Opportunism in Polarization: Presidential Success in the U.S. Senate, 1953-2006∗ Matthew N. Beckmann Assistant Professor Department of Political Science & Center for the Study of Democracy University of California, Irvine [email protected]

Vimal Kumar Ph.D. Candidate Department of Economics University of California, Irvine [email protected]

∗ Paper

prepared for the Symposium on Legislatures Conference, sponsored by the Center for the Study of Law, the Gould School of Law at the University of Southern California. For their thoughtful comments, we thank Bernard Grofman, Anthony McGann, Gary Cox, and Mathew McCubbins. Also, UC Irvine and UCI’s Center for the Study of Democracy generously supported this project.

1

Abstract That Congress has experienced increased polarization is clear, and burgeoning is the literature investigating its causes and consequences. Here we examine a counterintuitive wrinkle on the latter. Drawing from a simple game-theoretic model in which a president strategically allocates scarce “political capital” to induce changes in legislators’ votes, we show congressional polarization can actually improve a president’s prospects for winning key roll-call votes – a hypothesis that emerges inasmuch as polarization enables presidents to concentrate their resources lobbying fewer members (compared to a more homogenous chamber). We test this hypothesis by investigating presidents’ success on “key” Senate roll-call votes, 1953-2006.

2

That the last half-century has seen increased polarization in Washington is clear. Holders of most key posts and key votes just a few decades back, conservative Democrats and liberal Republicans are now few and far between, so much so that one pundit recently referred to the former as “endangered species” and the latter as “essentially extinct” (Roll Call, October 27, 2005). Replacing these moderates have been resolute ideologues and loyal partisans (Fleisher and Bond (2004); McCarty et al. (2006); Poole and Rosenthal (1984, 1997); Sinclair (2006); Theriault (2004)). Thus Jim Hightower’s aphorism that “There’s nothing in the middle of the road but yellow stripes and dead armadillos” remains instructive, and more so with each passing year. In light of Congress’ increasingly polarized character, a burgeoning literature has sought to uncover the causes and consequences. While a myriad of factors have been identified as contributing to congressional polarization (Carson et al. (2007); Fleisher and Bond (2004); Brady and Han (2007), Jacobson (2000); McCarty et al. (2006); Poole and Rosenthal (1997); Sinclair (2006); Stonecash et al. (2003); Theriault (2004)), the hypothesized effect is comparatively clear: legislative gridlock. Partisan polarization (via divided government) is thought to engender gridlock by promoting posturing over compromising (Gilmour (1995); Groseclose and McCarty (2001); Sinclair (2006)), and ideological polarization is predicted to encourage gridlock by reducing the range of status quos that can be beat by a coalition preferring something else (Brady and Volden (1998); Krehbiel (1998)). But, of course, gridlock is not an all-or-nothing proposition. Although polarization certainly inhibits lawmaking(Binder (1999, 2003); Coleman (1999); Edwards III et al. (1997); Howell et al. (2000); Jones (2001); Kelly (1993)), significant laws continue to pass – under unified

3

and divided government, and even in the face of substantial polarization (see esp. Mayhew (1991)). This paper considers these exceptions to the general rule. Specifically, developing a simple game-theoretic model in which the president allocates scarce political capital to induce changes in senators’ votes, we show how a polarized chamber, compared to one with more homogenous preferences, can actually improve a president’s prospects for prevailing on important roll-call votes and passing preferred legislation.1 This hypothesis is tested against data on presidents’ success on key Senate roll-call votes from 1953-2006.

1

Polarization as a Presidential Opportunity

To investigate the relationship between polarization, presidents, and policymaking, let us consider a simple, complete information game-theoretic model with two player-types. The first of these is the President, P, who has an ideal policy preference, p. Building on Neustadt’s (1960) seminal insight that presidential power turns on informal persuasion (rather than formal prerogatives), we assume the president is exogenous to the lawmaking game per se but wields “political capital” he can allocate to lobby lawmakers.2 The president values both 1 We

