VIEWS: 44 PAGES: 59 POSTED ON: 5/30/2009
Revealin g Leade rship Th rough Roll Ca ll Analysis : Howard Baker as Senate Majority Leader 1981-1982 by Dr. Michael P. Bobic Emmanuel College firstname.lastname@example.org Prepared for the Southern Political Science Association Annual Conference January 8-11, 2004 Revealing Leadership Through Roll Call Analysis: Howard Baker as Senate Majority Leader 1981-1982 Leading the United States Senate is much like coaxing a timid cat from a cashmere sweater. Without patience, one can get scratched, get an upset cat, or get a shredded sweater. Senator Bob Dole (R-KS) once said leading the Senate is like taking kindergartners to the bathroom. Howard Baker faced an unusual challenge as Senate leader in 1981. The Senate, previously a collegial institution, had become more fragmented and partisan. The elections of 1980 gave Baker a Republican majority, but many of the new freshman Senators were extremely conservative members bent on changing the way the Senate did business. They were unlikely to compromise or form coalitions with more liberal members, which in effect reduced Baker’s ability to get legislation passed. On the other hand, Ronald Reagan’s election by such a stunning margin gave Baker leverage against moderate Democrats. By most accounts, Baker excelled at the position of majority leader. Scholars and Senate observers as early as 1982 were comparing Baker to such great Senate leaders as Lyndon Johnson and Mike Mansfield (Arieff, 1981; Tolchin, 1982). The anecdotal evidence indicates that Baker was remarkably successful in adapting to his new role. His leadership of the Senate during the 97 th Congress was a time of great personal and political success. The empirical evidence gleaned from the roll call data for the 97th Congress supports this perception. Despite deepening ideological divisions within his party, a decreased sense of bipartisanship (Arieff, 1981; Ornstein, Peabody and Rohde, 1989), changes in Senate procedures and norms (Ornstein, Peabody and Rohde, 1989), and the relative inexperience of the Republican Party in leading the Senate, Baker has an enviable record for building coalitions of members, creating party cohesion, and winning on the floor of the Senate. This paper presents evidence to support these claims. The first section provides a brief historical overview of the 97th Congress, focusing on the policy and institutional situation Baker faced. The second part of the analysis presents results from an analysis of roll call voting patterns in the 97th Congress. This analysis reveals that Baker was remarkably successful in unifying Republicans and winning votes on the Bobic, SPSA 2004 Revea ling Lea dership Page 2 floor. Much of Reagan's success can be attributed to Baker's efforts on behalf of the president. The final section summarizes the main findings. 1 Baker as Majority Leader, 1981-1982. The elections of 1980 consummated a long trend of Republican gains throughout the nation including such Democratic strongholds as the south and midwest (Bailey, 1988; Dodd and Oppenheimer, 1990; Ornstein, Peabody and Rohde, 1989). The Republican capture of the Senate for the first time in over two decades promoted Baker to the position of majority leader in an uncontested party election in January of 1981 (Annis, 1995). Of the fifty-four Republicans serving in the Senate in 1981, fu lly thirty-three were in their first term, and sixteen of those had been elected in 1980. These freshmen were much more conservative than most of their more senior colleagues, including Baker himself (Ornstein, Peabody and Rohde, 1989). Many were also political "amateurs" (Overby, 1993): eight of the sixteen senators elected in 1980 had never held public office before. They were very loyal to Ronald Reagan (Baker, 1989) but few of them knew Baker (Ornstein, Peabody and Rohde, 1989) and most did not understand the way the Senate functioned (Hurley, 1991 ). These freshmen entered the Senate committed to bringing very conservative social issues such as abortion, school prayer and bussing to the floor for debate (Ornstein, Peabody and Rohde, 1989). They believed that they had been elected to pursue a specific set of policies, and because they did not intend to serve more than one or two terms, they wanted to act on these policies as quickly as possible (Hurley, 1991). A core of senior members, who were for the most part more moderate or liberal than the freshmen, opposed their efforts to bring these issues to the floor (Ornstein, Peabody and Rohde, 1989). In order to pacify both wings of his party, Baker promised that if freshmen would delay bringing their measures to the floor for debate until the Senate had cleared most of the Reagan agenda, Baker would see 1 This project is part of a larger study of H oward Baker’s leadership from 1977 to 1984. Because I hav e extracted this from a larger work, some informat ion pres ented w ill be nec essarily a bbrevia ted. Fo r a fuller disc ussion of the va riables an d meth odolog ical issue s, conta ct Mic hael Bo bic at mbobic@ eclions.net. Bobic, SPSA 2004 Revea ling Lea dership Page 3 to it that a free floor debate and vote was taken on each of these issues (CQWR, September 12, 1981). Members generally abided by this agreement for most of 1981. Baker's task of coordinating this fragmented party was made much easier by the leadership and abilities of the newly elected Republican president from California. Unlike his predecessor(s), Reagan espoused a philosophy of smaller government, greater incentives for private industry, foreign relations based upon realism and a powerful military (Ripley, 1988; Malbin, 1989). His economic agenda was "the largest and broadest piece of legislation ever considered by Congress" (Tate, 1981a). Although most observers argue that Reagan's agenda was "limited," Ripley (1988) properly notes that the agenda was not so limited in scope or ambition. The Omnibus Reconciliation Package which contained the key provisions of Reagan's economic agenda (SCR 9 in 1981) touched on almost every program or agency in the federal government, usually by cutting budgets or eliminating programs. Reagan's focus on cutting government may have seemed narrow, but its scope was staggering. In order to understand the policy dynamics affecting Baker's leadership a brief overview of that agenda seems warranted. What follows is a summary of President Reagan's key domestic and foreign policy agenda items. Domestic Policy in the 97th Congress. Reagan's domestic agenda was founded on the Economic Recovery and Tax Act (ERTA) of 1981 which contained a massive reduction of personal income taxes, changes in depreciation rates for businesses, a cut in capital gains taxes and the indexing of tax rates to the inflation rate. It also lowered corporate tax rates, added small increases in taxes on the middle class and generally attempted to scale back the growth of federal revenues (Rudder, 1989). One case in which divisions in the Republican Party emerged dealt with a Reagan proposal to cut $400 million from benefits to veterans. As many Republican senators offered amendments to restore these funds as did Democrats, although Baker, Reagan and Republican leaders were able to defeat each of them (Gregg, 1981). Bobic, SPSA 2004 Revea ling Lea dership Page 4 Although much of what Reagan wanted in economic policy passed, the battle over the 1982 budget took up most of 1981, and many of the other agenda items were left for the 1982 session. This would mean that 1982, already looking to be a difficult year as the recession of 1981 worsened and unemployment rates climbed, would be a session in which Republicans would face many difficult choices and divisive issues (Tate, 1981c). Despite appearances, the Reagan agenda in 1981 was far more than a set of economic policies. Reagan also attempted to place the Social Security program on a secure financial basis (Tate, 1981b) and tried to eliminate the Legal Services Corporation, a federal program which provided free or low-cost legal assistance to poor citizens (Cohodas, 1981a). Other moves included placing many federal social programs such as the food stamps program into block grants (Donnelly, 1981a), attempting to eliminate the Department of Education (Annis, 1995), and reforming the Clean Air Act (Annis, 1995). Reagan's domestic agenda in 1982 was an ironic reversal of his efforts at reform in 1981. The passage of the ERTA of 1981 was followed by a deep recession and a sharp increase in the federal deficit. Senators quickly came under intense political pressure to do something about the economy. Initially, Reagan ignored signs that he would have to allow a tax increase and sacrifice continued expansion of the military, but pressure from Baker and other Senate Repu blicans eventually persuaded him to change his position (Tate, 1982a). As a result, the largest tax cut in history in 1981 was followed in 1982 with the largest single tax increase in history, the Tax Equity and Fiscal Responsibility Act (TEFRA) of 1982 (Rudder, 1989). This act scaled back tax breaks for corporations, created an accelerated schedule for paying off the deficit, and provided for an increase in the minimum tax rate for those earning more than $50,000 per year (Rudder, 1990). Baker and the Repu blicans did manage to hand the President a small victory by keeping the cuts in individual taxes. A key point in Reagan's 1980 campaign was a push for a balanced budget amendment to the Constitution. In 1982, the Senate debated and passed such an amendment. Although debate was heated and many amendments were offered, Republicans defeated most of them. In its final form, the bill called Bobic, SPSA 2004 Revea ling Lea dership Page 5 for a balanced budget at the beginning of each fiscal year unless a three-fifths majority of Congress agreed to deficit spending, with the proviso that the amendment could be waived in time or declared war (SJR 58; CQ vote 288; CQ Almanac, 1982; 49S). Conflict in the Senate meant that of the 13 appropriations bills necessary to fund the government, only five had passed by late September 1982 (Plattner, 1982a). As elections drew nearer, Jesse Helms and other conservative members used the continuing resolution (a temporary funding measure whose passage was necessary to keep the government functioning) to pursue their social agenda. These conflicts eventually forced Baker and Reagan to call for a lame duck session after the November elections (Tate, 1982b). This session was designed to deal with the remaining appropriations bills and perhaps to finish work on the continuing resolution. A filibuster by Don Nickles of Oklahoma, Jesse Helms of North Carolina and Gorton Humphrey of New Hampshire.(Sarasohn, 1982) against a proposed increase of 5 cents per gallon to the gas tax (HR 6211) threatened to bring the session to a halt. Baker attempted to break the filibuster by placing the continuing resolution on the schedule to be debated after the debate on HR 6211, but this effort, as well as efforts to invoke cloture failed. Debate lasted from December 13 to December 23, when the Senate finally invoked cloture and passed the measure. The lame duck session typified the events surrounding domestic policy in 1982. What had started out as a promising and surprisingly successful presidency in 1981 faltered in 1982, grounded by a slow economy, a Senate Repu blican Party divided over social issues, and a president who was somewhat less involved in the policy process than he had been in 1981. The Reagan domestic agenda focused primarily on the policy arenas of government management of the economy, social welfare, and civil liberties. Although issues like abortion, bussing, school prayer and an extension of the Voting Rights Act of 1965 also found places on the agenda, Reagan and Baker made it clear that they wanted the Senate to clear the Reagan economic programs first. As majority leader, Baker had considerable discretion in pursuing that goal. Bobic, SPSA 2004 Revea ling Lea dership Page 6 Foreign Policy in the 97th Congress. There were essentially four main thrusts to Reagan's foreign policy in 1981: a harder line with the Soviet Union, increased involvement in Latin America and the Caribbean Basin, expansion of the military, and increased foreign military and economic aid (Annis, 1995; Malbin, 1989). On the defense side, furthermore, Reagan pushed development of the MX ballistic missile system (Davidson and Oleszek, 1990), the B-1B strategic bomber program (Congress and the Nation, 1984), and the expansion of the navy by at least two aircraft carriers. Reagan also increased proposed spending for foreign aid, a move that many in the Senate found hypocritical, given Reagan's desire to cut domestic programs so dramatically. Reagan fought for his foreign aid bill, the first such bill to emerge from the Senate in over three years at that time (Whittle, 1981). Baker defended the bill, arguing that without approval of these funds, the US would not be able to meet its international commitments, and that the aid would foster good will for the US abroad (Whittle, 1981). Finally, a review of Reagan's foreign policy agenda would be incomplete without a mention of his successful bid to sell advanced aircraft to Saudi Arabia despite substantial opposition. The Reagan foreign policy program of 1982 was a continuation of his programs in 1981. The only new agenda item