferences; relate these estimates to certain aspects of legislative behavior (often roll call decisions); and then attempt to make inferences about strategy and influence (the relative importance of leadership pressure versus member preferences, and so on). Typically, the goal is to isolate a significant effect for party above and beyond the effects of preferences as measured by a roll-call based ideological score. There are obvious problems with this approach to measurement. For one, all vote-based measures, including the DW-NOMINATE values estimated by Poole and Rosenthal, are themselves determined in part by partisan imperatives.4 Poole and Rosenthal (1997) demonstrate that the vast majority of roll calls cast on Capitol Hill can be effectively structured with two 4 It should be emphasized that Poole and Rosenthal know exactly what they are tapping with the DW-NOMINATE measures. Their primary focus in Poole and Rosenthal (1997), for instance, is on the evolution of voting patterns over time. 4 underling dimensions of evaluation. Indeed, for most of congressional history a single ideological dimension will suffice. We also know that party leaders seek to maintain their positions by minimizing conflict within their own caucuses and by winning on the floor (Sinclair, 1981). Leaders can promote both goals by focusing the floor agenda on items that unite the majority caucus internally and by convincing potentially recalcitrant legislators to toe the party line on roll calls. Throughout American history, the policy programs of the two parties have generally reflected the main ideological fault lines of the day. Thus, it comes as no surprise that there is an enduring ideological structure to the roll call record. Indeed, this structure is in part a consequence of party strategizing and influence. For purposes of illustration, consider a standard story of party leadership influence, which might go something like this. As a major floor vote nears, members are cross-pressured by party and constituency concerns and lean toward voting with the district and against the party. In response, the leadership attempts to convince wavering legislators to approach the issue from an ideological rather than a district perspective by emphasizing national concerns during the lobbying process, making promises and threats, and so on. As a result, the cross-pressured members choose to downplay local interests in their political calculations, cast ideologicallymotivated votes, and the party position carries the day. If we use a member’s relative placement within the ideological structure of the roll call record as a measure of her preferences, the preference indicator will perfectly predict behavior leaving little variance for party to explain – even though partisan imperatives determined behavior.5 The scholarly community would benefit from access to systematic evidence about the policy positions of legislators prior to the end of leadership lobbying on significant roll calls (Fenno, 1985).6 With such data, researchers would be able to gauge the changes that occur in member positions during the lobbying process and also whether these changes are toward the leadership position. For both conceptual and empirical reasons, then, we believe that legislative scholars should look beyond the simple spatial model and focus sustained attention on the process through which legislators formulate policy positions. Through extensive research in the papers 5 Lee (2005) finds that much of the partisan/ideological behavior that we observe on Capitol Hill occurs on issues that do not inherently divide liberals from conservatives. 6 More sophisticated research designs and estimation techniques probably will not be sufficient to produce a consensus about the importance of the congressional parties based on the roll call record alone. Consult, for example, the exchange between Snyder and Groseclose, 2000, 2001; McCarty, Poole, and Rosenthal, 2001; and Cox and Poole, 2002. 5 of former congressional leaders, one of us (Evans) has gathered extensive archival records for nearly all of the whip counts conducted by House Democrats from 1955-86 and House Republicans from 1973-80 and 1989-90; important periods of transition in the role of party leaders in Congress. The records include completed whip pads, memos and lists, formal tally sheets, completed House worksheets (small pamphlets with the names and positions of members denoted by state), lists of members that the leadership is targeting for lobbying, the names of the individuals tapped to do the lobbying, and other evidence about the coalition-building process. As of January 2006, over 10,000 pages of archival records had been photocopied for the project. Several scholars have already produced important studies of the House Democratic whip operation, including examinations of the results of past whip counts.7 Ripley (1964, 1967) analyzed the whip process from 1962 to 1964, providing a highly valuable treatment of the early institutional development of the office. Dodd (1978; and with Sullivan, 1981) examined dozens of House Democratic counts from the 93rd Congress (1973-74); his studies are an important portrait of the whip system during a time of transition for the congressional parties. More recently, Burden and Frisby (2004) used whip count records from 1971-72 to evaluate the coalition building effectiveness of the Democratic leadership. These studies all focus on relatively short time intervals prior to the 1980s emergence of strong party leadership in Congress, however, and their empirical reach does not extend to the Republicans or the Senate. The broader project of which this paper is a part is far more comprehensive in scope; the aim is to conduct a systematic analysis of partisan coalition building over many decades of recent congressional history. Whip Logistics How does the whip process work? For several decades, the party whip process in the House generally has commenced with a request from the top party leadership for a count of member preferences on a pending matter. Within the House leadership structure, the whip is the third ranking position on the majority side (behind the Speaker and Majority Leader) and second ranking for the minority (behind the Minority Leader). The Whip communicates the request (including a deadline for completing the poll) to members of the extended whip organization for 7 Sullivan has also authored a number of important studies using data from presidential head counts, which resemble whip counts in certain ways (e.g., Sullivan 1990a, 1990b). See also Covington (1987). 6 the relevant party. In the House, the Democratic whip operation is comprised of about two dozen deputy, zone, and “at large” whips. The GOP operation includes a smaller number of regional and assistant regional whips. Both parties divide their memberships into zones (or regional areas), with a member of the extended whip operation assigned responsibility for tracking viewpoints in that area. The two party whip teams seldom share information, focus on preferences within their own caucuses, and often poll on different matters. Once the request for a whip count is made, the zone whips contact the offices of the members within their areas, either directly or through staff, and ask for their positions on the item being polled. The question being polled is almost always phrased so that a “yes” response is in favor of the position of the leadership. As they hear from their colleagues, the zone whips report back to the Whip with their tallies, singling out particular members as potential problems or needing contact or persuasion from the leadership. The results of the initial tallies often are recorded on small pads, with a separate sheet used for each zone. Initial reports typically include many nonresponders; members who decline comment or are unreachable. As the vote nears, the number of nonresponders drops and positions evolve. That information also is forwarded to the Whip’s office. At this point, the information in the whip pads is often transferred into list or memo form for ready reference during meetings of the leadership. Based on our catalogue of polled questions, there has been a fairly steady increase in the number of whip counts over time, ranging from about 20 per Congress for House Democrats during the 1950s to about 50 per Congresses for the party during the 1980s (Evans 2004). For both parties, there was a significant spike in the number of counts during the late 1970s. Indeed, the Democrats conducted over 100 whip counts during the 95th Congress alone (1977-78). For the two Congresses for which data from both parties are now available (1977-80), the majority Democrats conducted almost twice as many whip counts as did the minority Republicans. The difference is not surprising. The majority party is primarily responsible for managing the floor agenda, and, as a result, the informational value of whip intelligence should be higher for the majority than it is for the minority. As mentioned, at this point we have only compiled whip count records for both Democrats and Republicans for 1977-80. During the period, House Democratic leaders polled their fellow partisans on over 150 distinct questions and Republican leaders contacted their members on about 70 questions. In selecting legislation to whip, party leaders tend to focus on 7 high visibility items that touch