A Sabermetric Test of Neustadt’s Skills Hypothesis or Why Ronald Reagan is like Bobby Cox and Lyndon Johnson is like Joe Torre Jon R. Bond Texas A&M University [email protected] and Manny Teodoro Texas A&M University [email protected] Prepared for Presentation at the 73rd Annual Meeting of the Midwest Political Science Association Chicago, Illinois April 16-19, 2015 Abstract A Sabermetric Test of Neustadt’s Skills Hypothesis* or Why Ronald Reagan is like Bobby Cox and Lyndon Johnson is like Joe Torre Jon R. Bond Texas A&M University And Manny Teodoro Texas A&M University This paper presents a new test of Neustadt’s Political Skills Hypothesis. We believe an indirect test comparing actual wins against expected wins offers the most systematic test of this key hypothesis. We update and replicate findings of Wins Above Expectations (WAE) from the traditional analysis of residuals from multiple regression. Then we adapt the Pythagorean Expectations formula from sabermetrics to estimate WAE based on votes supporting and votes opposing the president. Both approaches predict presidential success rates with high accuracy across all years and presidents. The WAEs from the two approaches, however, are uncorrelated. The two approaches also identify different presidents as above and below expectations. This suggests that the two approaches model different aspects of the underlying data generation process. More research is needed to sort out what all this means. *We are grateful to the Department of Political Science Texas A&M University for supporting this research. A Sabermetric Test of Neustadt’s Skills Hypothesis “The other teams could make trouble for us if they win.” Yogi Berra Presidential-congressional relations, like baseball, is all about winning. What the manager of a baseball team needs most from his players is runs: his hitters must score runs, his pitchers and defenders must prevent them. What the president needs most from members of Congress is their votes: he must get members to vote for his favored legislation and against legislation he opposes. But important as winning is, what professional observers (sports geeks and political scientists) really find most intriguing is Wins Above Expectations (WAE)—did the team do better or worse than expected? If a team wins more games than expected, was it due to an advantageous set of circumstances, simple “luck,” or did the manager’s skill and leadership produce wins when the team should have lost? Using different statistical methods, sabermetricians and political scientists have attempted to answer these questions for their respective fields.1 This paper applies sabermetric methods to analyze presidential success in Congress from Eisenhower to Obama. Specifically, we adapt Bill James’ (1982) Pythagorean Expectations (PE) formula, which estimates how many games a baseball team should have won based only on the number of runs scored and runs allowed, to estimate WAE for presidents. If a roll call vote on which the president expresses a position is analogous to a game, and votes for and against the president are analogous to runs scored and runs allowed, can the PE formula predict presidential success rates over the course of a season (year) as accurately as it does for professional baseball teams? If so, how well do WAE from PE correlate with residuals from multiple regression models used by political scientists to identify presidents who won significantly more or less often that predicted (Bond and Fleisher 1990, chap. 8; Cohen, Bond, and Fleisher 2013a; Fleisher and Bond 1983, 1992; Fleisher, Bond, and 1 Bill James named his approach to analyzing baseball “sabermetrics” in honor of the Society for American Baseball Research (SABR). 1 2 Wood 2008)? What are the lifetime WAEs of presidents from Eisenhower to Obama? Which presidents, if any, stand out as uncommonly successful or unsuccessful? The paper begins with a discussion of the fundamentals of presidential success in Congress. Political science research finds that a regression model with three independent variables—party control, presidential approval, and party polarization—and interactions successfully models the fundamentals of presidential success on roll call votes in Congress. We update this model through 2014 (Obama’s sixth year), and use the residuals to estimate WAE relative to this updated baseline. Next, we adapt the Pythagorean Expectations (PE) approach to predict presidential success rates based only on votes won and lost, and present WAE relative to this parsimonious atheoretical baseline. We find that both approaches predict presidential success rates with high accuracy across all years and presidents. WAE from the two approaches, however, are uncorrelated. This suggests that the two approaches model different aspects of the underlying data generation process. The Fundamentals of Presidential-Congressional Relations Richard Neustadt is rightly considered the dean of the behavioral study of the presidency. His message is clear and compelling. To understand the presidency, study presidential behavior—what the president “can do, as one man