focus the model and empirics on the U.S. Senate not only to build on previous work (see esp. Brady and Volden (1998); Krehbiel (1998)), but also because the House majority party’s role in agenda-setting is sure to condition polarization’s legislative impact in the lower chamber (see Aldrich 1995; Aldrich and Rohde (2000, 2001) ; Cox and McCubbins (2005), Cox and McCubbins (1993); Rohde (1991)). So even as we think our model offers general insights into polarization’s legislative impact, we also believe the open rules that govern the Senate make it more directly applicable. 2 To elaborate, we conceptualize “political capital” as the selective-incentives presidents control and may allocate to induce changes in lawmakers’ votes. So even as tactical applications may vary – e.g., arm-twisting, brow-beating, horse-trading, etc.– all serve the same purpose: allocating the presidency’s unique reservoir of persuasive resources to bargain with particular lawmakers. See also Beckmann (2008); Beckmann and McGann (2008).

4

the policy outcome and his political capital; his utility can be given as:

uP = −A · (y − p)2 + (WP − B)

(1)

Here, y is the policy outcome, B the investment of political capital needed to get legislators to adopt the president’s proposal, and A is a coefficient indicating the weight the president assigns to prevailing on the policy versus preserving his political capital. Because our interest is assessing presidents’ potential influence in Congress, we assume a non-trivial A - that is, the president is willing to invest some political capital in order to achieve a policy outcome nearer his ideal. The second player-type in our game is are lawmakers, organized along a continuum, spread across an interval [− 12 , 12 ], with a pivotal voter m. We assume each legislator has an ideal policy position, but she can be induced to change that position if “bought off” or “beat down” by the president’s political capital. So legislator i derives utility ui (x, xi , Bi ) if the policy she votes for is x and the political capital the president allocates to lobbying her is Bi . Her utility function can be given as: ui (x, xi , Bi ) = −(x − xi )2 + Bi

(2)

This equation assumes that i’s utility is additively separable in policy and capital; the policy component is quadratic in nature while the capital component is linear. We also assume that the legislator i’s utility function is single-peaked in policy, which peaks if she votes for her most preferred policy. And absent any capital, we assume legislators vote sincerely for 5

the policy nearest to their ideal. President Lobbying (Vote-Centered)

Exogenous Status Quo (sq)

President Proposes !(p,WP)

Congress Proposes 'm'

Voting Takes Place

Figure 1: The Legislative Endgame with Presidential Lobbying Turning to the game, the sequence of play is illustrated in Figure 1. After nature provides the status quo, and assuming majority rule, the basic legislative game follows two steps: the median voter begins by proposing a policy alternative, which she then supports with her (decisive) vote. In between these moves, however, the president can insert himself into the legislative process by first proposing an alternative, α, and second by lobbying lawmakers. When lobbying, we assume the president executes a vote-centered strategy - one where he allocates his political capital first with the median voter, and then, as the pivot moves toward the president’s position, with the interval of lawmakers between the median and that point. As such, if two proposed alternatives are the chamber’s median, m, and president’s proposal α, v(i)|(m,α) denotes legislator i’s preference intensity in policy for voting for m over α. v(i)(m,α) = ui (m, xi ) − ui (α, xi )

(3)

While negative vi indicates the member already supports the president’s proposal, a positive v indicates a member i’s willingness to vote for m instead of α. For these otherwise oppositional members, the president can invest political capital lobbying them to change their vote. In 6

doing so, the following equation shows the political capital the president must invest lobbying pivotal voters in order to guarantee passage of this particular α, where f (x) gives lawmakers’ preference distribution.

B(α) =

!

α 2

M =0

=

!

α 2

M =0

−[ui (α, x) − ui (0, x)]f (x)dx

α>0

(4)

α(α − 2x)f (x)dx

Again, the president follows two rules in allocating political capital: first, he only seeks minimum winning coalitions, and second, we assume lawmakers who are indifferent between alternatives will vote with the president, so the president’s goal is making pivotal voters indifferent between his proposal and the median’s. Before turning to comparative statics analysis of presidents’ (conditional) influence, a preliminary point warrants emphasis: a president’s influence depends on his willingness and ability to spend political capital. If the president either chooses not to get involved (A = 0) or lacks political capital to spend (B = 0), the median voter will propose and pass her preferred bill. In such circumstances, the chamber’s preference distribution does not matter; the president will have no influence. In other circumstances, however, the president not only seeks to exert influence on Capitol Hill, but also wields some political capital to employ to that end. Such cases have been commonplace since Franklin D. Roosevelt entered the Oval office more than 75 years back. We now turn to these cases, and in doing so uncover how presidents’ influence turns on more than his supply of political capital and the location of the pivotal voter; it also depends on