on major party priorities and for which the outcome is not a completely forgone conclusion. Not surprisingly, there is considerable overlap in the legislation targeted for whip activity by Democratic and Republican leaders. Indeed, almost all of the bills that generated GOP whip activities during 1977-80 were also the subject of one or more counts by the Democrats. The specific questions posed on a measure by Democrats and Republicans often differed, however, reflecting divergent party interests and strategic concerns. In this paper, we focus on the polled items from 1977-80 for which Democratic and Republicans asked their members identical or nearly identical questions. We do so for two main reasons. By most accounts the 95th and 96th Congresses constitute an unfriendly period for uncovering party effects. Although the Democratic majorities at the time were very large (292 members in the 95th Congress and 277 members in the 96th), the Democratic Caucus was rife with divisions.8 The traditional New Deal coalition that had structured the national party since the 1930s was withering and the 1980s resurgence of mass partisanship had yet to emerge (Bartels 2000, Hetherington, 2001). The proportion of party unity roll calls (floor votes on which a majority of one party votes “yes” and a majority of the other votes “no”) was just 37.3 percent in the 95th Congress and 47.2 percent in the 96th; significantly less than the 67.4 percent that characterized the 104th Congress (1995-96), arguably the recent high-water mark for conditional party government (Aldrich and Rohde, 1998). Moreover, Jimmy Carter was in the White House during 1977-90 and is not generally regarded as an effective legislative president (Jones, 1988). Democratic congressional leaders could not count on Carter to help them build majorities behind the party program. If we can uncover systematic evidence of party influence and efficacious coalition building by the leadership during this era, the result will be strong support for assertions that the congressional parties matter in internal decision making. There also is a more practical reason for our focus on 1977-80. Only within this time period do we have coalition-building data for both House Democrats and House Republicans. To our knowledge, few scholars have systematically evaluated majority and minority party influence on a common set of issues.9 Four decades of social choice theorizing suggests that floor majorities in legislatures can be highly unstable. It is possible, then, that the vote mobilization efforts of the majority leadership are counter-balanced to some extent by the 8 9 See Cooper and Sieberer (2005) for an analysis of the linkages between party size, cohesion, and success. Jenkins, Crespin, and Carson (2005) is a noteworthy exception. 8 lobbying activities of the minority party and its interest group allies. Integrating data from both parties also enables us to rigorously test for the impact of partisan differences on the roll call choices of wavering members, controlling for other factors. More generally, minority party leaders (within their own caucuses) have certain of the prerogatives afforded to the majority leadership. As much as possible, we believe, the leading theories of party in Congress should be evaluated with evidence from both the majority and minority parties. 2. Whip Counts and Roll Calls Table 1 is a list of the items on which Democratic and Republican leaders posed the same questions to their members during 1977-80. There were dozens of additional bills on which both party whip systems were active, but they focused on different aspects of the legislation. The Republicans might poll on key amendments from minority party members, while the Democrats might emphasize the rule and final passage votes. In contrast, the items in Table 1 constitute the legislative equivalent of a “jump ball.” Usually, leaders only conduct whip counts on major priorities where the outcome is somewhat in doubt. Whip counts are time consuming, require the investment of nontrivial political capital, and can aggravate rank-and-file members if conducted indiscriminately. When both parties poll on the same question, the implication is that the matter is of major importance and both sides view it as at least potentially winnable. Included in Table 1 are longstanding issues of contention between the two parties; the minimum wage, weapons programs, campaign finance, and also Jimmy Carter’s landmark proposal to overhaul U.S. energy policy. The column at the right denotes the party position that prevailed on the floor. In one case, the Amtrak reorganization, the outcome is categorized as ambiguous. Here, the Republican leadership formally opposed the amendment in question, which was offered on the floor by Al Gore, D-Tenn. But the Carter administration was also lobbying against the amendment, and based on documentation in the whip files it is clear that Democratic leaders did not take a formal position on the matter. As a result, the defeat of Gore’s amendment was a GOP victory, but not really a loss for the Democrats. The outcome of the Budget Targets fight is categorized as a victory for both leaderships. Prior to the May 1977 roll call, Democratic and GOP leaders struck a compromise that eventually passed by a margin of 221-177. However, the vast majority of 9 Republican members broke with their leadership and joined 70 Democrats in opposing the accord. Of the remaining 22 items, thirteen (59.1 percent) are victories for the majority Democrats and nine are wins for the minority party. As mentioned, the Democratic Caucus was very large during the 95th and 96th Congress (approaching 300 members) and a win rate of less than 60 percent may strike some observers as unimpressive. But it should be emphasized that the policy disputes in Table 1 do not constitute a random sample of floor fights during the period. Indeed, they are not even a random sample of the more partisan fights that occurred in the chamber. Instead, the fact that the same questions were whipped by both parties means that the items were judged to be significantly in play and that both party organizations (and their interest group allies) were actively engaged. Especially given the significant cleavages within the Democratic Caucus of the day, the outcomes in Table 1 do suggest a degree of efficacious coalition building by the Democratic leadership. Table 2 provides summary information about the challenges confronting the two leaderships at the beginning of the floor endgame on the jointly polled items. Included are the standard response positions (yes, leaning yes, undecided, leaning no, and no). Remaining responses (no response, no comment, unreachable, etc.) are collapsed into a single “other” category to simplify matters. We also have organized the data so that a “yes” response is always in favor of the position embraced by the relevant leadership.10 There are some noteworthy differences between the parties. In percentage terms, the GOP “base” as revealed by the whip counts (“yes” plus “leaning yes” responders) was significantly higher than was the case for the majority Democrats (67.1 percent versus 54.5 percent). The difference needs to be considered from the perspective of party size. The party support percentages in Table 2 translate into 150160 Democratic votes and 96-106 GOP votes. Still, the Republicans were relatively more united even though the overall levels of party unity on all roll calls during the relevant Congresses were 10 It is not unusual for the archival records of whip operations to include multiple drafts of a count, with the positions of some members changing from draft to draft as the lobbying process continues and the roll call nears. Indeed, for some issues five or more lists of the results may be compiled in the day or two before the relevant vote. The main changes from earlier to later drafts typically are a reduction in the number of nonresponders. To simplify matters, when the positions of members change across drafts, we code the position that maximizes the level of response ambiguity. If a member is “undecided” on one draft and “leaning yes” on the next, the position is coded as “undecided.” If there is a tie in terms of distance from undecided (e.g., shifts form “no” to “yes,” or from “leaning no” to “leaning yes”), then we coded the response that was furthest from the leadership position. A nonresponse is coded if and only if the member is nonresponsive across all drafts. 