was an effort to tighten the law covering those who exposed the identity of US agents serving in other countries (CQ vote 53; S391). Other foreign policy issues arising in 1982 included an effort by Congress requiring that the president keep US strategic stockpile of petroleum reserves filled at a certain rate (300,000 barrels a day) (Plattner, 1982b) Thus, Reagan's 1982 foreign policy program followed the course laid out in 1981. His efforts to provide stronger defenses against the Soviet Union were matched by his efforts to increase the US presence in the Caribbean and in Latin and South America. His willingness to associate with the leaders of El Salvador and Panama signalled an end to the emphasis on human rights of the Carter foreign policy and a return to a policy based on rea lism. Overall Reagan was more successful in winning support for his foreign policy initiatives, even among his Democratic opponents. Bobic, SPSA 2004 Revea ling Lea dership Page 7 The discussion of the events surrounding Reagan's 1982 agenda indicate that the political environment in which he and Baker were laboring had changed. In 1981, Republicans were willing to set aside their policy differences to pursue the Reagan agenda. Democrats, shocked by their loss of control of the Senate and wary of the power of the president's electoral victory had been very cooperative in passing his programs. However, when the economy did not respond as Reagan officials had predicted, both Democrats and Republicans began to chart their own political courses. Thus, the leadership task in 1982 appeared to be more difficult than in 1981. Most commentators note that Reagan's stunning success in 1981 was never matched for the rest of his presidency (Ripley, 1988; Fleisher and Bond, 1983; 1993). Party unity and cohesion suffered as well. Baker seemed to lose control of the Senate late in 1982, and as the lame duck session collapsed into filibusters and late-session maneuvers, many people saw this as a sign that something was wrong with the Senate. However, the empirical questions remain: was Baker successful in leading the Senate in the 97th Congress? Did his fortunes change over the course of the session? Was Baker able to leave discernable imprints on voting patterns found in the 97th Congress. These questions are addressed in the following sections. Senate leadership involves a number of activities, but most scholars argue that the key measure of leadership is the ability to get votes passed on the Floor of the Senate (Oleszek, 1989; Docc, 1983; Arnold, 1990; Ripley, 1988). Senate leaders are expected to affect votes. Patterson argues that effective leaders invest time and energy discovering and affecting the vote intentions of party memb ers (Patterson, in Kornacki, 1990; 46). Sinclair (in Deering, 1989; 135-36) argued that constituents expect Senate leaders to pass legislation. Hurley (in Deering, 1989; 150) indicates that Senate leaders use their floor scheduling powers to affect vote outcomes. In simple terms, a large part of Senate leadership involves building coalitions of members, holding those coalitions together on the floor, and winning votes. The following section explores Baker’s ability to build coalitions among Senators in 1981 and 1982. The data for this analysis include all party-line roll call votes taken in the 97th Congress. These roll Bobic, SPSA 2004 Revea ling Lea dership Page 8 call votes were then divided into the standard policy categories as proposed by Aage Clausen (Clausen, 1973. These policies included agricultural, social welfare, foreign policy, government management of the economy, and civil liberties policies. However, recent work indicates that governments also propose and dispose of legislation dealing with how government does its job-- how government operates (Lowi, 1972; 1974). Thus, I added a category to represent policies dealing with such issues as committee organization, member pay (HJR325, 1981), and of course SR 20, Howard Baker’s unsuccessful effort to televise Senate proceedings. Building Coalitions of Agreeme nt in the 97th Congress. The evidence presented below supports the belief that Baker was able to build coalitions of members successfully in 1981 and 1982. However, some of the expectations presented in the anecdotal discussion of the session are not borne out. The Democratic Party ma y have supported Reagan's proposals on final passage, but even in 1981 there is little indication that they were playing any role other than loyal opposition. To develop an understanding of Baker’s ability to build coalitions, I computed a PHI coefficient for every memb er and Baker, across all six policy categories. Once I had these PHI scores for every policy type, I used a univariate clustering routine developed by Hamparsum Bozdogan (1990) to identify unique clusters of members within each policy type. The results of the cluster analysis are presented graphically in Figure 1. Figure 1 Here Three points should be clear. First, Baker fa ced different clusters or coalitions of Senators in each policy type. Second, those policy coalitions differed from year to year. Finally, the level of support Baker had varied, from relatively cohesive and high support in tax policy in 1981, to dispersed support in 1982 in foreign policy. The clustering routine used revealed that the PHI scores were homoskedastic, but that the mean clusters were significant. The clusters discovered were from a possible maximum clustering of 9. Describing Coalition Structure. Bobic, SPSA 2004 Revea ling Lea dership Page 9 Cluster analysis of agreement with Baker's positions indicate that even in 1981, the structure of support and opposition was very much bimodal. Table 1 presents the cluster solutions for 1981 and 1982 using the information theoretic routine which generated the graphs in figure 1. The findings in Table 1 seem to contradict the historical evidence of Democratic support for Ronald Reagan in 1981 and a breakdown of Republican unity in 1982. Were those expectations met, one should see odd numbers of clusters in 1981 and only a few two-cluster solutions in 1982. The tables do not bear out these expectations. In 1981, for example, three arenas show two clusters, four show even numbers of clusters. A graphical plot of foreign policy, with five clusters, nevertheless demonstrates stray bipolar patterns (see Appendix), however the cluster solutions do not indicate party membership. If the split is along party lines, then the historical assumption of Democratic support for Reagan seems in error. They may be have voted with Reagan on final passage, but the clusters hint that Democrats were still the "loyal opposition". The CAIC values indicate a great deal of volatility in cluster patterns for 1981. Even though agriculture, government management of the economy, and government operations all have two clusters, those clusters are not very similar at all. Government management of the economy's CAIC value (202.8) is much lower than Agriculture or Operations indicating much less variance about the cluster means than for agricultural or government operations issues. Variance measures the spread of a distribution about the mean: the low government management CAIC indicates that the poles supporting and opposing Reagan's budget policies were strongly committed to those positions. The appendix presents a graphical interpretation of these variances. Civil Liberties has the lowest CAIC value, but there were only six votes in this category and the results are uninformative. Insert Table 1 Here The dominant cluster solution for 1982 is two groups, with three clusters in social welfare and four in government management of the economy. These results contradict the anecdotal evidence of the Bobic, SPSA 2004 Revea ling Lea dership Page 10 breakdown in party unity, although the four cluster solution for government management does reflect the great conflict surrounding Reagan's 1983 budget. The three cluster solution for social welfare is actually composed of two groups centered at .6 and -.6, with a small middle group centered at zero. Close examination of the CAIC values sheds light on the apparent contradiction between these results and the historical evidence. In every case except for social welfare, CAIC values for 1982 are greater than those for 1981, even if the number of clusters is smaller. Foreign policy, for example, went from a five cluster solution in 1981 to a two cluster solution in 1982, but the CAIC value increased by five points. This pattern of variance reveals that while there may be two dominant clusters of support, cluster members were much less unified in 1982 than in 1981. Assuming those clusters are composed of partisans, these data confirm a breakdown in Republican unity. To determine whether or not the assumption of a partisan split in the clusters is warranted, a closer examination of these clusters is necessary. As in the 95th Congress, the characteristics of interest for these members include freshman status, party identification, and ideology (presented in Table 2). The data for 1981 confirm some expectations about member behavior and challenge others. First of all, Baker could count on voting support from an average of 52 senators. In contrast, those who voted in disagreement constituted a group of only 34 senators on average, with 14 members on average remaining neutral or indifferent. Baker had support from enough individual members to win most votes most of the time. The partisan composition of Baker's support coalition presents a clear picture of Democratic behavior. For the most part, Baker's support did not come from Democratic defectors. Only in agricultural and civil liberties policies could Baker count on a large bloc of Democrats. In most arenas, he had fewer than two Democratic senators to vote with him noticeably frequently. Baker could count on almost all of the 16 Freshman Republicans elected in 1980. On average, 15 of the 16 freshmen senators voted with Baker most of the time. This support would lead one to expect Bobic, SPSA 2004 Revea ling Lea dership Page 11 a very conservative cast to Baker's core supporters, but that was not the case. Baker's core support was not as conservative as it was in the 95th Congress. The minimum average ADA occurred in social welfare policies, with a score of 24.0. The maximum ADA score occurred in agricultural policies, with a value of 38.7. Overall, the values generally varied little from each other across policies, which would indicate that membership in the different policy clusters was not as volatile as it was in the 95th Congress. Thus, Baker's core support in 1981 was composed of moderate to conservative Republican senators. While he had support from almost all of the freshmen elected in 1980, that bloc of members did not place an ideological stamp on his core support. Table 2 Here The results for 1982 are different from those in 1981. First, those who voted in agreement with Baker (a PHI score close to one) constituted a voting bloc of only forty-eight senators on average. Fortytwo senators generally formed a cluster of members far from Baker's preferred position. This mean difference meant that on most floor votes, Baker would have enou gh support to win, albeit by a slim majority of only six votes. This is very different from the majority of eighteen senators he held in 1981. The characteristics of supporters remained much as it was in 1981. Baker never garnered more than nine Democratic supporters in any one policy arena in 1982. Of the 18 new members elected in 1980, Baker could count on 15 most of the time in any policy arena, much as he could in 1981. Baker's core support was composed of more conservative senators than had been the case in 1981, however. Table 3 presents a summary of characteristics of cluster members in 1982. Baker supporters, therefore, appear to be moderate to conservative Republicans, including most freshman members. The consistency in ADA scores and the smaller variations in CAIC scores in Table 1 give strong indications that unlike the 95th Congress, support for Baker was not only stronger (as would be represented by higher mean PHI scores), but was also more consistent. Members who supported Baker in one policy arena appear to support him in other arenas as well. However, these data cannot confirm that expectation. Bobic, SPSA 2004 Revea ling Lea dership Page 12 Table 3 Here Baker's leadership task would be a great deal easier if the same senators supported him across different arenas. To determine whether that support base remained stable across policy arenas, a test of the relative rankings of members across policy categories was conducted, using the Kendall's Coefficient of Concordance. Testing all 100 senators' rankings in the six policy categories for 1981 produced a W coefficient of .5616 (P2 <=99 =333.62, p=.000). This is significant at "=.01 indicating that member rankings are similar across the six policy arenas in 1981. This finding supports the Smith (1981) and Poole and Rosenthal (1984) hypothesis that the Congress was becoming a more partisan and ideologically divided institution. To understand the nature of member consistency in 1981, Table 4 presents the top twenty senators in each policy area. A cursory examination of the table indicates that many of the same names appear in much the same locations in the table. The rankings still show interesting