7

the level polarization. Let us explain. Up to this point we have treated the pivotal voter as an atomized individual. Yet pivotal voters do not exist in a vacuum; instead, they are defined by their position in a preference distribution, and any number of preference distributions can yield the same pivot point. For a strategic president looking to pass a particular α, this has important implications for how much political capital he will need. Indeed, we can show the capital the president requires depends, in part, on legislators’ distribution of ideal points – in other words, polarization. Equation 4 defines this amount, which, using the president’s utility function from equation 1, can be simplified as

2

uP (α) = −A · (α − p) + WP −

!

α 2

M =0

α(α − 2x)f (x)dx

(5)

Now, president will propose α∗ which maximizes his utility given in equation 5. The first order condition for the maximization can be given by:

∗

A(p − α ) −

!

α∗ 2

M =0

(α∗ − x)f (x)dx = 0

(6)

Here we can see how polarization increases presidents’ potential influence. In equation 6, the president proposes the most ambitious proposal possible (the first term) given the cost of ensuring its passage by lobbying (the second term). Notice, as the value of f (x) decreases in the neighborhood of the median – i.e., as polarization leaves pivotal voters without many like-minded colleagues – the cost of lobbying also decreases, thereby permitting the president to propose an even more ambitious proposal. Thus the general equilibrium result is clear: 8

Figure 2: Three Archetype Preference Distributions of the Legislature. polarization around the pivotal voter generally works to enhance the president’s influence. To clarify this result, let us consider three archetype distributions - one unimodal, one uniform, and one bimodal - as shown in Figure 2 and defined in Appendix A. Though the three archetype distributions all have the same median voter, they vary substantially in terms of polarization, from low (unimodal) to moderate (uniform) to high (bimodal). So assuming majority rule, by comparing presidents influence in each, we can see how polarization qua polarization affects presidents’ legislative influence, all else equal. Figure 3 illustrates the president’s influence on the policy outcome given different levels of congressional polarization, assuming the president’s goal is to get an outcome as far to the right as possible (to 12 ). Perhaps the first noteworthy point is that presidents endowed with political capital and executing a vote-centered lobbying strategy can always improve their prospects for success by lobbying; for each distribution, even modest supplies of political capital permit presidents to pull the outcome toward their preferred outcome. And, of course, 9

added supplies of political capital only serve to further enhance presidents policymaking influence.

Figure 3: Presidential Influence with Different Preference Distributions But beyond these general findings is the more intriguing result: as polarization grows, increasing are the president’s policy returns on lobbying investments. By allowing the president to focus his political resources on only a handful of swing voters, polarization permits presidents to exert greater influence than would be possible in a chamber packed with so-called centrists. So given some level of political capital, the president exerts the least influence in the non-polarized (unimodal) legislature and the greatest influence in the polarized (bimodal) one. This illustration thus shows plainly the implications of the general result proved earlier: all else equal, polarization around the pivotal voter improves presidents’ prospects for exerting influence, prevailing on key votes, and securing legislative success.