10 similar across parties.11 Again, whip count data are useful in part because they focus our attention on the items that are directly relevant to the party agendas. Moreover, the reduced Democratic loyalty on the counts does not derive from member ambivalence about party priorities. The percentage of legislators who are “undecided” is actually very similar across parties (16.1% for Democrats and 15.1% for Republicans), as is the proportion in the “other” category (11.1% and 9.1%, respectively). The main difference is in the percentage answering “no” and “leaning no,” which is almost twice as high for the majority Democrats. How successful were the majority and minority leaderships at retaining their respective bases and at persuading undecideds and potential defectors to toe the party line? Table 3 portrays roll call support for the party position for the 23 jointly polled items that resulted in a vote.12 “Yes” and “leaning yes” responses are combined into a single category (Y/LY), as are “no” and “leaning no” responses (N/LN). Not surprisingly, for both parties the vast majority of the members in the Y/LY category ended up supporting the position of the relevant leadership. Still, there are some exceptions, especially for the majority Democrats. On four of the measures, twenty or more Democrats responded as “yes” or “leaning yes” on the whip poll but ended up voting the other way. The political context changed or opposition lobbying dominated the vote gathering efforts of the Democratic leadership. For both parties, most of the N/LN responders eventually voted against the position of the relevant leadership. But there also are some exceptions here. Democrats were somewhat successful in converting likely defectors on two of the minimum wage votes, funding for the B-1 bomber, the foreign intelligence bill, the natural gas substitute, the procedure for the energy conference report, and measures relating to PAC contributions, hospital costs, and revenue sharing. On six of the items – financial ethics, minimum wage indexing, the natural gas procedure, windfall profits, hospital costs, and revenue sharing – the Republicans and their interest group allies actually were able to convert all or almost all of the N/LN responders to the party position. Party leaders tend to view the undecided members as most “in play” on whipped legislation. For the undecideds, the most striking characteristic of Table 3 is the level of 11 The average percent of Democrats voting the party position on all party unity roll calls was 72.7% in the 95th Congress and 76.6% in the 96th. For the GOP, the analogous percentages were 76.1% and 79.0%. The party unity data were gathered by Joe Cooper and Garry Young. See http://www.jhu.edu/polysci/faculty/cooper/papers.htm. 12 For purposes of comparison, the Budget Targets and Amtrak Reorganization items are included. There is no entry for Universal Voter Registration: The Democrats pulled the measure in part because of unpromising whip count results and no roll call occurred. 11 variation that exists from item to item. Democratic and Republican leaders, for instance, retaining almost all of their undecided members on the financial ethics roll call, but the undecideds within both parties split fairly evenly on the labor law revision, consumer protection, and the Panama Canal measure. There also are potentially instructive asymmetries across the parties. While the Democratic undecideds were divided in their roll call choices on the hospital cost measure, for example, Republican leaders were able to retain 34 of the 38 GOP members who responded as “undecided” on the whip count. Similar issue specific variation is apparent for the “other” category.”13 In Table 4, we present the party retention rates for each response category, aggregated across the 21 jointly whipped items for which the parties took divergent positions and there was a roll call. Four points are important. First, as would be expected, the party support rates for both parties fall the further the whip count response is from “yes.” Second, for both parties the support levels for members in the “other” category is less than the support levels overall. Most of the entries categorized as “other” were for members who failed to respond to the relevant whip poll, suggesting that nonresponse behavior is partially strategic. Member who are potential opponents of the party position apparently are disproportionately likely to duck questions about their intentions from the whips. Third, both parties converted a larger portion of the “no” and “leaning no” responders than they lost from the “yes” and “leaning yes” categories, which strongly suggests that the vote gathering efforts of the leadership are efficacious. The difference is especially striking for the GOP. While only about 2.5 percent of initial supporters defected on the associated roll calls, about 43 percent of “no” and “leaning no” responders ended up staying on the partisan reservation. Fourth, across all categories of poll response, the Republicans were more successful (in percentages terms) at retaining their members. Once again, inferences about the relative success of Democratic and Republican Party leaders during the lobbying endgame need to be tempered by the significant differences in party size. In both Congresses, there were about twice as many Democrats as Republicans in the chamber. The results in Tables 3 and 4 are highly suggestive that party leaders indeed make a difference in the coalition-building process. According to the spatial model, the probability that 13 The issue-specific variation in relative party retention and conversion rates may be related to differences across policy areas in party advantage a la Petrocik’s (1996) concept of “issue ownership.” Alternatively, the differences may relate to aspects of the procedural context on a measure. These questions will be explored in further research, both for the jointly polled items in 1977-80 and for the hundreds of additional questions for which whip data are available. 12 a member will be undecided on a roll call should be symmetrically distributed over the position (on the underlying ideological continuum) that is located midway between the two alternatives under consideration (Poole, 2005). If the question is about final passage, for instance, the two alternatives under consideration are the bill as amended and the status quo of existing law. The members who most prefer an outcome (along the relevant ideological dimension) located midway between the bill and the status quo should be most likely to be indifferent and thus answer as “undecided” on the whip count. Based on “preferences” alone, the likelihood of a “yes” or “no” roll call for these members should be about the same. Yet, for both Democrats and Republicans, undecided legislators break disproportionately toward their party positions. Of course, it is possible (although unlikely) that the Democratic undecideds all have policy preferences located near the midpoint but still somewhat closer to the position of the Democratic leadership. Similarly, it is feasible that GOP undecideds have preferences near the point of indifference, but still falling on the side closest to the Republican positions. As a result, although highly suggestive, the loyalty rates of undecided members in Table 4 are not definitive evidence of party effects. Moreover, about 40 percent of Democratic undecideds and 30 percent of Republican undecideds ended up voting against their party positions. We need to dig a little deeper. Table 5 reports the results of a multivariate analysis of the voting choices of undecided members pooled across the two parties. The dependent variable is dichotomous, taking the value of one if a member voted the Democratic position and otherwise is zero. Only observations from the 21 questions for which the party leadership took clearly divergent positions and a roll call occurred are included. The independent variable of primary interest is party, which takes the value of one if a member is a Democrat and zero if she is a Republican. Our goal is to construct a “conservative” test and stack the decks against uncovering party effects. As a result, we also consider three independent variables that are often used as proxies for member “preferences” in the literature. The first is an indicator of constituency ideology; voting patterns within the district at the presidential level. Here, we use percentage support for Richard Nixon in 1972 because the relationship between presidential voting and district ideology was murky in 1976 – Democrat Jimmy Carter was from Georgia and did particularly well in the conservative south because of his strong ties to the region. Since voters tend to elect candidates to Congress reflecting their ideological views, the representatives from 13 relatively conservative (pro-Nixon) districts should have policy preferences that are right of center, with the opposite association holding for legislators representing liberal constituencies. Along with constituency ideology, we also consider the placement of members along the first and second dimension DW-NOMINATE scales. The first-dimension values, of course, capture the relative liberalism of a legislator’s voting record as it relates to the roll calls cast by other members of Congress. The substantive meaning of the second dimension scores varies over time, but during the 1970s it chiefly captured regional conflicts between the north and south (e.g., lingering disagreements over civil rights and increasingly over social issues). From 1990 onward, the second dimension scores are of limited value for predicting behavior, but in the 1970s the dimension was still consequential and needs to be included in our analysis. Once again, we emphasize that the Poole-Rosenthal measures include party effects and are themselves shaped by party agenda setting and leadership strategies. By including both scores and the measure of constituency ideology, we are making it difficult for the party dummy variable to exhibit an independent impact on roll call choice. Moreover, our focus is on member decision making in the days and weeks prior to some of the most important and highly publicized roll calls of the session. Much of the influence exerted by the extended party leadership