differences. One good example is Bob Dole. Dole appears in the top 20 on four of the lists in 1981. His ranking is never lower than tenth, and twice he ranks first, but he does not rank in the top twenty in either foreign policy or civil liberties issues. Table 4 A test of senator rankings in 1982 produced similar findings. The W score of .5649 (P2 <=99 =335.52, p=.000) also demonstrates significant agreement in member rankings across policy arenas. In fact, the ranking results for 1982 show a small increase in consistency over 1981. If Republican defections were an indication of underlying fractures in Republican unity, this result is unexpected. The rankings support the historical evidence of a breakdown in Republican discipline in 1982. For example, Bob Dole ranked no lower than eighth in any arena in 1982, but he only appears in three of the six policy dimensions. Dole is not ranked in the top twenty supporters in foreign policy, civil liberties or government operations issues. Pete Domenici (R-NM), who appeared in only three policy rankings in 1981, appears in all but civil liberties policies in 1982, but for the most part, ranks among the lowest supporters. Bobic, SPSA 2004 Revea ling Lea dership Page 13 Comparing the two lists, one finds that many of the same names appear in similar policies in both years. In agricultural policies, fourteen of the top ranked members appear in both 1981 and 1982. Fourteen names are also common in social welfare issues in both years. Both government management of the economy and foreign policy dimensions reveal thirteen names in common from 1981 to 1982. The rankings appear below, in Table 5. It follows the same format as Table 4. Table 5 Civil liberties and government operations seem to be the critical arenas for breakdowns in party unity. Only five names are common to both civil liberties lists, and only nine to government opera tions. The breakdown in consistency in civil liberties is understandable, but the difference in the lists for government operations is unexpected. The tables of member rankings include the mean level of support these twenty senators offered Baker in 1981 and 1982. In agricultural policy, an area in which Baker had little interest, the mean PHI score was .69. The data reveal that the average member on the list would vote with Baker on almost eighty percent of votes taken in this policy dimension. PHI scores for other policy arenas likewise show high levels of support. The data for 1982 indicate that support usually declined. While Baker had some success in controlling debate over issues like abortion, school prayer and bussing, one can see from the decrease in civil liberties scores that these debates magnified the divide in the party along ideological lines. Baker could count on the average member eighty percent of the time in 1981, but he could count on that member only sixty or sixty five percent of the time in 1982. Foreign policy voting support also declined, due to efforts to reduce the rate of growth in military spending. These data present a good summary of the characteristics and strengths of Baker's core support, but they do not reveal the dynamics creating these levels or patterns of support. To do that, a different methodology must be employed. By modeling support scores (PHI scores) as a function of partisan, Bobic, SPSA 2004 Revea ling Lea dership Page 14 personal and environmental pressures legislators face, a greater understanding of that support, should be possible. Modeling Coalition Structure. The support scores in each policy dimension are modeled using a common set of variables exploring partisan, personal and environmental pressures Senators faced. The variables used to explain support for Baker (high PHI scores) are presented in List I. To explore the foundations of Baker’s coalition support, member scores in each policy dimension are regressed against the twenty-two variables defined above. An information theoretic scoring routine determines which variables best capture variations in the phi scores per policy arena. The resulting models are solved simultaneously through SUR modeling techniques (Zellner, 1963; Hocking and Smith, 1968; McElroy, 1977; Schmidt, 1977). The results of this analysis are as follows. The models to be entered into the SUR equations are generated from an all possible models selection procedure in which variables are kept or rejected according to their contribution to model fit. The models were selected based upon the minimum Consistent Akaike’s Information Criterion value (Bozdogan, 1987) The selection procedure generated the regression models presented below: Best Models to Explain PHI scores, 1981 and 1982. 1981: AG SOCWEL GOVMAN FORPOL CIVLIB GOVOPS Party, Democratic Committee Chair, % in State Employed by Federal Government, Party Unity Party, Medicaid Spending, Social Program Expenditure Rank, ADA, Party Unity, Presidential Support. Party, Tenure, Party Unity, Conservative Coalition. Party, Tenure, CCUS, Presidential Support, Party Unity, % in Government Employment. Party, Percent African-American Population, Presidential Support, Conservative Coalition, AFLCIO. Party, Tenure, Party Unity. 1982: AG SOCWEL GOVMAN FORPOL CIVLIB GOVOPS Party, Republican Leader, Democrat Committee Chair, ACA. Party, Tenure, Percent Government Employment, AFL-CIO, Party Unity. Party, Tenure, Repub lican Leader, Percent in Government Em ployment, AFL -CIO, Medicaid Spending, Party Unity, Presidential Support, Conservative Coalition. Party, South, CCUS, Presidential Support, Party Unity. Party, Republican Leader, Republican Committee Chair, % in Government Employment, State Population, AFL-CIO, Party Unity, Conservative Coalition. Party, Republican Leader, AFL-CIO, Party Unity, Conservative Coalition. Bobic, SPSA 2004 Revea ling Lea dership Page 15 The maximum likelihood estimation procedure used determined that these models were most likely to have generated the observed PHI scores observed in the data. These models seem to indicate that the dimension of most importance will be the personal dimension, followed by the partisan dimension. Note that party unity (PARTYUNI) appears in almost all the models as a key variable, confirming the historical and Poole and Rosenthal evidence of increased partisanship in the Senate. One would expect from the W scores that the intermodel correlations would be quite high. On the other hand, given the observed differences between rankings, one might expect that the intermodel correlations would be quite low. The results of the modeling process confirm the second expectation: there is very little correlation in member behavior across the different policy types. Table 6 presents the intermodel correlation matrix for 1981. Table 6 here The model coefficients produced indicate that the personal dimension seems to carry the most explanatory power in understanding variations in PHI scores. Appeals to party unity appear to have been quite successful in increasing support for Baker's position. In every policy arena in which it appears, party unity is a significant variable. Personal characteristic variables such as ideology are significant as well. Table 7 presents the results of the SUR analysis of the coalition scores. The first column of Table 7 contains the three dimensions of pressures. The variables operationalizing those dimensions appear in column 2. The successive columns after that contain the coefficients and standard errors for the different policy dimensions. Coefficients significant at "<.05 are double lined. The bottom two rows of the table present the univariate F-test and R-squared statistics. If a variable was not selected for inclusion in the model originally, the cell corresponding to that variable is left blank. Note that due to rounding of the standard errors (in parentheses), it may not appear that a boxed term is significant. Bobic, SPSA 2004 Revea ling Lea dership Page 16 Partisan dimension variables do not perform particularly well. It is only in the dimension of social welfare that party identification makes a contribution to changes in PHI scores. This could be an indication of Democratic defection, or it could be an indication of weakened party loyalty among senators. Environmental factors, especially economic ones, do not seem to exert that much influence. In fact, this dimension exerted so little influence that most of its variables were eliminated in the model selection stage. When environmental variables are included in a model2 , they are statistically significant. Support for Baker in social welfare policies was in part a function of a state's expenditures for social welfare programs. The more a state spent for social support programs, the less support its senator(s) gave to Baker and Reagan. This finding is not unexpected, given that Reagan had proposed to make many of these programs block grants rather than federally administered programs. While states would gain more control over and responsibility for these programs, funding for them would be reduced. This was a financial burden many governors opposed. The different models highlight interesting patterns within policy dimensions. For example, Baker's support in agricultural issues was primarily partisan, although he did gain support from Democratic committee chairs. Baker supported Reagan's Farm Bill revisions in 1981, which probably accounts for this finding. Social welfare issues were a very diverse lot in 1981, but Baker's support was generally partisan and ideological. The amount a state spent on social welfare programs tended to decrease support for Baker's positions, as did increased medicaid spending. Government management of the economy issues were, as one would expect, very partisan, with party and party unity dominating other factors. Support for Baker on foreign policy matters was a function of personal factors such as tenure (older members were more supportive) and of environmental factors such as the percent of a state's workforce who were employed by the government. Not surprisingly, support for Baker's position also 2 Social Welfare and Civil Liberties arenas. Bobic, SPSA 2004 Revea ling Lea dership Page 17 correlated to party unity voting. Baker's support in civil liberties and government operations issues was almost entirely partisan and ideological. The model fit statistics are quite strong. The Adjusted R-Square statistics indicate that these models explain more than half of the variations in these models. The models which do not fit well are in agriculture and civil liberties, which is to be expected. The F-tests show a significant model fit in all cases. However, it is important to notice that even after a rigorous model selection procedure, most model terms were nonsignificant. Personal considerations were the most consistent influence in determining support for Baker's positions. Table 7: SUR Solutions for Baker's Support by Policy Arena: 1981 The data for 1982 follow similar patterns. The intermodel correlations (Table 8) are not very strong and many more of them are negative, which indicates that increases in support in one arena would indicate decreases in support in another. Overall, this table and the one for 1981 belie the findings of the cluster and rankings analyses: despite growing polarization between the parties and greater consistency in member rankings, senators still reacted to different policies in different ways. This finding certainly challenges the Poole and Rosenthal hypothesis of unidimensional voting in Congress. Table 8 here The models in 1982 tell a marginally different story than those in 1981. In 1982, the partisan dimension appears in four of the six policy dimensions. Party variables achieve significance in three policy arenas. Republican party leaders supported Baker more strongly than the rank and file in government operations issues, while committee chairmanship matters in civil liberties issues. Oddly, the committee chairs generally offered less support for Baker's positions in civil liberties issues than did the average senator, with a mean PHI score .25 points lower than the average senator's score. Not only were Republican committee chairs less supportive, overall Republican support for Baker's positions in civil liberties issues was .845 points lower than their Democratic counterparts. This result is clearly a function of Baker's failed efforts to invoke cloture on several filibusters over abortion, school prayer and bussing. Bobic, SPSA 2004 Revea ling Lea dership Page 18 The personal dimension also appears to have exerted important influence on voting decisions. However, unlike 1981, in which ideological measures seemed most consistently to affect voting, it appears support for the president in 1981 predicted voting patterns in 1982 almost as consistently as 1981 AFLCIO voting support scores. Party unity is by far the most consistent term appearing in the data. The coefficients for party unity are somewhat larger in 1982, indicating a bigger impact on voting decisions in 1982 than in 1981. Once again, environmental pressures failed to appear that frequently, but they were significant when they did appear. Medicaid spending levels increased support in government management of the economy issues, unlike 1981. A distinct regional effect appeared in foreign policy matters. This is the first time in which this model term appears and contributes to Baker's