10

2

Testing Polarization and Presidential Success

In the model above, the foremost hypothesis is that a president’s prospects for securing legislative success can actually improve in a legislature characterized by polarization compared to one characterized by homogeneity. To offer a first test this prediction, let us new investigate presidents’ success on Congressional Quarterly’s tally of “key” Senate roll-call votes between 1953 and 2006, where “key votes” are designated as such because they comprise “a matter of major controversy, a test of presidential or political power, and a decision of potentially great impact on the nation and on lives of Americans” (CQ Weekly 1 January 2007, 60). We opt for this test not only because “key” votes often serve as the ultimate arbiter in presidential-congressional relations (see Bond and Fleisher (1990); Edwards (1989); Shull and Vanderleeuw (1987)), but also because this fifty-four year period includes substantial variations in senators’ preference distributions, from normal to bimodal. Returning to the model, three basic factors are thought to predict a president’s ability to prevail on crucial roll-call votes: the default preference of the pivotal voter, the polarization around her, and the president’s willingness (and ability) to spend his capital to win those votes. We measure the first two factors – the pivotal voter’s predisposition and the polarization around her – using DW-NOMINATE scores (Poole and Rosenthal (1997)). The pivotal voter’s predisposition is coded as the filibuster pivot’s “distance” from the president’s side; polarization is calculated as the “space” between the median quintile of senators (i.e., the 40th and 60th senators). In essence, the pivotal voter measure indicates how likely the pivotal voter is to support the president’s side absent any presidential lobbying while the polarization measure indicates how densely-packed other senators are between that position

11

and one closer to the president’s. Because there are no direct measures of presidents’ supply or allocation of political capital, we instead developed an indirect measure of presidents’ political capital. In doing so, we drew on Zaller (1998) argument and analysis showing that presidents’ political standing is rooted in substantive reality (see also Kramer (1971); Fiorina (1981)). For our purposes, the assumption is that presidents lobby on all key votes, but sometime those efforts are more effective than others - namely when the president enjoys high approval ratings and stronger economy. Table 1: Presidents’ Annual Political Capital, 1953-2006 Honeymoon Mean of Lowest Highest Year All Years Year Year Eisenhower .87 .67 .33 1.00 Kennedy .76 .81 .73 .94 Johnson .88 .71 .45 .97 Nixon .65 .55 0 .71 Ford na .32 0 .61 Carter .78 .51 .24 .78 Reagan .60 .57 .15 .82 Bush (41) .74 .60 .40 .74 Clinton .50 .61 .49 .74 Bush (43) .59 .56 .38 .70 Mean .71 .59 .32 .80 (s.d.) (.13) (.13) (.23) (.13)

A president’s annual political capital is measured as a factor score of his yearly average job approval (Gallup) and the nation’s change in GDP from the previous year (Bureau of Economic Analysis, U.S. Department of Commerce). It is scaled 0-1, low to high.

12

Using this measure, Table 1 displays presidents’ political capital for every administration since Dwight Eisenhower. The patterns are interesting. First, a president’s first year in office does typically yield an especially favorable political environment – a so-called “honeymoon” period. Second, presidents have varied substantially in their political capital, not only from administration-to-administration (from Gerald Ford at the bottom to John Kennedy at the top) but also from year-to-year (e.g., Reagan during the recession of 1982 versus Reagan during the recovery of 1984). So even as this indicator, like any other, is not perfect – perhaps most importantly, it does not account for the effort presidents exert lobbying on each vote – it nevertheless provides a theoretically-grounded, empirically-precise measure that comports nicely with contemporaneous accounts of presidents’ political clout. In the analysis below, we break the political capital measure into thirds – low (1), medium (2), and high (3) – to facilitate interpretation. While the three factors above – pivots, polarization, and political capital – comprise the only moving parts in our theory, previous research has identified several related factors as important to explaining presidents’ legislative fate. To isolate our model’s predictions, the analysis that follows thus includes several other germane variables as controls. The first three are dichotomous variables, one for divided government (i.e., when the president’s party was the Senate’s minority party), one for a newly-elected president’s first year in office, a period when presidents are thought to fare especially well with Congress (Beckmann and Godfrey 2007; Grossback, Peterson, and Stimson 2006; Light 1999), and finally one more to control for a president’s lame duck status (during the last year of his second term). To account for two presidencies thesis, which suggests presidents get extra support in foreign affairs compared to domestic affairs, we coded each vote as to it subject matter, foreign or domestic. 13