may occur at the agenda setting stage or during committee deliberations. Indeed, Cox and McCubbins (2005) assert that party influence should be less pronounced when members make decisions about floor roll calls because these decisions are so visible and open to scrutiny by interest groups and constituents. In Table 5, we analyze the effects of party on the roll call choices of undecided legislators via three specifications. Notice that the parameter estimates for party are statistically significant and have the expected sign across all three models. The coefficient for district conservatism has the appropriate sign, but loses its statistical significance when the DW-NOMINATE measures are included. Not surprisingly, both DW-NOMINATE scores are strong predictors of the vote. Undecided members who are relatively conservative overall and tend toward the conservative position on civil rights/social issues (larger DW-NOMINATE values along both dimensions) are significantly less likely to break toward the Democrats. Further evidence about the substantive importance of these factors – and about their relationships to one other – is provided in Table 6 and Figure 1. If only party is included as a predictor, the probability that an undecided member will vote the Democratic position is .285 for 14 Republicans and .606 for Democrats; a difference of .321. The impact of party drops marginally when constituency ideology is added to the regression. When the two roll-call based indicators of member ideology are added, the substantive impact of the party dummy variable falls to .127, but is still substantial. Notice that for Republican undecideds, the probability of Democratic support falls below the cut-point of .5 if the other variables are set to their means. For the undecided Democrats, the probability of supporting the party position is above the .5 cut-point (again, with other independent variables taking on their mean values). On the margin for undecided members, then, party effects are large enough to make the difference between voting the leadership position and breaking toward the opposition. Figure 1 portrays the likelihood of a pro-Democratic vote for undecided members at different points of the liberal-conservative continuum. The relationship is captured separately for each party, setting other explanatory variables to their means. Across the entire ideological dimension, the probability of a pro-Democrat vote is higher for Democratic members, but the magnitude of the difference is largest for moderates located nears the midpoint of the PooleRosenthal measure. We believe that the sharp drop-off in the estimated importance of party across the three model specifications underscores why it is helpful to look beyond voting scores for gauging the policy preferences of members. Significant party effects are imbedded in all roll-call based measures of member ideology. As with voting scores, the positions that members adopt on nose counts are themselves shaped by party pressures. The fact that a question is being polled signals to legislators that the leadership is looking for support on the matter. And how a question is phrased during the whip process generally signals the response that the leadership is looking for. The whips also engage in persuasive efforts during the polling process. Still, party nose counts are issue-specific and occur prior to the roll call. The polling records enable us to track movements toward and away from the leadership position. Our analysis of these data shows that partisan vote-gathering efforts matter. 3. Coalition Building Efficacy Leadership influence may matter at the individual-level, but the aim of the whip process is to affect the outcome of the legislative process. To what extent is the endgame lobbying conducted by party leaders consequential for legislative outcomes on the floor? The minority 15 Republicans tended to retain a higher percentage of members in each response category, but the reason might have been that Democrats needed to mobilize a much smaller portion of their caucus to carry the day. Indeed, the leadership may tend to not lean on undecided member once a coalition of majority size has been achieved. To do otherwise might waste valuable political capital and create unnecessary problems for fellow partisans at home (King and Zeckhauser, 2003; Sinclair, 1995). Undecided or wavering Democrats may be less likely to stay loyal than is the case for their GOP counterparts because their votes are less likely to be needed. We can explore such possibilities by considering the size of the leadership base on each item in our sample, the number of additional votes necessary to secure victory, and whether or not the leadership was able to secure the needed increment. The information is provided in Table 7A for Democrats and Table 7B for the GOP.14 In both panels of Table 7, the second column – “intra-party base” – refers to the number of members within the relevant party who responded “yes” or “leaning yes” on the whip count for an item. In Table 7A, the third column denotes the number of Republicans who told their whips that they intended to (or were likely to) vote with the Democrats: These individuals responded as “against” or “leaning against” the position of the GOP leadership during the whip process. Similarly, the third column of Table 7B contains the number of Democrats who took positions favorable to the GOP position on the relevant poll. In the panels of Table 7, the sum of columns two and three yields the total base for the relevant leadership at the beginning of the floor endgame. Do Democratic leaders actually know how many Republicans are leaning toward the Democratic position prior to floor action, and do Republican leaders have similar information about likely Democratic defectors? Based on the archival record and existing scholarship, the answer appears to be “yes.” Democratic leaders will seldom approach wavering Republicans directly, but instead will ask a Democratic member who knows the GOP lawmaker personally to make the contact for them.15 Alternatively, the leadership may approach members of the other 14 Because they do not affect the official results, paired votes are not included in this part of the analysis and the treatment of the energy bill in Table 9. 15 During Newt Gingrich’s tenure as House GOP whip, 1989-94, he constructed a shadow whip system to gather position intelligence from Democratic members. During the 101st Congress, the shadow organization was headed by Olympia Snowe, Maine, and Thomas Bliley, Va., and included as assistants more than a dozen Republicans with ties to members across the partisan aisle. These Republicans were each assigned a set of Democrats to contact with whom they had geographic, committee, or personal links. Gingrich’s whip staff tabulated likely positions for Democrats on at least a half dozen polled questions in 1989-90 alone. The organizational complexity of Gingrich’s 16 party through allied interest groups that are actively lobbying both sides of the aisle on the matter. On labor issues, for example, the AFL-CIO collected position information about Republicans, as well as Democrats, and (based on archival records) was willing to share the political intelligence with friends in the Democratic leadership. The consumer movement provided the Democratic leadership with similar data about Republican positions on the consumer protection measure in 1977-78. In short, party leaders are generally able to weigh support from both parties when estimating the magnitude of the vote-gathering challenge before them. The fourth column in Tables 7A and 7B – “minimum necessary pickup” – is an estimate of that challenge; the number of additional votes that the leadership needs to prevail on the floor.16 The fifth column contains the actual pickup on each item, and the column at the far right denotes the margin of victory or defeat. For the 22 items (Universal Voter Registration is included here because we can gauge the outcome), notice first that the Democratic leadership had the votes necessary to win on the floor prior to endgame lobbing on two measures: Financial Ethics and the extension of ERA.17 Partisan imperatives may have been pivotal to the outcome by shaping member decision making in committee or the initial positions that members took on the relevant whip counts. But on these initiatives, the majority leadership already had the support necessary to prevail before endgame lobbying began in earnest. The initial margins on the whip counts were small – nine votes and seven votes, respectively – and the Democratic leadership obviously felt the need to shore up its majorities. But for these items, the whip process probably did not alter the outcome. For the Republican leadership, the minimum necessary pickup is positive across all 22 items in Table 7B. We can usefully collapse the items that were “in play” into three categories based on the magnitude of the challenge confronting the leadership: minimum necessary pickups of 30 votes or less (low pickup items); minimum necessary pickups between 31 and 60 (medium pickup items); and items for which the minimum necessary pickup was 61 votes or more (high pickup items). Not surprisingly, there is a reverse symmetry across the two parties. Items that require shadow system was unusual, however, and there is no evidence that House Democrats or Republicans maintained anything of this magnitude during 1977-80. 