support. This lack of impact from environmental factors may seem puzzling at first, but recall that much of Baker's support is derived from the 16 freshmen Republicans elected in 1980. Several studies of these members since 1986 indicate that some of them entered the Senate intent on being "amateurs" in the legislature. These members believed they had been elected to pursue specific policy changes and were not overly concerned with constituent service (Overby, 1993). They paid little attention to their constituents, making few trips home and spending relatively little on franked mailings. The model coefficients in Table 9 generally support expectations about why members supported Baker's positions. His support in agriculture, social welfare, and tax issues was primarily driven by partisan and ideological factors, while civil liberties and operations found some environmental factors important. Being a fellow Southerner only increased support for Baker's position in issues of foreign policy. Party identification was a consistent predictor of support, while committee position is not. This is probably a result of high party unity among Republicans. The model diagnostics indicate that all policy dimensions were modeled quite well: in most cases the model R-squared was .70 or greater. This likely was due to a number of factors: a president who had a small agenda and Baker's leadership position and his successful appeals to party unity and increased Bobic, SPSA 2004 Revea ling Lea dership Page 19 ideological divisions in the chamber. Once again, the personal dimension was most consistent in affecting support. Table 9 Here A comparison of the results for 1981 and 1982 reveals several patterns. Agricultural policy in both years was most affected by partisan factors. Republicans were less supportive of Baker's positions in social welfare in both years, although in 1981, support for Reagan translated into support for Baker. Party unity's import has already been noted. Government management of the economy, foreign policy, civil liberties and operations issues were essentially driven by personal factors in both years. Additionally, the civil liberties and operations arenas shared a marked increase in the significance of partisan factors in 1982. These tables contain counterintuitive findings in regard to voting support patterns. The findings regarding party, party leader, and committee chair or ranking member support are not what one would expect. More often than not, Republicans supported Baker's positions less than did Democrats on average. This finding is probably more a result of the strength of Republican Party support than it is actual voting differences between partisans and leaders. Being one of five key party leaders had almost no effect on voting. This indicates that the Republican leadership and the Democratic leadership were no more supportive of Baker's positions than the average senator. The same finding holds for committee chairs and ranking members as well. In 1981 only Democratic committee chairs and ranking members showed significant levels of support, and they voted with Baker more often than did the average senator. Repub lican chairs provided notable support in 1982's government operations arena. On most issues, party rank and file members were statistically indistinguishable from either committee chairs or party leaders. This is not consistent with literature citing distinct career, policy goal, and voting differences between rank and file members and the leadership (e.g., Grumm, 1963; Andrain, 1964; Hibb ing 1991). Bobic, SPSA 2004 Revea ling Lea dership Page 20 The divergence in the rankings, the intermodel correlations, and the models of support indicate that while Baker had stronger support, it was still fragmented support requiring new efforts as bills or votes from different policy arenas came to the floor. Despite these limitations, Baker successfully built coalitions of members around his positions. Was he able to mobilize this support when he needed it on the senate floor? Was Baker as successful creating partisan cohesion as he was in building coalitions? That is the question addressed in the next section. Building Party Cohesion in the Sena te in the 97th Congress. Describing Cohesion Structure. The PHI scores reported in tables 4 and 5 would lead one to expect that the task of coordinating individual member support into blocs of votes on specific measures would be a relatively easy task for Baker and Reagan in 1981 and 1982. Moreover, since the arenas in which Baker and Reagan exercised their most aggressive efforts to build those blocs included government management of the economy, foreign policy, social welfare, and government operations issues, one would expect to see particularly strong Rice scores for those policy areas. However, the anecdotal evidence indicates that in 1982, Republicans departed somewhat from the policy agenda Reagan had outlined to pursue their own policies. Moreover, several Republican senators began to push for floor debate on such social issues as abortion, bussing and school prayer. Thus, one would expect to see that Republican cohesion declined in government management of the economy and civil liberties issues in 1982 compared to 1981. These expectations are confirmed by the patterns in Table 10. This table presents mean Rice scores (Rice, 1923)by policy area, year and presidential agenda status. These tables present several interesting patterns of support. Table 10 Here In 1981, Republicans were remarkably cohesive. The Rice scores indicate that Baker was able to marshall on average seventy-five percent of Republicans to vote with him on any given issue. The highest levels of cohesion exist for government management, foreign policy and social welfare issues. Cohesion Bobic, SPSA 2004 Revea ling Lea dership Page 21 increased on those votes embodying the heart of the Reagan agenda. With the exception of civil liberties, agenda cohesion increased dramatically over nonagenda cohesion. CQ vote #132, HR 3512, a civil liberties issue on Reagan's agenda, produced a cohesive bloc of Republicans voting against Baker's position (Rice Index score: -24.53). The data for 1982 differ from 1981, but not in a consistent manner. Republican cohesion about Baker's preferred position still remained strongest in social welfare, government management and foreign policy issues, but the specific score values differ from the Rice scores in 1981. Cohesion rates for civil liberties issues declined, supporting anecdotal evidence of a split among Repub licans. Agenda items did not seem to have the impact on cohesion in 1982 that they did in 1981. The most likely explanation for this pattern is that Reagan was less personally involved in the legislative process in 1982 (Ripley 1988). Modeling Republican Cohesion Structure. The data presented in Table 10 indicate that the Repub lican party voted as a bloc to a remarkable degree. These data support the anecdotal evidence presented at the beginning of this chapter that Baker forged an unprecedented cohesiveness in the party. According to Ornstein, Peabody and Rhode (1989), Baker appealed to party loyalty and used the floor schedule to create this level of party cohesion. However, as inflation and unemployment continued to rise, Republicans facing reelection pressured the administration to soften its tax and budget program. This internal partisan conflict over the shape of the 1982 and 1983 budgets should have decreased cohesion rates. Table 10 indicates that this is not the case. Cohesion within a party responds to changes in environmental, bill related, and leadership factors. These factors should manifest severa l patterns, given the historical evidence presented. Specifically, economic variables should decrease cohesion rates, while new spending or bills which increase spending by ten percent or more should increase cohesion. Agriculture, civil liberties, and social welfare variables should be significantly different from government operations. The Baker-related variable of year should decrease cohesion, while television appearances and agenda status should increase cohesion. To test Bobic, SPSA 2004 Revea ling Lea dership Page 22 Baker’s ability to draw members toward his position, I created sets of variables representing Baker-specific issues, environmental issues, and bill-specific pressures. These are presented in List II in the Appendix. Cohesion explores outcomes across bills. Senators are no longer the unit of analysis, specific bills are. The Dependent variable for this analysis is a ratio of those Senators supporting Baker divided by the total number of Senators voting on a given bill. This is sometimes called ratio logistic regression. The results are similar to standard 0,1 regression (Agresti, 1990). The advantage is that the model predicts an increase in the number of Senators supporting Baker, ie: an increase in Cohesion. The resulting regression line indicates the change in the log of the odds of Senators supporting Baker and opposing his position. Cohesion is a party-specific behavior: Baker’s ability to increase cohesion among Republicans implies a loss of ability to attract Democrats to his position. Thus, we explore Baker’s effects on Republican and Democratic cohesion. The results of this analysis for Republicans appear in Table 11. The columns of the table present the logistic regression coefficients (column 3), their standard errors (column 4), the test statistics and probabilities (columns 5 and 6), and finally, the odds ratio of observing high rates of cohesion for each successive unit increase in a predictor variable. As with the data for the 95th Congress, the -2 loglikelihood chi-squared value and the AIC value indicate the model does capture Republican cohesion. The difference in the null and model AIC values (538.52) show that the model significantly increases the ability to predict increases in cohesion patterns. The Gamma and Somer's D measures indicate a moderate association between the model prediction and the actual cohesion rates. The findings in Table 11 confirm the anecdotal assessment of Republican cohesion in the 97th Congress. It is clear from these data that the worsening economic environment harmed Republican cohesion. As inflation and the gross domestic product increased, Republican cohesion in the Senate was less likely to occur. On the other hand, Republicans were more likely to vote as a bloc as unemployment Bobic, SPSA 2004 Revea ling Lea dership Page 23 rates soared. This finding is consistent with literature on the effects economic indicators have on presidential success. Bill specific results show that Baker was more likely to gain votes in social welfare, government management of the economy and foreign policy than in government operations issues, and less likely otherwise. The odds of maintaining Republican cohesion were 1.46 times greater for bills which decreased spending by more than ten percent than for those which did not affect spending. The odds were also 1.2 times greater for continuing programs than for new proposals (Newbill). Table 11 Here Public attention to a policy, as measured by the variable "salience", tended to decrease cohesion, indicating that members were responding to public pressures to alter the Reagan agenda. Overall, bill specific issues exerted substantial influence in Republican voting behavior. Most leader specific variables significantly affected cohesion rates. The odds of maintaining Republican cohesion were 1.37 times greater in 1981 than in 1982. Baker's appearances on television to defend the Reagan agenda had little impact on Republican cohesion. Republicans were 1.32 times more likely to vote for programs on the Reagan agenda than they were to vote for nonagenda items. Overall, these findings support the anecdotal evidence and expectations. Republican cohesion declined significantly in 1982 from the remarkably high levels found in 1981. That decline seems to be a function of the recession, members putting civil liberties issues on the agenda, and the change in policy emphasis from 1981 to 1982. Moreover, the coefficient for year would indicate that Baker had a honeymoon of sorts as majority leader much like the one he had as minority leader. This is an interesting finding, given the importance the honeymoon effect plays in understanding presidential success in getting legislation passed. These data indicate that Baker was successful in turning Senate Republicans into a powerful voting bloc in support of the Reagan agenda. However, as majority leader Baker was also responsible for persuading Senate Democrats to support Reagan's proposals. Thus, unlike the analysis of the 95th Bobic, SPSA 2004 Revea ling Lea dership Page 24 Congress, an understanding of Baker's leadership in the majority requires an examination of his ability to affect Democratic votes. The data indicate that, while a majority of Democrats did not support the Reagan agenda, Baker nevertheless had an impact on Democratic voting patterns. Modeling Democratic Cohesion Structure. According to Senate observers, Democrats supported the Reagan agenda in 1981 at very high rates. While this support was probably due to D emocratic uncertainty about the nature of Reagan's election, observers say their support helped insure that Reagan's program emerged from the Senate relatively intact (Ornstein, Peabody, and Rohde, 