We also incorporate a trend variable (a natural log transformation of the Congress number) to control for non-polarization changes occurring in Washington over the last half-century, including increased incivility (Uslaner 1993) and the proliferation of interest groups (Baumgartner and Leech 1998). Finally, we include a dummy variable for the president’s party to account for potential partisan asymmetries in presidents’ success. Turning to the probit multiple regression results in Table 2, we see the model and its hypotheses find clear support on all Senate key votes, but especially on contested ones – those where fewer than 80% of both party’s members vote together. Since both sets of results are so similar, but the contested ones are more interesting and reliable, we will limit our discussion to the contested vote analysis. First, as previous research has shown, the further away the pivotal voter’s predisposition from the president’s side, the lower his chances for prevailing on “key” contested Senate votes (b = −2.66, se = 1.27, p < .05). Holding everything else at its 2006 value, the president’s predicted probability of winning a key, contested vote runs from .42 to .82 across the observed range of filibuster pivot predispositions (farthest to closest), with the median distance yielding a .51 predicted probability of presidential success. Plainly, the greater the ideological distance between the president and pivotal voter, the worse the president’s prospects for winning an important, controversial floor vote in the Senate. But if a president’s prospects for winning important votes dims as the pivotal senator’s predisposition diverges from his own, that reality is mitigated if there is lots of “open space” in the interval between them (b = 1.50, se = .63, p < .05). So even if confronting the far-off pivotal voter of 2006, the predicted probability of presidential success jumps from .29 to .49 14

Table 2: Multiple Regression (Probit) Results for Success of the President’s Position on “Key” Senate Votes, 1953-2006 Coef. (Robust SE) a All Key Votes Contested Key Votes Filibuster Pivot’s Position (DW-Nominate “distance” to the president’s side)

-2.39* (1.32)

-2.66** (1.27)

Polarization (DW-Nominate “space” between 40th and 60th senators)

1.26* (.71)

1.50** (.63)

President’s Political Capital (1-3=bottom 1/3rd (1), middle 1/3rd (2), and top 1/3rd (3))

.09* (.05)

.08* (.05)

Divided Government (1=Senate Divided)

-.09 (.20)

-.08 (.21)

Honeymoon Year (1=year of inauguration)

.23 (.17)

.16 (.19)

Lame Duck Year (1=final year of 2nd term)

.35** (0.10)

0.36** (0.11)

Foreign Policy (1=foreign policy vote)

0.09 (0.08)

0.12 (0.08)

President’s Party (1=Democrat)

-.19 (.17)

-.18 (.16)

Congress, Natural Log (natural of Congress #, 83rd to 109th)

-.04** (.01)

-.04** (.01)

Constant

18.95** (4.33)

20.38** (4.77)

N

678

644

Log Pseudolikelihood

-435.04

-412.03

* p <.10; ** p <.05 (two-tailed) Dependent variable: success of the president’s position on CQ’s “key” roll-call votes in the Senate, 1953-2006 (1 if the president’s position prevailed, 0 if it did not). The president’s preferred position was identified by CQ based on public statements, and absent that, we assumed the president’s position was his typical one (liberal for Democrats, conservative for Republicans). In the test of “contested” votes, we excluded those roll-calls where more than 80% of each party’s senators voted together. a Standard

errors are adjusted for clustering on President.

15

if the Senate’s level of polarization is extreme (the maximum observed value) rather than moderate (the median observed value). Therefore, that presidents’ distance to the Senate’s pivotal voter has correlated with increased polarization (r=.46) is, in some ways, good news for those in the West Wing; all else equal, this polarization gives presidents a better chance for passing their preferred policies.a The final important piece in our theoretical model – presidents’ political capital – also finds support in these analyses. Presidents operating under the specter of strong economy and high approval ratings get an important, albeit moderate, increase in their chances for prevailing on “key” Senate roll-call votes (b = .08, se = .05, p < .10). Figure 2 displays the substantive implications of these results in the context of polarization, showing that going from the lower-third of political capital to the upper-third increases presidents’ chances for success by 6 points. Thus political capital’s impact does provide an important boost to presidents’ success on Capitol Hill, but it is certainly not potent enough to overcome basic congressional realities. Political capital is just strong enough to put a presidential thumb on the congressional scales, which often will not matter, but can in close cases. Lastly, it is interesting to note that only two of the various control variables reliably predict presidents’ success or failure on key Senate votes. The first is the president’s lame-duck status (b = .36, se = .11, p < .05), which shows that presidents tend to fare better as they leave office without the possibility of being reelected – 14 points better, holding all else at its 2006 values. The other significant control variable is the one accounting for non-polarization a However,

if the president is low on political capital, added polarization can hurt his prospects for signing preferred proposals into law. For absent a strong president, polarization does promote gridlock (Brady and Volden (1998), Krehbiel (1998)).