16 The number of votes necessary to win is not necessarily 218 due to absences. We calculated the minimum necessary pickup based on the number of members who actually cast “live” votes on the relevant roll call. 17 The base behind the position of the Democratic leadership was also sufficient to prevail on the Budget Targets item, which is excluded from Tables 7A and 7B because the two leaderships were in agreement. 17 low pickups for the Democrats tend to call for high pickups by the GOP and vice-versa. Still, the relationship is not perfect. Campaign finance reform and the Humphrey-Hawkins fullemployment measure were low pickup bills for the Democrats, but fell in the medium range for the GOP. The foreign intelligence proposal was a medium pickup item for the Democrats and a high pickup item for Republicans, mostly because of the high number of undecided members in both parties unsure about the underlying policy issue. Interestingly, success rates across the parties are roughly similar within pickup categories. Among their respective low pickup items, the Democrats prevailed on six of seven and the GOP on four of five. Among their high pickup items, the Democrats lost five of seven and the GOP failed on nine of ten. The parties basically fought to a draw on their medium pickup items – here, the Democrats won on three of six, while the Republicans prevailed on three of the seven questions that fell in the middle category for them. Clearly, the aggregate results do not suggest that the majority party dominates the endgame lobbying process – at least not for jointly whipped items during the 1977-80 period. Minority party leaders have many of the same resources for influencing their members that majority party leaders have vis-à-vis the majority caucus. Included would be valuable committee assignments, other forms of party patronage, access to party-controlled campaign funds, and simply the interpersonal pressure that members of a group feel to contribute to collective goals.18 High levels of unity on major roll calls can help foster a coherent and favorable name brand for the minority party, as well as for the partisan majority. Perhaps most important, party leaders do not whip rank-in-file members in isolation from the broader legislative environment. Minority party leaders may lack many of the formal prerogatives extended to the majority leadership, but whipping efforts on the minority side of the aisle are often reinforced by the party’s allies in the interest group community. Consider the Windfall Profits proposal, which was a high pickup item for both leaderships in June 1979. The provision was a substitute amendment offered by W. Henson Moore, R-La., and James R. Jones, D-Ok. It was intended to weaken a windfall tax measure that had been reported on a 20-16 vote by the Committee on Ways and Means earlier that month. The oil industry staunchly opposed any new tax, but preferred the Moore-Jones proposal to the 18 We thank Rep. Dan Lipinski, D-Ill., for repeatedly emphasizing this very obvious, but nonetheless overlooked, point in personal conversations. 18 committee bill. The whip count revealed deep divisions among chamber Democrats, largely along ideological and regional lines. When the matter came to a roll call, Republican members voted for the proposal by a margin of 146-10, and they were joined by 90 conservative and moderate Democrats. The key factor behind House passage of the Moore-Jones substitute was not the relative efficacy of the coalition-building efforts of GOP leaders. By most accounts, the added support for the proposal derived from an intensive lobbying campaign orchestrated by the oil industry. In gauging the relative effectiveness of the two whip systems, then, it is important to keep in mind that party leaders do not lobby in a vacuum. In the contemporary Congress, most major issues will activate a range of outside interest groups on both sides of the question. Studies that purport to uncover the exercise of legislative influence by the majority or minority party are really capturing the impact of the parties and of the interest groups that are allied with them on a matter. More generally, the potential efficacy of lobbying operations on the minority side means that scholarly theories and empirical tests about party influence need to consider the minority, as well as the majority. Too often, studies of party leadership focus almost exclusively on the majority party.19 Still, a closer examination of the coalition-building data in Tables 7A and 7B indicates that the majority Democrats were particularly effective when the stakes were highest. On two of the most significant party priorities, the minimum wage and the massive energy reform legislation, the Democrats orchestrated impressive “come from behind” victories. The specific questions related to the “youth differential” amendment to the minimum wage measure and the natural gas pricing substitute. Although both items were firmly in the “high pickup” range for Democrats, the leadership prevailed after intensive lobbying campaigns aimed at retaining wavering members. In addition, the two jointly polled items in our sample that were decided by a single vote – the youth differential amendment and the procedure for considering the conference report on the energy bill – were both won by the Democrats, a strong indicator of effective lobbying by the majority. Indeed, it is useful at this point to consider in more depth coalition-building efforts on the energy reform matter. By most accounts, the bill was the most 19 Krehbiel and Wiseman (2005) make precisely this point in their study of minority party committee assignments during the Cannon era. 19 significant and hotly-contested piece of legislation during the Carter administration. And it was the subject of more whip activity than any other single measure during 1977-80.20 The energy reform fight was launched early in 1977 by President Jimmy Carter and quickly became the top domestic policy priority for Democratic leaders in both chambers. During the previous Congress, efforts to enact an energy bill had floundered in the House, largely because the decentralized committee system proved incapable of producing a coherent proposal that could win on the floor (Oppenheimer, 1980; Jones and Strahan, 1985). As a result, early in 1977 Speaker O’Neill established a temporary Ad Hoc Committee on Energy and charged it with coordinating the activities of the standing committees with jurisdiction in the area. Although Carter aggressively attempted to sell his energy program on Capitol Hill and to the general public, opinion polls indicated that most Americans did not view passage of the administration plan as critical. Within Congress, the issue evoked cross-cutting cleavages that divided members along partisan, ideological, and regional/economic lines. In the House, O’Neill devoted the full powers of his office to shepherding a version of the Carter plan to passage. In late July 1977, as floor action neared, the Democratic whips conducted nose counts on seven distinct questions, and within a week followed up with two more. The questions are presented in Table 8. The first set of polled items concerned the rule, four likely amendments, and the expected GOP motion to recommit, as well as passage of the measure as reported by the ad hoc committee. The most significant issues were all subject to whip counts, indicating the importance of the bill to the leadership and the precariousness of the coalition forming behind the legislation. The key issue was natural gas pricing. It was the topic of two Democratic whip questions (“Will you support the Ad Hoc Committee amendment on natural gas pricing?” “Will you oppose any natural gas deregulation substitute amendment?”) and one nose count on the GOP side (“Will you support the deregulation of new natural gas?”). The coalition-building challenge confronting the leadership is summarized in Table 9 (as well as Tables 7A and 7B for the jointly polled items). Because the two sets of polled items do not completely overlap, Table 9 provides the actual roll calls cast by the opposing party as an estimate of likely support from that side of the aisle (column 2), as well as the intra-party base 20 Our treatment of the energy policy fight draws in part on coverage provided in various issues of Congressional Quarterly Weekly Report. 