1989; Ripley, 1988). Thus, Democratic cohesion Rice scores should be positive or close to zero in 1981. Recall that the modification of the Rice Index proposed in Chapter Three uses negative values to indicate cohesion in opposition to a given position. Table 12 presents the average Democratic cohesion rates by policy area and presidential interest for 1981 and 1982. The data do not support expectations. First, it is evident from the table that Democrats were not particula rly supportive of Baker's position. In 1981, the only positive Rice Valu e is for nonagenda agricultural policies. Moreover, Rice scores for presidential agenda items were not consistently greater or lesser than those for nonagenda items. If the hypothesis of Democratic acquiescence to the Reagan mandate were true, one would expect to see smaller negative values for agenda items than for nonagenda items. This expectation holds for government management of the economy issues, but not for social welfare or foreign policy issues. Table 12 Here The data for 1982 are as one would expect, given the policy interests of the Democratic Party and its leadership's criticisms of the Reagan program. While Reagan argued that his reforms were designed to stimulate the economy, Democrats reacted to the Reagan program as if it were a restructuring of social welfare policies. Democrats made many efforts late in 1981 and throughout 1982 to restore funds to such social welfare policy programs as unemployment insurance, and supported a new jobs bill and a Bobic, SPSA 2004 Revea ling Lea dership Page 25 program to encourage new housing construction. Thus, the increase in cohesion in 1982 in social welfare programs is to be expected. The agenda data are the most interesting. It would appear that in 1982, Democrats were less unified in their opposition to Reagan initiatives than they had been in 1981. For the most part, the Rice scores on agenda items in 1982 are less than the scores for 1981. While this could indicate increased support for the Reagan agenda in 1982, it is more likely a reflection that while Democrats opposed much of the Reagan agenda, the party leadership was not very effective in presenting coherent alternatives to the Republican initiatives. Reagan may have been less personally involved in the legislative process in 1982, but the policy agenda still bore his marks and narrowed the scope of the policy debate. Democrats took the downturn in the economy as a signal to change the Reagan program in some fundamental ways. It also allowed Democrats to argue that the cuts made in various social welfare programs had been too deep and that funds to these programs should be restored. Thus, models of Democratic cohesion should be driven in part by economic variables. Democratic cohesion should also respond differently in different policy arenas as well. First of all, since ethics and campaign finance reforms were occurring at this time, one would expect to find government operations issues appearing as a significant model term. Government management of the economy effects should be about as strong as government operations (demonstrated by a nonsignificant coefficient), while the impact of foreign policy and civil liberties issues should be weaker than that of government operations (signified by significant but negatively signed coefficients). Since the empirical question is Baker's ability to draw Democratic supporters into his camp, this impact would be represented in the logistic models by positive coefficients for the other policy related variables. On the other hand, Senate observers note that Baker made special efforts to tailor his leadership in a way that would appeal to Democrats as much as possible. Therefore, one would expect to see some of the Baker related variables show strong positive effects on the odds of Democrats supporting Baker's positions. Year should not be one of those variables. The honeymoon effect should be more pronounced Bobic, SPSA 2004 Revea ling Lea dership Page 26 for Democrats than for Republicans, evidenced in the odds ratio value being much larger for Democrats than the 1.32 value for Republicans. The results of the Democratic cohesion logistic model are reported in Table 13. The model statistics again indicate that the model predicts Democratic cohesion well, but with less power than the Republican model. The difference in AICs is smaller, as is the Chi-Squared value. Baker had a smaller impact on Democratic cohesion than he did on Republican cohesion. This conclusion is supported by the measures of association as well. Both the Gamma and Somer's D values are smaller than they were for the Republican model. They indicate a weak association between the model prediction and actual Democratic cohesion. Table 13 Here The first point to note in these results is the impact of economic variables. As the economy soured, Democrats were far less likely to support Baker's positions and by extension, the president's initiatives. Only as inflation increased (as measured by the Consumer Price Index) did Democratic support increase. Bill specific variables perform largely as expected. Baker was more likely to gain Democratic support for agricultural arena issues than he was for government operations issues. He was less likely to have Democratic support for government management of the economy issues than for government operations. None of the other policy variables were significant. Newness of proposals increased the odds of Democratic support. Baker specific issues show very interesting patterns. First of all, the variable for year is nonsignificant. Baker may have had a honeymoon with Republicans, but it did not carry across the aisle. Baker's television appearances did increase the odds of Democratic support. One interpretive note is in order. The intercept value for Table 13 implies that all things being equal, Democrats were more likely to support Baker's positions than not. However, this result is somewhat artifactual, given the performance of economic variables. For example, by inserting the mean Bobic, SPSA 2004 Revea ling Lea dership Page 27 values for economic variables in 1981 and multiplying those values by the model terms reported above, one finds that only a five percent probability of support for Baker's positions existed. The computation of the probabilities is easy. It is found by the following equation: Where B= the coefficients in Table 13, and X= the particular variable values for 1981. The evidence presented above supports the anecdotal evidence. Baker was able to build coalitions of Republicans and then turn that support into cohesive voting blocs when it mattered. Despite the downturn in the economy late in 19 81, Republican cohesion remained remarkably high. Democratic cohesion was not nearly as solid as Republican cohesion, possibly reflecting disarray among Democratic leaders to present a unified alternative to the Reagan agenda. Was Baker able to use this support in his party to deliver the votes necessary to win passage of the Reagan agenda? Clearly the answer to this question is yes. The next section explores some of the reasons behind that remarkable success. Winning Votes in the 97th Congress. Describing Winning. Table 14 presents Baker's win and loss record by policy type and presidential interest for 1981 and 1982. The data explore Baker’s success in passing bills on the President’s Agenda and off the agenda. It is clear from the table that Baker's position prevailed frequently. The table also demonstrates how the flow of policies shifted between 1981 and 1982. In 1981, the most common type of policy before the Senate was in the government management of the economy arena. Second most frequent were social welfare policies. The president's agenda was composed primarily of social welfare and government management of the economy policies. In both cases, Baker's ability to win votes was remarkable. Baker's position prevailed almost eighty percent of the time on nonagenda issues, and almost eighty-five percent of the time on agenda items. Bobic, SPSA 2004 Revea ling Lea dership Page 28 Table 14 Here Nineteen eighty-two displays similar patterns. Once again, the distribution of policies shows a preponderance of social welfare and government management of the economy issues. However, there is a decline in success in social welfare issues in 1982. While Baker's position prevailed on non-agenda items eighty-five percent of the time in 1981, it prevailed only sixty-five percent of the time in 1982. He faced a similar, but less dramatic decline in success on government management of the economy measures as well. Presidential interest did not seem to help much. Social welfare items on the President's agenda were almost four times more likely to lose in 1982 than they were in 1981. Reagan's position still prevailed in seven of ten cases in 1982, but that is a far cry from 1981. While these patterns are interesting, they do not explain what factors led to Baker’s success. Modeling success uses the same variables as for cohesion. The difference is that several interactive terms and model specifications will be tested. The variables used are in List III in the Appendix. Modeling Winning. These data suggest that interactions between several model terms exist. For example, the data strongly indicate that there is probably an interaction between year and presidential agenda. There also seems to be an interaction between year and policy type. The interaction between year and policy, recall, captures the dynamic nature of the policy process. Policy arenas which dominate politics in one year often fade and are replaced by a different arena in subsequent years. This is especially true of presidential politics. If the president does not press for his agenda quickly, the policy process simply marches on without him. President Reagan was aware of this fact of political life and acted very quickly in 1981 to bring his proposals to Congress. Even when control of the agenda slipped from his hands somewhat in 1982, he had substantially limited the agenda to budgetary and deficit issues, so that unlike the issue volatility present in the 95th Congress, both years of the 97th Congress were dominated by social welfare and government management of the economy issues (Rudder, 1989). The number of bills declined in 1982, Bobic, SPSA 2004 Revea ling Lea dership Page 29 as well as the proportions of bills in these two principle arenas. Thus, one would anticipate that modeling Baker's ability to win votes would require a model with at least a few interaction terms. Thus, five models of winning were used to understand these patterns. These models are discussed in Table 15. Essentially, they use the model terms, plus several interactions. Five models are fitted to the data. The Dependent variable is a 0 or a 1, indicating Baker’s position prevailed (1) or failed (0). Standard logistic regression is used to explore Baker’s success. The five models of success were fitted to the data and the Consistent Akaike's Information Criterion (CAIC) is used to determine the best-fitting model. One would anticipate that at least the agenda by year interaction should contribute to an understanding of Baker's ability to win votes on the floor of the Senate. Table 15 presents the model selection statistics derived from fitting the different interaction models to the data. The first column of the table identifies the model under examination. The second and third columns present the -2 loglikelihood P2 values and associated significance values for each model. The final column presents the CAIC value. As always, the minimum CAIC value indicates the model most likely to have generated the actual data. The best fitting model is denoted by two stars beside the CAIC value. Table 15 Here The CAIC value minimizes at the no interactions model as the best fitting model for these data. While this finding is somewhat surprising given the apparent strength of the year by agenda interaction, it is less surprising that the year by policy interaction was eliminated. Apparently, the year by president interaction simply do not contribute enough to reductions in model error to justify the additional information costs incurred in including them in the model terms. Fitting the no interactions model to the data produces a number of interesting findings. First of all, the model correctly predicts Baker's successes 341 times out of 351, but correctly classified a vote as a loss only 19 times out of 73 (Table 16). It appears to be easier to model floor success than it is to model floor defeats. Baker's tremendous success rate presents some difficulties in computing reliable model fit Bobic, SPSA 2004 Revea ling Lea dership Page 30 statistics. However, studies of the CAIC indicate that it is robust even in the presence of this kind of data (Bozdogan, 1994). Table 16 Here Fitting the no interaction model chosen in Table 15 produced the results in Table 17. The goodness of fit Chi-Square value indicate that this model overfits the data, but as in the model for the 95th Congress, variable selection routines did not change the overall patten of results. The Chi-Squared value indicates that this model captures the dynamics of winning. The Gamma and Somer's D values show a strong association, although they are less reliable due to the large number of cases in the joint win-win cell (Asher, Weisberg, Kessel