16

+(/010%2%.3!/4!.?)!+()5%,)6.=5!+/5%.%/6!S%66%6&!

?

"HB BH# +()5%,)6.5=! +/2%.%-12!! 81@%.12!

BHL BHK BHF

O/P Q),%'M! R%&?

BHD BHJ BHI BH* BH" BHB E*!5H,H

E"!5H,H

M)16 +/21(%>1.%/6 ?

N"!5H,H

N*!5H,H!

! $%&'()!*! Figure 4: Predicted Probability of Presidential Success on Contested “Key” Senate Votes +(),%-.),!+(/010%2%.3!/4!+()5%,)6.%12!7'--)55!/6!8/6.)5.),! " by Senators’ Polarization and the President’s Political Capital. Figure generated holding all 9:)3;!7)61.)!1.%/6!16,!.?)!+()5%,)6.=5!+/2%.%-12!81@%.12 ! ! else at their values in the 109th Congress (2005-2006) "

!$%&'()!&)6)(1.),!?/2,%6&!122!)25)!1.!.?)%(!A12')5!%6!.?)!"B#.?!8/6&()55!C*BBDE*BBFGH!

!

17

!

changes occurring in Washington over the last 50 years (measured as a natural log of the Congress number). Results for that variable show senators have been less willing to support the president on key, contested roll call votes, all else equal (b = −2.66, se = 1.27, p < .05). Coefficients on all other control variables are all in the expected direction, but none reaches conventional thresholds of statistical significance. In sum, the multiple regression results corroborate the basic model and its principal hypothesis: polarization around that pivotal voter’s position provides presidents with a better opportunity to win important, contested roll-call votes. This is especially true if the president is backed by high public approval and buoyed by a strong economy. By contrast, a president confronting a far-off pivotal voter surrounded by like-minded colleagues has few options for achieving legislative success, regardless of his political muscle.

3

Discussion

Washington politics have transformed in the postwar period, as Keith Poole and Howard Rosenthal have noted: “Beginning in the mid-1970s, congressional politics became much more divisive. More Democrats staked out consistently liberal positions, and more Republicans supported wholly conservative ones” (2007, 104). Part and parcel of strong partisans’ ascendance has been ideological moderates’ demise. The result is a Capitol characterized by a few centrists caught between two opposing camps of resolute ideologues - a pattern labeled polarization and sure to persist. In light of the Capitol’s increasingly polarized character, the scholarly task is clear: understanding the implications - both for the legislative process and the outcomes that result. Here 18

we aimed to build on previous research that finds congressional polarization often, but not always, produces legislative gridlock. Specifically, our theory helps specify the conditions when polarization will lead to gridlock and when it will not. So even as we find polarization does make coalition-building more difficult when the president lacks political capital or chooses not to use it promoting legislation, we also uncover a somewhat unintuitive prediction: polarization around the pivotal voter can actually provide presidents a unique opportunity not just to pass legislation, but to pass legislation closer to their ideal than would be possible in chamber filled with ideological moderates. Indeed, by allowing presidents to focus their efforts on fewer members, we hypothesize polarization can improve presidents’ prospects for winning key votes, securing legislative success, and influencing national legislation. Using of all Senate CQ key votes from 1954 to 2006 allowed a first test of this account, and results corroborated all principal predictions. Most important, we found clear support for the proposition that polarization qua polarization boosts presidents’ chances for prevailing on important, contested roll-call votes, especially when enjoying high approval ratings and strong economic growth. Before concluding, it is important to recall that polarization and the gridlock it engenders are not necessarily scourges on American democracy. The United States’ founding fathers never intended federal lawmaking to be easy and, in fact, they erected a constitutional design – including bicameralism and vetoes, staggered terms and separated constituencies – to ensure America’s public officials could not easily impose new laws on their constituents. As such, it is worth underscoring that “gridlock” is not necessarily symptomatic of a poorly functioning republic. Indeed, as George Will artfully expressed: “Gridlock is not an American problem, it is an American achievement” (Washington Post, Nov 4, 1999: A 35). 19