20 (“yes” and “leaning yes” responses on the relevant nose count) and the margin of victory or defeat. Notice first that only 140 Democrats were “yes” or “leaning yes” on the ad hoc committee’s natural gas proposal, while 121 of the 142 Republicans who answered their party’s whip call were against the initiative or leaning that way. After an intensive lobbying campaign that featured personal intervention by O’Neill and (especially) Majority Leader James Wright, D-Tex., the Democrats were able to defeat the main Republican alternative, an amendment offered by Rep. Clarence Brown, R-Ohio, on a vote of 199-227. Based on whip counts from both parties, the leadership could only count on 143 votes in the days leading up to the roll call. The minimum necessary pickup was a full 70 votes, an enormous challenge for Democratic leaders. Still, they managed to pick up 84 additional votes and prevailed on the floor by a comfortable margin. During this stage of the fight over energy reform, the Republicans won on three polled items; the ad hoc committee’s proposed gasoline tax (which lost on the floor); an administrationback substitute on the gasoline tax sponsored by James J. Howard, D-N.J., (which also went down to defeat); and a minor proposal from Walter Flowers, R-Ala., concerning electric power plants (which was adopted). The natural gas item was the single most important issue, however. Here, as well as on other major aspects of the floor fight, the position of the Democratic leadership prevailed even though the whip counts indicated that O’Neill often needed to pick up fifty or more votes. After more than a year of bicameral conflict over the conference report, the energy bill returned to the House floor during the waning days of the 95th Congress. At this point, the key issue was procedural; whether or not to divide the package and consider the natural gas provisions separately from the remainder of the conference report. Democratic leaders favored a single vote; Republican supported separate consideration. The majority polled the matter on October 4, 1978, asking for member positions on the previous question on the rule and on the rule itself. Republicans began their whip count a week earlier, asking their members whether (1) they would support an effort to bring down the proposed rule and thereby separate the package, and (2) whether they would then vote against the natural gas portions of the conference report. Based on the whip counts, O’Neill could count on 182 Democratic votes. If we factor in the 13 21 Republicans who had signaled to their own leadership that they would be supporting the Democrats on procedure, the Speaker needed to pick up at least 12 more votes to win. The lobbying endgame continued right down to the wire. The White House attempted to help convince wavering Democrats to support O’Neill on the rule but congressional liaison was never Jimmy Carter’s strong suit. Thomas “Lud” Ashley, the Democratic chair of the Ad Hoc Committee on Energy, later described his experience working with the President during the floor lobbying process. I called Carter on one occasion…. I said, “Mr. President, … I’m going to lose this natural gas bill unless I can get some guys to change their vote. And I have a list of ten people that we can work with but we got to get ‘em and we got to get ‘em today.” He said, “It will be done. Thanks, Lud. Good-bye.” So I went to Tip with the same list. I got back to my office and there was a call from the White House and it was the President. I said, “What is it, Mr. President?” He said, “There is a natural gas title that is being considered in the energy bill and I have got to have your vote on it.” I said, “Does this sound familiar by any chance, Mr. President? For crissakes. I called you twenty-five minutes ago. How about crossing off the fellow at the top of the list and going from there? I think you got me.”21 O’Neill and the whips were more effective. The day of the vote, as the roll call proceeded, it became clear that the outcome could go either way. With the roll call tied at 206-206 and time running out, the deciding vote was cast by Rep. Robert Carr, D-Mich., who earlier had answered as “undecided” on all of the relevant whip counts. With O’Neill standing in the well of the House, watching him closely, Carr voted in favor of the motion on the previous question for the rule, the Democrats prevailed on procedure, and the conference report was adopted. By all accounts, the lobbying efforts of the House Democratic leadership were pivotal to the outcome. 4. Conclusion Our analysis of endgame lobbying on major floor roll calls in the House during 1977-80 provides clear evidence that parties and leaders can make a difference. Even on major issues toward the final stages of the House legislative process, large portions of the membership remain undecided about how to vote. As others have demonstrated, parties influence agenda setting and the deliberations that occur in committee. But they also exert a significant impact on the process 21 As quoted in Farrell (2001), 506-7. 22 through which rank-and-file members develop positions on policy issues on the floor. Much of what constitutes party influence, we believe, concerns the formation of preferences. The 1990s transformation of legislative studies has shifted scholarly attention away from member goals as theoretical primitives and toward member preferences as the central conceptual building block. In the words of one prominent practitioner, the spatial model has become “the workhorse theory of modern legislative studies” (Cox, 2001, pg. 189). Most applications of the spatial model to the field, however, posit that the preferences of legislators are fixed and exogenously determined. Not surprisingly, efforts to incorporate a significant role for parties in these models have focused on formal prerogatives such as agenda control, preferential rights to propose policy changes, the placement of parties within the temporal sequence of decision making, committee assignment powers, and so on. However, if parties and leaders play a key role in shaping the process of preference formation, then theories rooted in assumptions about fixed or exogenously determined preferences are only going to take us so far toward understanding the congressional parties. An alternative approach would be to revisit the distinction between goals and preferences, or at least to develop a more refined sense of what constitutes member preferences in the legislative process. Social choice theorists define “preference” as an ordering of concrete alternatives rather than as a point on an ideological scale or left-right continuum. The best observational research about roll call decision making demonstrates persuasively that legislators confront a range of potentially competing preference orderings as they go about their work (Kingdon, 1973; Fenno, 1985). These discrete orderings emanate from overlapping subsets of constituents, from the constellation of interest groups, from a legislator’s own views about good public policy, from party leaders, and so forth. The roll calls that members cast result from a complex cognitive aggregation of these potentially competing preference orders. In other words, the preferences that members form over concrete alternatives during the legislative process are induced preferences. Both conceptually and empirically, it is difficult to separate out the various components. As a result, attempts to distinguish analytically between preferences and parties are problematic, to say the least. It is not at all clear to us, for instance, that party influence should be defined as a divergence of policy outcomes from the preferences of the chamber median toward centrist preferences within the majority caucus. If the positions of dozens of majority party members 23 often are in play right up to the occurrence of major floor votes, precisely what constitutes the chamber median or centrist preferences within the majority caucus? Appropriate conceptualizations and empirical tests of party influence, we believe, should emphasize the movement of member positions over time. The analysis of party influence in Congress also needs to pay more attention to the role of the minority. If anything, the GOP leadership was more effective than were the majority Democrats at whipping rank-and-file party members into line. The influence that minority party leaders exert over the roll call decisions of their members may not be as pronounced following the 1980s increase in party polarization. Or the influence of the majority leadership may have increased to the point that it now dominates the coalition-building tactics of the minority. We will have to extend our analysis of jointly polled items to more recent Congresses to address the matter. But regardless of party attachments, a vote is a vote on Capitol Hill. Minority party leaders have many of the resources for influencing rank-and-file members that are available to the majority leadership. We cannot understand the role of party in Congress without considering both the majority and the minority. Finally, an important implication of this paper is that scholars need to gather more and better data about the evolving positions of legislators. Vote-based measures of member ideology can teach us a lot about the basic structure of the roll call record. Poole and Rosenthal (1997) is rightly one of the three of four most influential scholarly books about Congress published over the past decade. But the search for party effects in Congress – indeed, the study of influence in the legislative process more generally – would benefit from systematic evidence about the preferences of key actors prior to roll call choice. As we have demonstrated, whip count data are one valuable source of pre-vote evidence about member positions. There are others. Similar information can be gleaned from the votes that members cast in committee – committee roll calls have been publicly available for all panels since 1970. The historical records kept by the National Archives also include committee roll call data for some panels dating to the 19th Century. VanDoren (1990) demonstrates how the creative use of hearings and other congressional documents can be used to infer the content of member positions at the very beginning of the legislative process. Accumulating such evidence will enable us to better track how the policy preferences of members and other key actors develop over time, link the changes that occur to leadership strategies and the broader political environment, and hopefully gauge in 24 a more systematic fashion precisely who exerts influence in the legislative process. 