among many, to carry his own choices through that maze of personalities and institutions called the government of the United States” (Neustadt 1960, i). Although he did not present a formal theory or use the language of the scientific method, his insightful analysis generated hypotheses that subsequent generations of political scientists set about testing. The three that have received the most attention include: 1. The Political Party Hypothesis (for which he expected null results): “What the Constitution separates our political parties do not combine” (Neustadt 1960, 33); 2. The Public Approval Hypothesis: ceteris paribus, popular presidents are more successful than are unpopular presidents (chap. 5); and 3. The Political Skills Hypothesis: ceteris paribus, presidents who are highly skilled at bargaining and persuasion will be uncommonly successful while unskilled presidents will be unsuccessful (chap. 4). 3 As tests of the Neustadtian theory of presidential power2, the Political Skills Hypothesis is clearly the most crucial because it focuses directly on what the president does to maximize his influence. With respect to public approval, Neustadt (1960, 87) did not expect a strong “one-to-one relationship”. Rather, he viewed public approval “as a conditioner, not the determinant” of presidential success. The Political Parties Hypothesis is the least important test. Neustadt expected null results, but rejecting the null hypothesis does not necessarily undermine the theory. What evidence has 40 years of scientific research testing these hypotheses produced? Overall, the correlation between the importance of the hypothesis to the theory and empirical support for it is -1.0. Political Parties Political science research has found strong support for the Political Parties Hypothesis. Large-n quantitative studies consistently reject the null hypothesis that parties do not provide a bridge between the president and Congress. Indeed, we find that party is by far the strongest determinant of presidential success in Congress— support from members of the president’s party is statistically and substantively higher than that of members of the opposition (Edwards 1989), and the president wins significantly more roll calls if his party has a majority (Bond and Fleisher 1990). Although Neustadt’s observation about the weakness of American parties was accurate, the pattern of robust party effects held during the period of his analysis—mid-twentieth century—when parties were at their weakest. Rising party polarization in Congress from the mid-1980s to the present has magnified the effects of party, though important chamber differences have emerged (Bond, Cohen, and Fleisher 2014, 2015; Bond, Fleisher, and Cohen 2012; Cohen, Bond, and Fleisher 2013a, 2013b). We have sound theoretical reasons to explain the strong influence of party on presidential support in Congress. Most obvious is simple arithmetic—majority presidents have more members on the floor with incentives to support their positions than do minority presidents. Because members of the same political party 2 Neustadt’s analysis, of course, is not a scientific theory, as we understand it in contemporary political science. Yet, the generation of testable hypotheses is arguably the most important thing that we want our theories to do. 4 must satisfy similar electoral coalitions, they have a wide range policy preferences in common. Consequently, majority presidents win more because they have more members on the floor who agree with their proposals. Furthermore, because the president’s co-partisans must run for reelection on his record as well as their own, they have an electoral incentive to help him succeed. Thus, simple arithmetic explains much of why majority presidents win more votes in Congress than do minority presidents. Yet, arithmetic is not the only, or even most important, reason why majority presidents win more floor votes than do minority presidents. A political party is more than just a bundle of shared preferences. More important than numbers, is control of key levers of power in Congress. Political parties have long been conceived of as competing teams (Downs 1957). Like sports teams, political parties are institutions with rules and procedures that channel and constrain behavior (Aldrich 2011). The goal of party institutions is to win—win elections and enact policies favored by party members. As a result, parties in Congress regularly take opposing positions on issues that have nothing to do with policy or ideology, just because they don’t want the other team to win (Lee 2009), and to score points for the next election. Majority party agenda control is a central component of the two leading partisan theories of Congress—Conditional Party Government (CPG) theory (Aldrich and Rohde 2000; Rohde 1991) and Cartel Theory (Cox and McCubbins 2005). When it comes to winning or losing roll call votes, then, the main advantage of majority party control is that the president’s team controls key levers of power in Congress—committees, access to the floor, rules governing debate, and which amendments are considered.