and Shively, 1984; p. 216). Of the three dimensions affecting the likelihood of a Baker win, environmental variables seem to exert the strongest influence. Changes in the gross domestic product decreased the odds of winning, while freshman support and increases in inflation increased the odds of success. Table 17 Unemployment rates had no impa ct on the odds of Baker's position prevailing on the floor. While these data cannot address the issue, it appears that Baker's efforts late in 1981 to modify Reagan's second round of tax cuts, and Republican efforts to reshape the 1983 budget in 1982 prevented the party and the president from suffering a number of embarrassing defeats on the floor. Certainly political commentators like Tolchin (1982) and Arieff (September 12, 1981) believed that was the case. Bill specific variables were a bit disappointing. The policy results indicate that the odds of success were no different for any policy as compared to government operations voting patterns. The only other bill specific variable which affected the odds of floor success was the size of the bill. The odds of success for Baker were eight times greater for bills which decreased spending by ten percent or more than for bills which left spending limits about what they had been. Finally, those variables associated with Baker were also somewhat disappointing. None of them achieved significance at the "=.05 level, although Baker's television appearances and presidential interest Bobic, SPSA 2004 Revea ling Lea dership Page 31 were close. The honeymoon effect so prominent in Republican cohesion patterns does not appear to translate into a honeymoon effect on the floor. Perhaps the reason no honeymoon effect appears is due to the difference in the tenure of a Senate leader and a president. While a president is limited to two terms in office, senators know their leaders are not limited as to terms of service. Thus, it is important that members work with their leaders not only for their own career opportunities in the future, but also for the sake of the party. The lack of a significant interaction term challenges the assumption that because Reagan was less involved in policy in 1982, he was less successful than in 1981. Assuming that Baker continued to pursue Reagan's agenda in 1982 to the same degree that he had in 1981, these data indicate that Reagan's interest in a policy only marginally affected the odds of winning. Taken as a whole, these data suggest that Baker's role in keeping the Reagan agenda alive and on track has not been overstated. These data indicate that Baker's efforts on behalf of the administration paid off in several unexpected ways. There are indications that Baker's ability to win votes was not bound to one or another specific policy dimension. He seemed equally likely to win across the whole policy domain in both years of the 97th Congress. While these data do not say so, it would not be unreasonable to argue that the evidence presented above would indicate that a great deal of the Reagan Revolution moved forward on the slightly stooped shoulders of the majority leader from Tennessee. Conclusion. The analysis presented in this chapter has explored Howard Baker's ability to build coalitions of senators, to organize those coalitions into cohesive voting blocs, and to win floor votes in the 97th Congress. The analysis confirmed some anecdotal assumptions about Baker's leadership in the Senate and challenged others. Overall, the analysis paints a picture of a very successful leader whose skills and abilities overcame a number of critical obstacles to bring the Reagan agenda to fruition. The analysis of coalitions indicated that contrary to expectations, Republicans and Democrats remained remarkably unified and at opposite ends of the policy spectrum. Baker's core supporters were Bobic, SPSA 2004 Revea ling Lea dership Page 32 composed of conservative Republicans, with most of the new freshman senators supporting the president's initiatives and thus supporting Baker. However, the data did not support the argument that Democrats were overly supportive of Reagan's policies even in 1981. While Democrats may have been with the president on final passage of key bills, they did not allow the President a free hand in redesigning government. At every step of the way, they offered modifying amendments, used delaying tactics, and whatever parliamentary maneuvers they could to protect the interests of their constituents. If they did support the president on final passage of key agenda items, it was because they had done all they could beforehand to modify his proposals. Additionally, despite Republican fragmentation over civil liberties issues late in 1981 and late in 1982, Baker's coalitions of supporters remained relatively stable and offered him very high levels of support. Baker could count on the average Republican's vote on seventy percent of all roll calls taken. The strong coalitions Baker built translated into phenomenally high levels of cohesion in both years of the 97th Congress. If Baker could count on a senator's vote seventy percent of the time, he could also count on seventy-five percent of fellow Republicans on the typical vote. That support increased in the policy arenas of special interest such as government management of the economy, social welfare, foreign policy or government operations. All of this translated into a record for winning that was unprecedented. Baker's position prevailed eighty percent of the time in floor voting. The model of success suggests that Baker's ability to modify proposals, to work with Democrats and Republicans, and to encourage party loyalty even among freshmen "amateurs" brought much of the Reagan program through the Senate relatively intact. This analysis supports the general opinion of Baker as a majority leader. Baker's ability to build coalitions, create cohesive blocs of voters and win votes on the floor of the Senate were remarkable accomplishments. The picture emerging from this analysis is that of a leader in control of his party and of the Senate floor, who was able to coordinate support from diverse members, and to gain passage of important legislation in critical times. Bobic, SPSA 2004 Revea ling Lea dership Page 33 Baker's leadership was not perfect. He himself acknowledged that his inability to control members' desires for debate on civil liberties issues represented a serious failure of leadership. The collapse of Senate decorum in the lame duck session of 1982 was another example of events overwhelming the majority leader and leaving him with little control over Senate proceedings. Nevertheless, in the aggregate, the data confirm the general impression that Baker was a very successful leader of the United States Senate. Was he as great a leader as Johnson or Mansfield? Without a similar analysis of these men, that question cannot be answered empirically. However, the record Baker posted for accomplishing the primary tasks of leadership suggests that such comparisons are not unwarranted. Bobic, SPSA 2004 Revea ling Lea dership Page 34 Table 1: Information Theoretic Cluster Solutions by Policy, 1981 and 1982 1981 Policy Maximum Clusters 1982 Best Solution CAIC Best Solution CAIC AG 7 2 118.3 2 146.6 SOCWEL 7 4 134.4 3 124.7 GOVMAN 7 2 102.8 4 136.2 FORPOL 7 5 120.9 2 125.5 CIVLIB 7 6 56.25 2 96.1 GOVOPS 7 2 154.2 2 169.3 Bobic, SPSA 2004 Revea ling Lea dership Page 35 Table 2: Mean Characteristics of Senators, by Policy Arena, PID, Ideology and Freshman Status, 1981 Policy/ Charac teristic Close to Baker Moderate Far From Baker AG PID R, 48; D, 15 Only Two Cluste rs R, 7; D, 31 AG ADA 38.7 (24.11) 58.9 (20.28) AG Fresh F, 14; Not, 49 F, 5; Not, 33 SOCW EL PID R, 42; D, 0 R, 7; D, 48 R, 1; D, 24 SOCWEL ADA 24.0 (19.16) 48.1 (16.23) 71.7 (14.4) SOCW EL Fresh F, 14; Not, 28 F, 3; Not, 31 F, 2; Not, 23 GOVM AN PID R, 54; D, 1 Only Two Cluste rs R, 1; D,45 GOVMAN ADA 30.00 (19.16) 60.86 (19.78) GOVM AN Fresh F, 16; Not, 39 F, 3; Not, 43 FORPO L PID R, 49; D, 1 R, 5; D, 20 R, 1; D, 25 FORPOL ADA 27.0 (17) 48.0 (17.34) 71.13 (14.69) FORPO L Fresh F, 16; Not, 34 F, 1; Not, 24 F, 2; Not, 24 CIVLIB PID R, 35; D, 28 R, 2; D, 6 R, 14; D, 26 CIVLIB ADA 28.34 (16.23) 50.01 (10.45) 63.45 (22.17) CIVLIB Fresh F, 12; Not, 41 F, 3; Not, 5 F, 13; Not, 36 GOVO PS PID R, 50; D, 1 Only two C lusters R, 4; D, 42 GOVOPS ADA 29.62 (18.09) 60.47 (24.62) GOVO PS Fresh R, 16; Not, 38 F, 3; Not, 42 R=Republicans, D=Democrats Numbers in Parentheses are Standard Deviations. F=Freshmen (members with six years of tenure or fewer) Not=Senators with more than six years of tenure. Bobic, SPSA 2004 Revea ling Lea dership Page 36 Table 3: Mean Characteristics of Senators, by Policy Arena, PID, Ideology and Freshman Status, 1982 Policy/ Charac teristic Close to Baker Moderate Far From Baker AG PID R, 42; D, 4 Only Two Cluste rs R, 13; D, 42 AG ADA 21.89 (24.41 53.91 (31.24) AG Fresh F, 12; Not, 34 F, 7; Not, 48 SOCW EL PID R, 46; D, 0 R, 8; D, 8 R, 1; D, 38 SOCWEL ADA 11.55 (11.12) 34.06 (12.14) 77.94 (17.24) SOCW EL Fresh F, 16; Not, 30 F, 1; Not, 15 F, 2; Not, 32 GOVM AN PID R, 41; D, 0 R, 12; D, 13 R, 0; D, 34 GOVMAN ADA 12.5 (12.29) 30.0 (16.3) 77.94 (19.24) GOVM AN Fresh F, 14; Not, 27 F, 3; Not, 23 F, 2; Not, 32 FORPO L PID R, 52; D, 5 Only Two Cluste rs R, 3; D, 41 FORPOL ADA 15.81 (14.26) 69.66 (22.43) FORPO L Fresh F, 16; Not, 41 F, 3; Not, 41 CIVLIB PID R, 37; D, 9 R, 15; D, 10 R, 8; D, 22 CIVLIB ADA 16.31 (16.38) 41.66 (27.69) 73.33 (22.56) CIVLIB Fresh R, 13; Not, 33 F, 3; Not, 22 F, 3; Not, 27 GOVO PS PID R, 52; D, 0 Only Two Cluste rs R, 3; D, 46 GOVOPS ADA 14.71 (14.02) 65.31 (25.19) GOVO PS Fresh F, 17; Not, 35 F, 2; Not, 47 R=Republicans, D=Democrats Numbers in Parentheses are Standard Deviations. F=Freshmen (members with six years of tenure or fewer) Not=Senators with more than six years of tenure. Bobic, SPSA 2004 Revea ling Lea dership Page 37 Table 4: Top 20 Correlates by Policy Arena, 1981 Rank AG SOCWEL GOVMAN FORPOL CIVLIB GOVOPS Humphrey (D,NH) 20 Armstrong ( R ,C O ) H e lm s (R,NC) Simpson ( R ,W Y) Tower ( R ,T X) Eas t (R,NC) 19 Goldwater (R,AZ) Danfor th ( R ,M O ) Warner ( R ,VA) Nickles ( R ,O K) Jepsen ( R ,IA) H atc h ( R ,U T ) 18 Schm itt (R,NM) Tower ( R,TX) Schm itt (R,NM) Wallop ( R ,W Y) Mattingly ( R ,GA) Chafee (R, RI) 17 Wallop ( R,WY) Mattingly ( R,GA) Per c y (R,IL) Jepson ( R ,IA) H atc h ( R ,U T ) Quayle ( R , IN ) 16 Cochran ( R,M S) Kasten (R, W I) Cochran ( R ,M S) Rudman (R,NH) Kasten (R, W I) Goldwater (R,AZ) 15 Domenici (R,NM) Grassley ( R,IA) Lugar ( R , IN ) Schm itt (R,NM) Lugar ( R , IN ) Stevens ( R ,AK) 14 Byrd, H ( I,VA) H a y ak a w a ( R,CA) Stevens ( R ,AK) Eas t (R,NC) Hollings (D,SC) Thurmond (R,SC) 13 Weicker ( R ,C T ) Warner ( R,VA) Denton ( D ,GA) Cochran ( R ,M S) Long ( D ,LA) D'Am ato ( R ,N Y) 12 Tower ( R,TX) Laxalt ( R,NV) Thurmond (R,SC) Denton ( D ,GA) Stennis ( D ,M S) Cochran ( R ,M S) 11 Dole ( R,KS) McClure ( R , ID ) H a y ak a w a ( R ,C A) Gorton ( R ,W A) Byrd, H ( I,VA) Denton ( D ,G A) 10 Mattingly ( R,GA) Stevens ( R,AK) Tower ( R ,T X) Simpson ( R ,W Y) Nunn ( D ,G A) Packwood (R,OR) 9 Thurmond (R,SC) Sym ms ( R , ID ) Goldwater (R,AZ) Mattingly ( R ,G A) Warner ( R ,VA) Sym ms ( R , ID ) 8 H a y ak a w a ( R,CA) Dole ( R,KS) Murkows ki ( R ,AK) Stevens ( R ,AK) Kassebaum ( R ,KS) Murkows ki ( R ,AK) 7 Lugar ( R , IN ) Denton (D,AL) Laxalt ( R ,N V) Lugar ( R , IN ) Wallop ( R ,W Y) Wallop ( R ,W Y) 6 Warner ( R,VA) Garn ( R ,U T ) Eas t (R,NC) D'Am ato ( R ,N Y) H a y ak a w a ( R ,C A) Tower ( R ,T X) 5 H e lm s (R,NC) Gorton ( R,WA) Quayle ( R , IN ) Laxalt ( R ,N V) Stevens ( R ,AK) McClure ( R , ID ) 4 Denton (D,AL) Wallop ( R,WY) McClure ( R , ID ) H a y ak a w a ( R ,C A) Cochran ( R ,M S) Laxalt ( R ,N V) 3 East (R,NC) Abdnor (R,SD) D om eni c i (R,NM) Garn ( R ,U T ) Schm itt (R,NM) H a y ak a w a ( R ,C A) 2 Laxalt ( R,NV) East (R,NC) Wallop ( R ,W Y) D om eni c i (R,NM) Tower ( R ,T X) Garn ( R ,U T ) 1 Murkows ki ( R,AK) Schm itt (R,NM) Dole ( R ,KS) Thurmond (R,SC) Simpson ( R ,W Y) Dole ( R ,KS) Mean Phi .67 .82 .85 .84 .88 .81 Bobic, SPSA 2004 Revea ling Lea dership Page 38 Table 5: Top 20 Correlates by Policy Arena, 1982 Rank AG SOCWEL GOVMAN FORPOL CIVLIB GOVOPS Jepsen ( R ,IA) 20 Schm itt (R,NM) McClure ( R , ID ) Mattingly ( R ,GA) Rudman (R,NH) Laxalt ( R ,N V) 19 Domenici (R,NM) Mattingly ( R,GA) Denton (D,AL) Laxalt ( R ,N V) Eas t (R,NC) McClure ( R , ID ) 18 Lugar ( R , IN ) Hatch ( R ,U T ) Sym ms ( R , ID ) H a y ak a w a ( R ,C A) Heflin (D,AL) Gorton ( R ,W A) 17 Cohen ( D,M E) Simpson ( R,WY) Rudman (R,NH) Denton (D,AL) Proxmire (R, W I) Armstrong ( R ,C O ) 16 Brady ( D,NJ) Warner ( R,VA) D om eni c i (R,NM) Hawkins (D,FL) Schm itt (R,NM) Quayle ( R , IN ) 15 Weicker ( D ,C T ) Stevens ( R,AK) Per c y (R,IL) D om eni c i (R,NM) Johnston ( R ,LA) Mattingly ( R ,G A) 14 Murkows ki ( R,AK) Abdnor (R,SD) McClure ( R , ID ) Eas t (R,NC) Dole ( R ,KS) Tower ( R ,T X) 13 Sym ms ( R , ID ) H e lm s (R,NC) Warner ( R ,VA) Jepsen ( R ,IA) Grassley ( R ,IA) Laxalt ( R ,N V) 12 McClure ( R , ID ) Sym ms ( R , ID ) Tower ( R ,T X) Stevens ( R ,AK) Kassebaum ( R ,KS) Kassebaum ( R ,KS) 11 Hawkins (R,FL) Gorton ( R,W A) Garn ( R ,U T ) Mattingly ( R ,GA) Humphrey (D,NH) Goldwater (R,AZ) 10 Rudman (R,NH) East (R,NC) H a y ak a w a ( R ,C A) Cochran ( R ,M S) D'Am ato ( R ,N Y) Brady ( R ,N J ) 9 Armstrong ( R ,C O ) Wallop ( R,WY) Thurmond (R,SC) Goldwater (R,AZ) Quayle ( R , IN ) Sym ms ( R , ID ) 8 Goldwater (R,AZ) Tower ( R,TX) Quayle ( R , IN ) Schm itt (R,NM) Abdnor (R,SD) D om eni c i R,(NM) 7 Dole ( R,KS) H a y ak a w a ( R,CA) Simpson ( R ,W Y) H atc h ( R ,U T ) Mattingly ( R ,GA) Thurmond (R,SC) 6 Mattingly ( R,GA) Denton ( D,GA) Cochran ( R ,M S) Kasten (R, W I) Kasten (R, W I) Rudman (R,NH) 5 H a y ak a w a ( R,CA) Dom enici (R,NM) Laxalt ( R ,N V) Tower ( R ,T X) Denton (D,AL) Lugar ( R , IN ) 4 H e lm s (R,NC) Thurmond (R,SC) Murkows ki ( R ,AK) Lugar ( R , IN ) Sasser (D,TN) Stevens ( R ,AK) 3 Denton ( D,GA) Laxalt ( R,NV) Dole ( R ,KS) Simpson ( R ,W Y) Lugar ( R , IN ) Murkows ki ( R ,AK) 2 Laxalt ( R,NV) Brady ( R,NJ) Brady ( R ,N J ) Garn ( R ,U T ) Murkows ki ( R ,AK) H a y ak a w a ( R ,C A) 1 Byrd, H ( I,VA) Dole ( R,KS) Stevens ( R ,AK) Brady ( D ,N J ) Cochran ( R ,M S) Garn ( R ,U T ) Mean Phi .69 .77 .84 .71 .52 .83 Bobic, SPSA 2004 Revea ling Lea dership Page 39 Table 6: Intermodel Correlation Matrix, 1981 AG S O CW E L GOVMA N FORPOL CIVLIB GOVOPS AG 1.0 -.225 .188 .136 .345 .299 SOCWEL -.225 1.0 .221 .005 -.181 .193 GOVMAN .188 .221 1.0 .014 -.216 .386 FORPOL .136 .005 .014 1.0 -201 .009 CIVLIB .345 -.181 -.216 -.201 1.0 -.043 GOVOPS .299 .193 .386 .009 -.043 1.0 N=100 Senators in 1981. Bobic, SPSA 2004 Revea ling Lea dership Page 40 Table 7: SUR Solutions for Baker's Support by Policy Arena: 19813 Dim P A R T I S A N Vars Intercept Party Rep Lead Dem Lead RepCom DemCom P E R S O N A L ACA ADA PRESSUP CONSCO AL Tenure Govemp AFLCIO CCUS Partyun E N V I R O N M E N T Pop Med Inc Black Exp Rev Medcaid South Individual F-Tests Adj. R-Square (OLS) -.068 (.025) -.003 (.001) .411 (.379) .006 (.002) .007 (.002) .010 (.001) .002 (.002) .010 (.002) .014 (.002) -1.56 (.978) -.004 (.001) .016 (.002) .001 (.000) .004 (.003) -.024 (.007) .031 (.004) .174 (.078) AG (se) .486 (.249) -.267 (.366) SOCWEL (se) -.370 (.208) -.612 (.197) GOVMAN (se) .086 (.100) -.294 (.154) FORPOL (se) -.342 (.201) -.518 (.191) CIVLIB (se) -.019 (.329) -.152 (.195) GOVOPS (se) .444 (.138) -1.02 (.271) .001 (.001) .006 (.002) 1.108 (.545) -.003 (.002) .007 (.003) 19.934,70 .505 214.796,68 .945 720.174,70 .975 287.016,68 .958 44.735,69 .747 180.863,71 .879 3 As with the tables in Chapter 4, the Standard Errors are truncated. Thus boxed values may not ap pear to be significant. Bobic, SPSA 2004 Revea ling Lea dership Page 41 Table 8: Intermodel Correlation Matrix, 1982 AG AG SOCWEL GOVMAN FORPOL CIVLIB GOVOPS 1.0 .214 .145 -.128 -.091 .020 SOCWEL .214 1.0 .179 -.255 .197 -.188 GOVMAN .145 .179 1.0 -.299 .034 -.033 FORPOL -.128 -.255 -.299 1.0 .099 -.033 CIVLIB -.091 .197 .034 .099 1.0 -.061 GOVOPS .020 -.188 -.033 -.033 -.061 1.0 N=100 Senators Bobic, SPSA 2004 Revea ling Lea dership Page 42 Table 9: SUR Solutions for Baker's Support by Policy Arena: 19824 Dim Variable AG(se) SOCW EL(se ) .276 (.117) -.495 (.185) GOVMAN (se) -.321 (.179) -.307 (.169) .008 (.048) FORPOL (se) -.458 (.194) -.221 (.193) CIVLIB(se) GOVO PS(se) P A R T I S A N Intercept Party Rep Lead Dem Lead RepCom DemCom -.723 (.100) .647 (.124) .285 (.172) -.217 (.294) -.845 (.351) .130 (.107) .466(.252) -.324 (.321) .221(.097) -.253 (.066) .171 (.109) .005 (.002) P E R S O N A L ACA ADA P R E S SU P CONSCOA L Tenu re Govemp AFLCIO CCUS Partyun .013 (.002) -.003 (.001) .007 (.002) .011 (.002) -.001 (.002) .001 (.002) .516 (.405) -.004 (.001) .001 (.002) -.096 (.812) -.001 (.001) .004 (.001) -1.58 (.694) .010 (.002) -.006 (.002) .010 (.002) .008 (.001) .005 (.002) .010 (.003) -.012 (.005) .007 (.003) E N V I R O N M E N T Pop Med Inc Black Exp Rev Medc aid South Individual F-Tests .015 (.013) .076 (.032) 22.484,70 .537 284.895,69 .951 301.129,65 .973 228.43 5,69 25.298,66 .724 133.825,69 .899 Adj R-Square (OLS) .939 4 The Standard Errors have been truncated, so boxed values may not appear significant. Bobic, SPSA 2004 Revea ling Lea dership Page 43 Table 10: Mean Rice Cohesion Scores, by Policy, President's Agenda, and Year, Republicans Policy Type Non Agenda AG SOCWEL GOVMAN FORPOL CIVLIB GOVOPS 7.25 (7) 62.34 (42) 68.01 (55) 68.64 (41) 51.34 (5) 53.91 (21) 1981 Agenda Non Agenda 1982 Agenda 64.17 (10) 77.23 (23) 88.00 (14) 63.98 (9) -24.53 (1) 100 (1) 30.71 (6) 54.63 (36) 66.32 (43) 45.52 (18) 28.68 (19) 57.01 (11) 32.08 (1) 59.56 (10) 60.19 (22) 61.10 (22) 14.79 (2) 68.72 (5) Mean Rice Score in Table, Parenthetical values indicate number of bills. Bobic, SPSA 2004 Revea ling Lea dership Page 44 Table 11: Model Terms: Republican Cohesion, 97th Congress Dim Vars B Se(B) Wald P(Wald) Odds Ratio E N V I R Intercept 10.986 1.832 35.981 .000 999.000 GDP -0.003 .001 7.321 .007 .997 Unemp .208 .047 19.412 .000 1.231 CPI(Urban) -0.016 .034 .216 .642 .984 B I L L S P E C I F I C AG -0.592 .086 47.356 .000 .553 SOCWEL .054 .067 .652 .419 1.055 GOVMAN .379 .065 33.744 .000 1.462 FORPOL .124 .069 3.180 .074 1.132 CIVLIB -0.525 .084 39.552 .000 .592 Bigbill .381 .072 28.416 .000 1.463 New bill -0.188 .065 8.576 .003 .828 Salience -0.148 .046 10.439 .001 .863 B A K E R Year -0.312 .090 12.036 .000 .732 TV .002 .003 .366 .545 1.002 Agenda .281 .042 45.308 .000 1.324 Model Diagnostics Statistic AIC (Null-Model) -2LL Value 538.52 566.517 df 14 Sig .000 Reduction in Error Measures Measure Value Somer's D .213 Gamma .217 Bobic, SPSA 2004 Revea ling Lea dership Page 45 Table 12: Mean Rice Cohesion Scores, by Policy, President's Agenda and Year, Democrats Policy Type Non Agenda AG SOCWEL GOVMAN FORPOL CIVLIB GOVOPS 13.94 (7) -41.09 (42) -61.51 (55) -54.07 (41) -11.00 (5) -51.33 (21) 1981 Agenda -45.43 (10) -51.62 (23) -46.15 (14) -47.61 (9) 9.52 (1) -14.29 (1) Non Agenda -40.57 (6) -54.99 (36) -61.66 (43) -33.07 (18) -20.23 (19) -53.32 (11) 1982 Agenda -90.91 (1) -39.59 (10) -34.88 (22) -34.88 (22) -11.36 (2) -25.20 (5) Mean Rice Score in Table, Parenthetical values indicate number of bills. Bobic, SPSA 2004 Revea ling Lea dership Page 46 Table 13: Model Terms: Democratic Cohesion, 97th Congress Dim Vars B SE(B) Wald p(Wald) Odds Ratio E N V I R Intercept 6.791 1.669 16.553 .000 889.214 GDP -0.006 .002 31.683 .000 .993 Unemp -0.036 .046 .618 .431 .964 CPI(Urban) .138 .034 16.624 .000 1.148 B I L L S P E C I F I C AG .339 .089 14.424 .000 1.405 SOCWEL -0.099 .067 2.178 .140 .906 GOVMAN -0.228 .066 12.024 .001 .795 FORPOL .087 .070 1.545 .214 1.091 CIVLIB .682 .087 61.983 .000 1.977 Bigbill .072 .062 1.328 .249 1.074 New bill .193 .064 9.212 .002 1.213 Salience .017 .045 .141 .707 1.017 B A K E R Year .153 .089 2.954 .0857 1.165 TV .008 .003 10.124 .002 1.008 Agenda .087 .039 4.794 .028 1.091 Model Diagnostics Statistic Value AIC (Null-Model) 280.80 -2 LL Chi-Square 308.801 Reduction in Error Measures Measure Value Somer's D .175 Gamma .179 df 14 P .000 Bobic, SPSA 2004 Revea ling Lea dership Page 47 Table 14: Baker's Win and Loss Rates, by Policy and President's Agenda, for 1981 and 1982 1981 Policy Type Nonagenda Agenda Nonagenda Agenda 1982 Loss Win Loss Win Loss Win Loss Win AG 2 6 2 8 3 3 1 0 SOCWEL 7 37 2 22 14 27 3 7 GOVMAN 8 49 0 14 8 35 4 18 FORPOL 9 34 3 6 5 14 2 20 CIVLIB 0 5 1 0 6 15 1 1 GOVOPS 5 16 0 1 3 8 0 5 Table 15: Model Fit Statistics: Chi-Square and CAIC Model Type Chi-Sq uare df P(Chi Square) CAIC No Interactions 378.272408 364.323403 378.530407 363.343402 377.974403 .8517 .9170 .8410 .9171 .8097 412.18** 429.63 418.23 435.58 440.91 Year by Policy Year by Reagan's Agenda Year by Policy Year by Agenda Year by Policy by Agenda Bobic, SPSA 2004 Revea ling Lea dership Page 48 Table 16: Best Fitting Model of Winning in the 97th Congress Classification Table, Baker Wins Predicted Actual Loss Loss Win 19 10 Win 54 341 Bobic, SPSA 2004 Revea ling Lea dership Page 49 Table 17: Model Coefficients, Baker Success in the 97th Congress Dim Vars B SE(b) Wald Sig Odds E N V I R O N Intercept 28.911 15.872 3.318 .068 999.00 GDP -0.029 0.009 9.587 .002 .972 Unemp 0.171 0.401 .181 .671 1.186 Fresh 0.227 0.040 31.659 .000 1.255 CPI(Urban) 0.651 0.282 5.327 .021 1.917 B I L L S P E C I F I C Clause n(5df) 3.749 .586 AG -0.599 0.461 1.690 .194 .549 SOCWEL -0.087 0.302 .084 .773 .916 GOVMAN 0.395 0.295 1.789 .181 1.484 FORPOL 0.206 0.325 .402 .526 1.228 CIVLIB 0.292 0.497 .346 .556 1.339 Bigbill 2.151 1.049 4.199 .040 8.594 New bill 0.108 0.570 .036 .851 1.114 Salient -0.466 0.376 1.532 .216 .627 B A K E R Year -0.811 0.785 1.068 .301 .444 TV 0.041 0.023 3.133 .077 1.042 Agenda 0.547 0.345 .025 .068 1.056 Model Diagnostics: Statistic Value -2LL 68.065 Goodness of Fit 378.272 Reduction in Error Measures: Statistic Value Somer's D 0.52 Gamma 0.85 df 15 408 Sig .000 .852 Bobic, SPSA 2004 Revea ling Lea dership Page 50 List I: Regression Variables to Explain Coalitions Model Component Variable Description Party RepLead Partisan DemLead Repcom 1 if Republican, 0 if Democrat. 1 if senator is of the 5 party leadership positions, 0 otherwise. 1 if senator is of the 5 party leadership positions, 0 otherwise. 1 if Republican committee chair or committee minority ranking member, 0 Else. 1 if Democratic committee chair or committee minority ranking member 0 Else. Americans for Constitutional Actio n score for previous year. Americans for Democratic Action Score for previous year. Presidential Support Sco re for previous year. Conservative Coalition Supp ort Score for previo us year. Years of service in the Senate. Percent of employees in state wh o work for the federal government. AFL-CIO Voting Score for the previo us year. Chamber of Commerce Score fo r the previous year. Party Unity Score f or the previous year. For analytical purposes, it made sense to multiply all Democratic PU scores by -1, to create a linear scale. As a raw variable, it is a U shape. State Pop ulation, in 1 00,000 s. Federal Spending Per Capita in state. State Median Income. Percent of State population that is African-American. States ranked in terms of their expe nditures for social programs, from Highest (1) to lowest (50) States ranked in terms of revenues fro m the federal government for soc ial programs from highest (1) to lo west (50). Medicaid benefits per capita in state. 1 if senator is From the tra ditionally defined So uth, 0 Else Demcom ACA ADA Personal PRESSUP CONSCOAL Tenure Govemp AFLCIO CCUS PARTYUN POP FEDPC$ Environmental Med Inc Black Exp Rev MEDCAID South Bobic, SPSA 2004 Revea ling Lea dership Page 51 List II: Factors explaining Cohesion toward Baker’s Position Model Dimension Variables Operational Definition GDP Environmental Unemployment Inflation Rate Gross Domestic Product for the US reported quarterly. Monthly Unemployment Rate, as reported by US Government. Monthly levels of the Consumer Price Index for urban commodities, by month. Bill is coded into one of the five dummy variable s. SOCWEL, 1 if SOCWEL, -1 if GOVOPS, GOVMAN, 1 if GOVMAN, -1 if GOVOPS, FORPOL, 1 if FORPOL, -1 if GOVOPS, CIVLIB, 1 if CIVLIB, -1 if GOVOPS, Agriculture, 1 if AG, -1 if GOVOPS. 1 if the bill increases or decreases spending more than 10 percent over the previous year, 0 otherwise. 1 if the bill is a new program or a new grant of power, 0 otherwise. 1 if issue has 5 roll calls and TV coverage, 0 otherwise. 0=first year of session (1977; 1981) 1=second year of session (1978; 1982) Count of television appe arances per mon th, from Vander bilt records. 1 if on president's agenda 0 otherwise. Policy Category Bill Specific Size Newness Salience Year Baker TV appearances Agenda Bobic, SPSA 2004 Revea ling Lea dership Page 52 List III: Success in the United States Senate Model Dimension Variables Operational Definition GDP Gross Domestic Product for the US reported monthly. Monthly unemployment rate, as reported by US Government. 1 if majority support Baker 0 else 1 if Majority support Baker 0 else Monthly levels of the Consumer Price Index for urban commodities, by month. Same definitions as for Cohesion 1 if bill alters spending more than 10%, 0 otherwise. 1 if the bill is a new program or a new grant of power, 0 otherwise. Same definition as for cohesion. 0=first year of session (1977; 1981) 1=second year of session (1978; 1982) Count of television appearances, monthly (Vanderbilt). 