25 Table 1. Items subject to both Democratic and Republican whip counts, 1977-80 Item Question Roll Call Date Roll Call Result Winner Financial Ethics Previous question motion Final passage 2/2/1977 Democrats Motion to concur to Senate amendments 5/17/1977 267-153: D264-15; R 3-138 205-217: D191-88; R 14-129 221-177: D 192-70; R 29-107 Final passage 6/7/1977 Democrats Final passage N/A 244-164: D 222-47; R22-117 No roll call occurred Substitute amendment Amendment 8/3/1977 Democrats Amendment 9/15/1977 Amendment 9/15/1977 Amendment 10/5/1977 Final passage 10/6/1977 B-1 Bomber Funding Amendment 10/20/1977 Agency for Consumer Protection Humphrey-Hawkins (Full Employment) Campaign Finance Reform ERA Deadline Extension Foreign Intelligence Surveillance Natural Gas Conference Report Panama Canal Treat Implementation Windfall Profits Tax Final passage of substitute Final passage 2/8/1978 Rule 3/21/1978 Final passage 8/15/1978 Amendment 9/7/1978 Previous question motion on rule Final passage 10/13/1978 Substitute amendment Amendment 6/28/1979 Rule 10/17/1979 Substitute amendment Substitute bill 11/15/1979 199-227: D 72-210; R 127-17 209-211: D 79-199; R 130-12 264-161: D132-150; R 132-11 223-193: D 97-178; R 126-15 168-247: D 65-210; D 103-37 257-163: D 221-59; R 36-104 194-204: D 80-183; R 114-21 189-227: D 172101; R 17-126 257-152: D 233-41; R 24-111 198-209: D 198-69; R 0-140 233-189: D 192-86; R 41-103 176-200: D 64-191; R 112-9 207-206: D 199-79; R 8-127 224-202: D 189-80; R 35-122 236-183: D 90-173; R 146-10 197-214: D 133124; R 64-90 228-182: D 206-51; R 22-131 234-166: D 99-158; R 135-8 255-118: D 138-99; R 117-19 Common Site Picketing Budget Targets Hatch Act Amendments Universal Voter Registration Natural Gas Pricing Minimum Wage: Youth Differential Minimum Wage: Tip Credit Minimum Wage: Indexing Labor Law Revision: Equal Time Labor Law Revision Amtrak Reorganization PAC Contribution Limits Hospital Cost Control Revenue Sharing 3/23/1977 9/15/1977 3/16/1978 6/21/1979 7/24/1979 11/13/1980 Republicans Bipartisan Republicans Democrats Republicans Republicans Democrats Democrats Democrats Republicans Democrats Republicans Democrats Democrats Democrats Democrats Republicans Ambiguous Democrats Republicans Republicans 26 Table 2. Distribution of responses on whip counts* Poll Response Democrats Republicans Yes 47.3% 64.0% Leaning Yes 7.2% 3.1% Undecided 16.1% 15.1% Leaning No 3.9% .9% No 14.5% 7.7% Other 11.1% 9.1% * Observations for Budget Targets and Amtrak Reorganization are not included because the Democratic and Republican leaderships did not endorse different positions. 27 Table 3. Roll Call Support for the party position by poll category Item Dem Y/LY 205-7 Dem U 26-0 Dem N.LN 6-7 Dem Other 26-2 GOP L/LY 112-2 GOP U 10-1 GOP N/LN 5-0 GOP Other 10-1 152-1 21-13 1-65 19-11 121-0 8-7 0-7 0-0 163-23 16-17 1-22 15-8 12-5 4-24 7-68 6-13 Hatch Act Amendments Natural Gas Pricing 174-6 20-9 7-25 26-8 97-2 7-3 5-8 11-9 129-5 34-7 13-44 35-17 119-2 7-6 0-7 1-2 Minimum Wage: Youth Differential Minimum Wage: Tip Credit Minimum Wage: Indexing Labor Law Rev: Equal Time Labor Law Revision 123-12 32-12 25-46 21-9 112-3 10-5 3-4 9-0 114-20 22-26 1-76 14-10 112-3 13-7 1-1 7-0 134-1 26-20 9-68 13-11 86-3 21-7 15-4 5-1 150-6 39-27 4-26 17-6 76-5 14-9 5-11 8-12 179-1 36-22 3-30 3-6 79-8 9-10 3-8 13-10 B-1 Bomber Funding Consumer Protection Humphrey-Hawkins (Full Employment) Campaign Finance Reform ERA Deadline Extension Foreign Intelligence Surveillance Natural Gas Conference Report Panama Canal Treat Implementation Windfall Profits Tax 138-3 17-10 12-63 20-9 87-1 13-5 5-13 14-4 138-19 26-25 3-54 11-7 118-1 8-6 2-11 0-0 172-2 40-9 4-33 21-1 103-6 10-12 0-7 2-1 165-17 19-22 4-26 17-9 123-0 13-0 2-0 4-0 166-2 16-21 4-58 8-6 71-0 9-5 16-33 7-3 104-18 49-18 10-16 37-15 70-2 28-5 2-2 20-0 167-11 14-9 14-58 6-2 66-2 38-5 12-1 14-2 144-6 22-20 6-39 17-15 83-5 22-15 5-15 12-0 124-10 20-18 4-37 25-25 89-0 30-5 7-3 20-2 78-28 27-28 18-60 13-11 56-7 16-19 12-21 6-17 147-12 31-19 13-13 18-7 91-2 14-4 6-12 23-4 99-15 34-25 13-45 16-16 73-1 34-4 10-0 21-3 64-29 13-30 11-52 13-30 78-2 11-10 11-6 17-1 Financial Ethics Common Site Picketing Budget Targets Amtrak Reorganization PAC Contribution Limits Hospital Cost Control Revenue Sharing 28 Table 4. Roll call retention by whip position and party* Poll Position Democrats Republicans 93.6% 97.5% 60.6% 71.5% 15.9% 42.9% 63.3% 79.8% 71.1% 87.1% Yes or Leaning Yes Undecided No or Leaning No Other All Categories * Observations for Budget Targets and Amtrak Reorganization are not included because the Democratic and Republican leaderships did not endorse different positions. Universal Voter Registration is also excluded because no roll call occurred. 29 Table 5. Probit analysis of voting behavior of “undecideds” Party A (n=1379) B (n=1368) C (n=1368) .838*** .796*** .321** -4.017*** -.260 Constituency DW Nom 1 -2.110*** DW Nom 2 -.540*** Constant -.569*** 2.018*** -.199 Correctly Predicted .642 .700 0.727 PRE .283 .397 0.450 Log likelihood -890.966 -825.906 -771.017 McKelveyZavoina R-SQ 0.135 0.255 0.346 ***p<.01, **p<.05 (one-tailed tests). Standard errors are adjusted for clustering on the poll/vote. Note: Observations for Budget Targets and Amtrak Reorganization are excluded because the party leaderships did not embrace different positions on the relevant questions. Universal Voter Registration is also excluded because no roll call occurred. The number of observations is slightly lower in models B and C due to missing values for the constituency ideology measure. 30 Table 6. Substantive significance of party across the three models Model Republican Democrat Difference A .285* .606 .321 B .300 .605 .305 C .411 .538 .127 *Cell entries are the probability of casting a “pro-Democratic” roll call for undecided members of each party under the alternative model specifications in Table 5, all other explanatory variables held at their means. 0 Prob of Voting Dem Position .2 .4 .6 .8 1 Figure 1. Party, “ideology” scores, and voting behavior of undecideds -1 -.5 0 DW-NOMINATE 1 Rep .5 1 Dem 31 Table 7A. Party coalition building: Democrats Intra-Party Base Initial GOP Support Actual Pickup Margin 5 Minimum Necessary Pickup -9 Financial Ethics 215 47 56 Common Site Picketing 154 7 51 44 -7 Hatch Act 191 13 1 40 39 Universal Voter Registration Natural Gas Pricing 131 3 83* -- -- 135 8 70 84 14 Min Wage Youth Differential Min Wage Tip Credit 141 7 62 63 1 135 2 76 24 -52 Min Wage Indexing 136 20 52 37 -15 Labor Equal Time 163 16 29 68 39 Labor Revision 185 12 14 60 46 B-1 Bomber 148 20 31 36 5 Consumer Protection 157 13 39 19 -20 Humphrey-Hawkins 176 8 21 73 52 Campaign Finance Reform ERA Extension 185 2 17 11 -6 169 50 -7 14 21 Foreign Intelligence Surveillance Natural Gas Conference 127 5 56 68 12 182 13 12 12 0 Panama Canal 154 21 39 49 10 Windfall Profits 139 10 61 34 -27 PAC Contributions 166 18 22 44 22 Hospital Costs 116 11 73 39 -34 Revenue Sharing 102 18 67 -2 -69 Item *Based on the assumption that all 433 of the individuals in the House at the time of the whip count would have participated in the roll call. NOTE: Cell totals differ from those in Table 3 because of members taking poll positions but not voting and the reverse. 32 Table 7B. Party coalition building: Republicans Intra-Party Base Initial GOP Support Actual Pickup Margin 14 Minimum Necessary Pickup 80 Financial Ethics 116 23 -57 Common Site Picketing 122 68 21 27 6 Hatch Act 101 33 70 30 -40 Universal Voter Registration Natural Gas Pricing 123 72 22* -- -- 121 57 36 21 -15 Min Wage Youth Differential Min Wage Tip Credit 116 72 23 21 -2 117 77 19 70 51 Min Wage Indexing 90 79 40 54 14 Labor Equal Time 83 30 95 55 -40 Labor Revision 90 33 87 40 -47 B-1 Bomber 90 78 32 26 -6 Consumer Protection 119 58 31 50 19 Humphrey-Hawkins 113 37 55 2 -53 Campaign Finance Reform ERA Extension 128 30 46 51 5 71 62 78 56 -22 Foreign Intelligence Surveillance Natural Gas Conference 78 29 82 69 -13 69 72 66 65 -1 Panama Canal 88 46 79 68 -11 Windfall Profits 89 41 80 106 26 PAC Contributions 95 28 82 59 -23 Hospital Costs 81 61 59 92 33 Revenue Sharing 96 71 20 88 68 Item *Based on the assumption that all 433 of the individuals in the House at the time of the whip count would have participated in the roll call. NOTE: Cell totals differ from those in Table 3 because of members taking poll positions but not voting and the reverse. 33 Table 8. Whip counts on the National Energy Act, 1977-78 Date 7/26/1977 7/27/1977 7/27/1977 7/27/1977 7/27/1977 7/27/1977 7/27/1977 8/1/1977 8/1/1977 8/5/1977 9/19/1978 10/4/1978 10/4/1978 Date 7/27/1977 8/2/1977 8/2/1977 9/28/1978 9/28/1978 Questions Polled by Democrats Will you vote for the previous question on the rule to accompany H.R. 8444, National Energy Act? Will you support the Ad Hoc Committee amendment on natural gas pricing? Will you oppose any natural gas deregulation substitute amendment? Will you support the Ad Hoc Committee amendment on business user taxes? Will you support the Ad Hoc Committee amendment on the gasoline tax? Will you oppose the Republican recommit motion? Will you vote for final passage of the bill? (pre amdts) Will you support the Howard gasoline tax amendment? Will you oppose the Jones amendment on plowback of crude oil tax receipts to oil producers? Will you vote for final passage of H.R. 8444, the National Energy Act? (post amdts) Will you support the Natural Gas Bill as reported by the Energy Conference Committee? Will you vote for the previous question on a rule combining the five energy reports into a single package? Will you vote for the rule combining the five energy conference reports into a single package? Questions Polled by Republicans Will you support the deregulation of new natural gas? Will you support the motion by Cong. Flowers to strike Sec. 548 of the energy bill, H.R. 8444, which, if enacted, would cause undue delay in construction of all electric generating plants and transmission facilities? Will you support the Republican substitute energy bill, H.R. 8555? Will you support an effort to separate the Natural Gas Bill from the rest of the energy package? If there is a separate vote on Natural Gas – Will you vote for or against it? 34 Table 9. Party coalition building on the National Energy Act A. Democrats Item Prev Ques Rule Ad Hoc Nat Gas Nat Gas Subst Ad Hoc Business Ad Hoc Gasoline Tax Motion to Recommit Passage (pre-amend) Howard Amdt Jones Amdt Passage (post-amend) Nat Gas Conf Prev Ques Rule Conf Rule on Conf Rept GOP roll calls in Support of Dems 5 --* 17 9 2 3 -6 15 13 -8 -- Intra-Party Base on Poll 211 140 135 114 77 179 173 65 131 220 238 182 161 Actual Margin 37 -14 11 -160 8 --129 12 33 -0 -- Dem roll calls in Support of Reps 72 127 29 79 -- Intra-Party Base on Poll 121 94 106 69 68** Actual Margin -15 48 -64 -1 -- B. Republicans Item Nat Gas Subst Flowers Amdt GOP Subst Rule Conf Separate Nat Gas * “--“ indicates that a roll call directly linked to the question did not occur. ** The cell entry for this GOP poll is the sum of “no” plus “leaning no” on the question as posed. 35 References Aldrich, John H. 1994. “A Model of a Legislature with Two Parties and a Committee System.” Legislative Studies Quarterly 19: 313-40. Aldrich, John H., and David W. Rohde. 1998. “Measuring Conditional Party Government.” Presented at the annual meeting of the Midwest Political Science Association, Chicago. Bach, Stanley, and Steven S. Smith. 1988. Managing Uncertainty in the House: Adaptation and Innovation in Special Rules. Washington, D.C.: Brookings Institution Press. Bartels, Larry M. 2000. “Partisanship and Voting Behavior, 1952-1996.” American Journal of Political Science 44: 35-50. Burden, Barry D., and Tammy M. Frisby. 2004. “Preferences, Partisanship, and Whip Activity in the House of Representatives.” Legislative Studies Quarterly: 29, 569-90. Cooper, Joseph, and David W. Brady. 1981. “Institutional Context and Leadership Style: The House from Cannon to Rayburn.” American Political Science Review 75: 411-35. Cooper, Joseph and Ulrich Sieberer. 2005. “The Importance of Majority Party Size in Congress,” manuscript, John Hopkins University, Baltimore, MD. http://www.jhu.edu/~polysci/faculty/cooper/MajorPartySize66.pdf. Covington, Cary R. 1987. “Staying Private: Gaining Congressional Support for Unpublicized Presidential Preferences on Roll Call Votes.” Journal of Politics 49: 737-55. Cox, Gary W. 2001. “Estimating Legislators’ Preferences with Roll Call Data: Introduction to the Special Issue.” Political Analysis 9, 189-91. Cox, Gary W., and Mathew D. McCubbins. 1993. Legislative Leviathan: Party Government in the House. Berkeley: University of California Press. Cox, Gary W., and Mathew D. McCubbins. 2005. Setting the Agenda: Responsible Party Government in the U.S. House of Representatives. Cambridge: Cambridge University Press. Dodd, Lawrence C. 1978. “The Expanded Roles of the House Democratic Whip System: The 93rd and 94th Congresses.” Congressional Studies 7: 27-56. Dodd, Lawrence C., and Terry Sullivan. 1981. “Majority Party Leadership and Partisan Vote Gathering: The House Democratic Whip System.” In Understanding Congressional Leadership, ed. Frank H. Mackaman. Washington, D.C.: CQ Press. Evans, C. Lawrence. 2001. “Committees, Leaders, and Message Politics.” In Congress Reconsidered, 7th ed, ed. Lawrence C. Dodd and Bruce I. Oppenheimer. Washington, D.C.: Congressional Quarterly, 217-43. 36 Evans, C. Lawrence. 2004. The House Whip System and Party Theories of Congress. NSF Proposal (SES-0417759), http://clevan.people.wm.edu/whip.htm. Farrell, John A. 2001. Tip O’Neill and the Democratic Century. Boston, MA: Little, Brown. Fenno, Richard F., Jr. 1973. Congressmen in Committees. Boston, MA: Little, Brown. Fenno, Richard F., Jr. 1985. “Observation, Context, and Sequence in the Study of Politics.” American Political Science Review 80: 3-15. Hetherington, Marc. 2001. “Resurgent Mass Partisanship: The Role of Elite Polarization.” American Political Science Review 95: 619-32. Jenkins, Jeffery A., Michael H. Crespin, and Jamie L. Carson. 2005. “Parties as Procedural Coalitions in Congress: An Examination of Differing Career Tracks.” Legislative Studies Quarterly 30: 365-89. Jones, Charles O. 1988. The Trusteeship Presidency: Jimmy Carter and the United States Congress. Baton Rouge: Louisiana State University Press. Jones, Charles O., and Randall Strahan. 1985. “The Effect of Energy Politics on Congressional and Executive Organization in the 1970s.” Legislative Studies Quarterly 10: 151-79. King, David C., and Richard J. Zeckhauser. 2003. “Congressional Vote Options.” Legislative Studies Quarterly 28: 387-412. Kingdon, John W. 1973. Congressmen’s Voting Decisions. New York: Harper and Row. Kingdon, John W. 1989. Congressmen’s Voting Decisions, 3rd ed. Ann Arbor, MI: The University of Michigan Press. Krehbiel, Keith. 1988. “Spatial Models of Legislative Choice.” Legislative Studies Quarterly 13, 259-319. Krehbiel, Keith. 1991. Information and Legislative Organization. Ann Arbor, MI: The University of Michigan Press. Krehbiel, Keith. 1998. Pivotal Politics: A Theory of U.S. Lawmaking. Chicago, IL: University of Chicago Press. Krehbiel, Keith. 2000. “Party Discipline and Measures of Partisanship.” American Journal of Political Science 44: 206-21. Krehbiel, Keith, and Alan E. Wiseman. 2005. “Joe Cannon and the Minority Party: Tyranny or Bipartisanship?” Legislative Studies Quarterly 30, 479-505. 37 Lee, Frances E. 2005. “Principled and Unprincipled Partisanship: Analyzing Conflict in Contemporary Senate Roll-Call Voting,” presented in the Department of Government colloquium series, College of William and Mary, September 23, 2005. Lawrence, Eric, Forrest Maltzman, and Steven S. Smith. 2005. “Changing Patterns of Party Effects in Congressional Voting,” presented at the annual meeting of the American Political Science Association, Washington, D.C. Mayhew, David. 1973. Congress: The Electoral Connection. New Haven: Yale University Press. McCarty, Nolan, Keith T. Poole, and Howard Rosenthal. 2001. “The Hunt for Party Discipline in Congress.” American Political Science Review 95: 673-88. Oppenheimer, Bruce I. 1980. “Policy Effects of U.S. House Reform: Decentralization and the Capacity to Resolve Energy Issues.” Legislative Studies Quarterly 5: 5-30. Petrocik, John R. 1996. “Issue Ownership in Presidential Elections, with a 1980 Case Study,” American Journal of Political Science 40: 825-50. Poole, Keith T. 2005. Spatial Models of Parliamentary Voting. Cambridge: Cambridge University Press. Poole, Keith T. and Howard Rosenthal. 1997. Congress: A Political-Economic History of Roll Call Voting. Oxford: Oxford University Press. Ripley, Randall B. 1964. “The Party Whip Organization in the United States House of Representatives.” American Political Science Review 58: 561-76. Ripley, Randall B. 1967. Party Leaders in the House of Representatives. Washington, D.C.: Brookings Institution Press. Rohde, David W. 1991. Parties and Leaders in the Postreform House. Chicago, IL: University of Chicago Press. Shepsle, Kenneth A., and Barry R. Weingast. 1995. Positive Theories of Congressional Institutions. Ann Arbor: University of Michigan Press. Sinclair, Barbara. 1983. Majority Leadership in the U.S. House. Baltimore, MD: Johns Hopkins University Press. Sinclair, Barbara. 1995. Legislators, Leaders, and Lawmaking: The U.S. House of Representatives in the Postreform Era. Baltimore, MD: Johns Hopkins University Press. 38 Smith, Steven S. 1989. Call to Order: Floor Politics in the House and Senate. Washington, D.C.: Brookings Institution Press. Snyder, James M., and Timothy Groseclose. 2000. “Estimating Party Influence in Congressional Roll Call Voting.” American Journal of Political Science 44: 193-211. Snyder, James M., and Timothy Groseclose. 2001. “Estimating Party Influence on Roll Call Voting: Regression Coefficients Versus Classification Success.” American Political Science Review 95: 689-98. Sullivan, Terry. 1990a. “Bargaining with the President: A Simple Game and New Evidence.” American Political Science Review 84: 1167-95. Sullivan, Terry. 1990b. “Explaining Why Presidents Count: Signaling and Information.” Journal of Politics 52: 939-62. Uslaner, Eric M. 1989. Shale Barrel Politics: Energy and Legislative Leadership, Stanford: Stanford University Press. VanDoren, Peter M. 1990. “Can We Learn the Causes of Congressional Decisions from Roll Call Data?” Legislative Studies Quarterly 15: 311-40. 39
© Copyright 2026 Paperzz