1 if on president's agenda 0 otherwise. Unemployment Environmental Republican Freshman Support Democrat Freshman Support Inflation Rate Policy Category Size Bill Specific Newness Salience Year Baker TV appearances Agenda Bobic, SPSA 2004 Revea ling Lea dership Page 54 BIBLIOGRAPHY _____, 1981. "Abortion Fight: Taste of Things to Come?", in Congressional Quarterly's Weekly Report, September 12, 1981; p. 1744. Akaike, H. 1973. "Information Theory and an Extension of the Maximum Likelihood Principle," in Second International Symposium on Information Theory. B.N. Petrov and F. Csaki (eds.) Budapest: Academiai Kaido, 267-281. Andrain, Charles F., 1964. "A Scale Analysis of Senators' Attitudes Toward Civil Rights," Western Political Quarterly 17:3:488-503. Annis, Lee, 1995. Howard Baker: Conciliator in an Age of Crisis. New York: Lenham Books. Arieff, Irwin B., 1981. "Under Baker's Leadrship Senate Republicans Maintain Unprecedented Voting Unity," in Congressional Quarterly's Weekly Report September 12; 1743-47. Asher, Herbert A., Herbert Weisberg, John Kessel, and W. Phillips Shively, 1984. Theory Building and Data Analysis in the Social Sciences. Knosville: The University of Tennessee Press. Bozdogan, Hamparsum, 1990. "On the Information-Based Measure of Covariance Complexity and Its Application to the Evaluation of Multivariate Linear Models," in Communications in Statistics, Theory and Methods 19:1:221-278. Bozdogan, Hamparsum, 1994. "Multivariate Regression", a MATLAB module, presented in class, Spring, 1994, University of Tennessee. Bozdogan, Hamparsum, 1987. "ICOMP" A New Model-Selection Criterion," in Classification and Related Methods of Data Analysis, Hans Herman Bock, ed. Amsterdam: NorthHolland, 599-608. Chappell, Henry and Motoshi Suzuki, 1993. "Aggregate Vote Functions for the U.S. Presidency, Senate and House," in Journal of Politics 55:1:207-217. Chappell, Henry and Motoshi Suzuki, 1991. "Aggregate Vote Functions for the U.S. Presidency, Senate and House," manuscript submitted for publication. Chappell, Henry, 1990. "Economic Performance, Voting, and Political Support: A Unified Approach," In The Review of Economics and Statistics, 72:313-320. Clausen, Aage, 1973. How Congressmen Decide. New York: St. Martins. Clausen, Aage R. and Clyde Wilcox, 1987. "Policy Partisanship in Legislative Leadership and Recruitment," Legislative Studies Quarterly 12: 243-263. Clausen, Aage, 1972. "State Party Influence on Congressional Policy Decision" in Midwest Journal of Political Science 16:1:77-101. Cohodas, Nadine, 1981a. "Strategic Choice Looms on Legal Services," in Congressional Quarterly's Weekly Report July 11, 1981; p. 1255. Bobic, SPSA 2004 Revea ling Lea dership Page 55 Crotty, William J. and Gary Jacobson, 1980. American Parties in Decline. Boston: Little, Brown and Company Davidson, Roger H., 1989. "The Senate: If Everyone Leads, Who Follows?" in Lawrence Dodd and Bruce Oppenheimer, Congress Reconsidered, 4th edition. Washington, D.C.: CQ Press. Deering, Chris, 1989. Congressional Politics. Chicago: Dorsey Press. Dodd, Lawrence, 1983. "Coalition-Building by Party Leaders: A Case Study of House Democrats," Congress and the Presidency 10: 2: 147-168. Donnelly, Harrison, 1981. "Scaled Down Block Grants Near Enactment," in Congressional Quarterly's Weekly Report July 4, 1981; p. 1182. Fleisher, Richard and Jon Bond, 1988. "Are There Two Presidencies? Yes, But Only for Republicans," in Journal of Politics 50:747-767. Fliesher, Richard and Jon R. Bond, 1993. "Assessing Presidential Support in the House II: Lessons from George Bush," American Journal of Political Science 36: 2: 525-541. Fliesher, Richard and Jon R. Bond, 1986. "Presidential-Congressional Relations on Economic Votes, 1957-1980," in Polity 18:498-513. Fliesher, Richard and Jon R. Bond, 1983. "Assessing Presidential Support in the House: Lessons from Reagan and Carter," Journal of Politics 45: 745-758. Fliesher, Richard and Jon R. Bond, 1982. "Beyond Committee Control: Floor Leadership in the House and Senate," American Politics Quarterly 11: 2: 131 ff. Fliesher, Richard, 1984. "Economic Benefit, Ideology and Senate Voting on the B-1 Bomber," American Politics Quarterly 13: 200 ff. Forshee, Richard A. and Russell D. Ranka, 1991. "Do Presidents Have Honeymoons?" Paper presented at the 1991 Annual Meeting of the Midwestern Political Science Association. Gibbons, Jean Dickenson, 1993. Nonparametric Measures of Association. London: Sage. Gregg, Gail, 1981. "GOP Senators Successful in Heading Off Attempts to Restore Veterans' Funds," in Congressional Quarterly's Weekly Report March 28, 1981; p. 547-548. Gross, Donald A., 1984. "Changing Patterns of Voting Agreement Among Senatorial Leadership," Western Political Quarterly 37: 120-142. Grumm, John G., 1963. "A Factor Analysis of Legislative Behavior," Midwestern Journal of Political Science 7: 336-356. Grumm, John G., 1965. "The Systematic Analysis of Blocs in the Study of Legislative Behavior," Western Political Quarterly 18: 2: 350-362. Bobic, SPSA 2004 Revea ling Lea dership Page 56 Hall, Richard, 1992. "Measuring Legislative Influence," in Legislative Studies Quarterly 17:2:205232. Hayes, Stephen, Stephen Shull Susan Howell and John Flaxbeard, 1984. "Presidential Support Among Senate Leaders and Followers,' American Politics Quarterly 12: 195 ff. Hibbing, John R., 1991. "Contours of the Modern Congressional Career," American Political Science Review 85: 405-428. Hinckley, Barbara, 1981. Congressional Elections. Washington D.D.: CQ Press. Hinckley, Barbara and Paul Brace, 1991. "The Structure of Presidential Approval: Constraints Within and Across Presidents," Journal of Politics 53: 4: 993-1017. Hinckley, Barbara, 1972. "Coalitions in Congress: Size and Ideological Distance," American Journal of Political Science 16: 197-207. Hocking, R. R. and William Smith, 1968. "Estimation of Parameters in the Multivariate Normal Distribution With Missing Observations," in American Statistical Association Journal 63:159-173. Hurley, Patricia, 1991. "Partisan Representation, Realignment and the Senate in the 1980s," in Journal of Politics 53:1:3-33. Hurley, Patricia and Richard Wilson, 1989. "Partisan Voting Patterns in the U.S. Senate 1877-1986," Legislative Studies Quarterly 14: 2: 225-250. Hurley, Patricia, 1989. "Parties and Coalitions in Congress," in Deering, Christopher (ed), Congressional Politics. Chicago: Dorsey Press. Jackson, John E., 1971. "Statistical Models of Senate Roll Call Voting" in American Political Science Review, 65:2:451-470. Jenkin-Smith, Hank C., Gilbert St. Clair, and Brian Woods, 1991. "Explaining Change in Policy Subsystems: Analysis of Coalition Stability and Defection over Time," American Journal of Political Science, 35: 4: 851-880. Koford, Kenneth, 1989. "Dimensions of Congressional Voting," American Political Science Review 83: 949-962. Krehbeil, Keith, 1993. "Constituency Characteristics and Legislative Preferences" in Public Choice 76:21-37. LaFeber, Walter, 1978. The Panama Canal: The Crisis in Historical Perspective. New York: Oxford University Press. LeLoup, Lance, 1977. "The Impact of Dominant Spending Patterns on Senate Support: An Examination in Three Policy Arenas," American Politics Quarterly 5: 291-336. Lepper, Susan, 1974. "Voting Behavior And Aggregate Policy Targets," in Public Choice 18:6782. Bobic, SPSA 2004 Revea ling Lea dership Page 57 Light, Paul C., 1982. "Passing Nonincremental Policy: Presidential Influence from Kennedy to Carter," Congress and the Presidency 9: 1: 61-82. Light, Paul C., 1989. "The Focusing Skill and Presidential Influence in Congress," in Christopher Deering, Congressional Politics. Chicago: Dorsey Press. Lowi, Theodore, 1972. "Four Systems of Policy, Politics and Choice" in Public Administration Review July/August, 1972; pp. 298-309. Lyons, William, 1977. "Per Capita Index Construction: A Defense" in American Journal of Political Science 21:1:177-183. MacRae, Duncan, 19 66. "Cluster Analysis of Congressional Votes with the BC-TRY System," Western Political Quarterly 19: 4: 631-638. Malbin, Michael, 1989. "Legislative-Executive Lessons from the Iran-Contra Affair," in Lawrence C. Dodd and Bruce Oppenheimer, eds., Congress Reconsidered Fourth Edition. Washington D.C.: Congressional Quarterly Press. McElroy, Marjorie, 1977. "Goodness of Fit For Seemingly Unrelated Regressions: Glahn's R2 and Hoopers r2 ," in Journal of Econometrics 6:381-387. Morrison, Richard, 1972. "A Statistical Model for Legislative Roll Call Analysis," Journal of Mathematical Sociology 2: 235-247. Nownes, Anthony, 1992. "Primaries, General Elections and Voter Turnout: A Multinomial Logit Model of the Decision to Vote," American Politics Quarterly 20: 2: 205-226. Oleszek, Walter, 1989. Congressional Procedures and the Policy Process. Washington D.C.: CQ Press. Ornstein, Norman J., Robert L Peabody and David W. Rhode, 1989. "Change in the Senate: Toward the 1990s", in Dodd, Lawrence and Bruce Oppenheimer, eds, Congress Reconsidered: Washington D.C.; CQ Press. Overby, L. Marvin, 1993. "Political Amateurism, Legislative Inexperience, and Incumbency Behavior: Southern Republican Senators, 1980-1986" in Polity 25:3:401-420. Plattner, Andy, 1982a. "Senate Holds Up Lame Duck Adjournment," in Congressional Quarterly's Weekly Report, December 18; p. 3039. Plattner, Andy, 1982b. "Senate Upholds Veto of Oil Allocation Bill," in Congressional Quarterly Weekly Report March 27, 1982; p. 663. Poole, Keith and Howard Rosenthal, 1985. "A Spatial Model for Legislative Roll Call Analysis," American Journal of Political Science 29: 357-384. Poole, Keith, 1984. "Least Squared Metric, Unidimensional Scaling of Multivariate Linear Models," Psychometrika 55: 123-149. Bobic, SPSA 2004 Revea ling Lea dership Page 58 Poole, Keith and Howard Rosenthal, 1987. "An Analysis of Congressional Coalition Patterns: A Unidimensional Spatial Model," in Legislative Studies Quarterly 12:55-75. Poole, Keith, 1981. "Dimensions of Interest Group Evaluations of the U. S. Senate," American Journal of Political Science 25:49-67. Poole, Keith and Howard Rosenthal, 1984. "The Polarization of American Politics," in Journal of Politics 46:1061-1079. Rice, Stuart, 1923. "The Identification of Blocs in Small Political Bodies," in American Political Science Review 21:3:619-627. Ripley, Randall, 1988. Congress: Process and Policy. New York: W.W. Norton and Co. Sarasohn, Judy, 1982. "Gas Tax-Transport Bill Wins Approval in House, But Is Stalled by Senate Foes," in Congressional Quarterly's Weekly Report, December 11, 1982; p. 2991. Schmidt, Peter, 1977. "Estimation of Seemingly Unrelated Regressions With Unequal Numbers of Observations," in Journal of Econometrics 5:365-377. Sinclair, Barbara, 1989. "Leadership Strategies in the Modern Congress," in Christopher Deering, ed., Congressional Politics, Chicago: Dorsey Press. Sinclair, Barbara, 1983. Majority Leadership in the U.S. House. Baltimore: Johns Hopkins University Press. Sinclair, Barbara, 1989. The Transformatoin of the U.S. Senate. Baltimore, Johns Hopskins University Press. Sinclair, Barbara, 1981. "Agenda and Alignment Changes in the House of Representatives," in Dodd and Oppenheimer (eds.), Congress Reconsidered. Washington D.C.: CQ Press. Sinclair, Barbara, 1981. "Agenda and Alignment Changes in the House of Representatives," in Dodd and Oppenheimer (eds.), Congress Reconsidered. Washington D.C.: CQ Press. Sinclair, Barbara, 1982. Congressional Realignment 1925-1978. Austin, Tex.: University of Texas Press. Sinclair, Barbara, 1982. "Coping with Uncertainty," in Barbara Sinclair, Congressional Realignment 1925-1978. Austin, Tex.: University of Texas Press. Smith, Steven, 1981. "The Constituency and Ideological Structure of U.S. Senate Voting Alignments 1957-19 76," American Journal of Political Science: 25: 780-795. Sullivan, Terry, 1991. "The Bank Account Presidency," American Journal of Political Science 35: 3: 686-723. Sullivan, Terry, 1987. "Presidential Leadership in Congress: Securing Commitments," in McCubbins and Sullivan, Congress: Structure and Policy. New York: Cambridge University Press. Bobic, SPSA 2004 Revea ling Lea dership Page 59 Tate, Dale, 1982b. "Reagan Requests Lame Duck Session on '83 Spending Bills," in Congressional Quarterly's Weekly Report, September 18, 1982; p. 2337. Tate, Dale, 1982a. "Reagan Pressed to Enter Budget Talks," Congressional Quarterly's Weekly Report March 27, 1982p. 671. Tate, Dale, 1981a. "Republicans Press Alternatives" in Congressional Quarterly's Weekly Report June 20; p. 1080. Tate, Dale, 1981b. "Conferees Nearing Accord on Specific Cuts," in Congressiona l Quarterly's Weekly Report July 25, 1981; p. 1327. Tate, Dale, 1981c. "GOP Leaders Claim Consensus on Budget," in Congressiona l Quarterly's Weekly Report, October 24; p. 2052. Todd, John R, and Kay Ellis, 1974. "Analyzing Factional Patterns in State Politics: Texas, 19441972" in Social Science Quarterly 55:3:718-731. Van Horn, Carl E., 1989. "Congressional Policymaking: Cloakroom Politics and Policy," in Christopher Deering, ed., Congressional Politics. Chicago: Dorsey Press. Van Doren, Peter, 1990. "Can We Learn the Causes of Congressional Decisions from Roll Call Data?" Legislative Studies Quarterly 15: 311 ff. Wagner, James R., 1977. Report in Congressional Quarterly's Weekly Report, June 18; p. 1223b. Wagner, James R., 1977. Report in Congressional Quarterly's Weekly Report, May 21, 1977; p. 962a. Ward, Daniel S., 1991. "The Continuing Search for Party Influence in Congress: A View from the Committees" in Legistive Studies Quarterly, 18:2:211-230. Weisberg, Herbert, 1978. "Evaluating Theories of Congressional Roll Call Voting," American Journal of Political Science 22: 3: 554-577. Weisberg, Herbert, 1983. "Alternative Baseline Models and Their Implications for Understanding Coalition Behavior in Congress," Journal of Politics 45: 3: 657-671. Welborn, David M., 1993. "Republican Party Leadership in the Senate: 1977-1985", paper prepared for presentation at the annual meeting of the Sou thern Political Science Association. Whittle, Richard, 1981. "Bypassing Traditional Rules, Senate Acts Before House on Foreign Aid Bill," in Congressional Quarterly's Weekly Report November 21, 1981; p. 2282. Wilcox, Clyde and Aage Clausen, 1991. "The Dimensionality of Roll Call Voting Reconsidered," Legislative Studies Quarterly 16: 3: 393-406. Argu es Zellner, Arnold, 1962. "An Efficient Method of Estimating Semmingly Unrelated Regressions and Tests for Aggregation Bias," in American Statistical Association Journal 57:350-375.
Pages to are hidden for
"Revealin g Leade rship Th rough Roll Ca ll Analysis"Please download to view full document