Yet if polarization and gridlock do not necessarily reveal a broken policymaking process, nor is government inaction always benign. Just as the founders did not want legislative coalition-building to be easy, nor did they want it to be impossible – not in addressing the nation’s pressing problems, not in answering citizens’ considered demands. It is in this regard that our study encourages a renewed focus on presidents’ policymaking actions. For in today’s political environment – where winning coalitions rarely emerge effortlessly – it places a premium on understanding what policies presidents support and, even more, what policies they are willing to invest resources promoting on Capitol Hill. As Harry Truman once explained, “The legislative job of the President is especially important I sometimes express it by saying the President is the only lobbyist that [most] Americans have” (10/25/56).

20

Appendix A. Uniform distribution:

Uni-modal Distribution

Bimodal Distribution

  1 xi ∈ [−0.5, 0.5] f (xi ) =  0 Everywhere else   4xi + 2 xi ∈ [−0.5, 0]    f (xi ) = −4xi + 2 xi ∈ [0, 0.5]     0 Everywhere else    8xi + 4       −8xi    f (xi ) = 8xi      −8xi + 4       0

21

(7)

(8)

xi ∈ [−0.5, −0.25] xi ∈ [−0.25, 0] xi ∈ [0, 0.25] xi ∈ [0.25, 0.5] Everywhere else

(9)

References John H. Aldrich and David W. Rohde. The consequences of party organization in the house: The role of the majority and minority parties in conditional party government. In Jon R. Bond and Richard. Fleisher, editors, Polarized Politics: Congress and The President in a Partisan Era, number 31-72. CQ Press, Washington, DC, 2000. John H. Aldrich and David W. Rohde. The logic of conditional party government: Revisiting the electoral connection. In Lawrence C. Dodd and Bruce I. Oppenheimer, editors, Congress Reconsidered, pages 269–292. CQ Press, Washington, DC, 2001. Matthew N. Beckmann. President’s playbook: White house strategies for lobbying congress. The Journal of Politics, 70(2), 2008. Matthew N. Beckmann and Anthony McGann. Navigating the legislative divide: Polarization, presidents, and policymaking in the us. Journal of Theoretical Politics, 20(2), 2008. Sarah A. Binder. Stalemate: Causes and Consequences of Legislative Gridlock. Brookings Institution Press. Washington, DC, 2003. Sarah A. Binder. The dynamics of legislative gridlock. American Political Science Review, 93(3):519–33, 1999. Jon R. Bond and Richard. Fleisher. The President in the Legislative Arena. University of Chicago Press, Chicago., 1990. David W. Brady and Hahrie Han. A delayed return to historical norms: Congressionl party

22

polarization after the second world war. British Journal of Political Science, 37:505–531, 2007. David W. Brady and Craig. Volden. Revolving Gridlock: Politics and Policy from Carter to Clinton. Westview. Boulder, CO., 1998. Jamie L. Carson, Michael H. Crespin, Charles J. Finocchiaro, and David W. Rohde. Redistricting and party polarization in the us house of representatives. American Politics Research, 35(6):878–904, 2007. John J. Coleman. Unified government, divided government, and party responsiveness. American Political Science Review, 93(4):821–35, 1999. Gary W. Cox and Mathew D. McCubbins. Setting the Agenda: Responsible Party Government in the U.S. House of Representatives. Cambridge University Press., 2005. Gary W. Cox and Mathew D. McCubbins. Legislative Leviathan: Party Government in the House. University of California Press, Berkeley, 1993. George C. III. Edwards. At the Margins: Presidential Leadership of Congress. Yale University Press. Yale University Press, New Haven., 1989. George C. Edwards III, Andrew. Barrett, and Peake Jefferey. The legislative impact of divided government. American Journal of Political Science, 41(2):545–63, 1997. Morris P. Fiorina. Retrospective Voting in American National Elections. Yale University Press., New Haven, CT, 1981. Richard. Fleisher and Jon R. Bond. The shrinking middle in the us congress. British Journal of Political Science, 34(3):429–452, 2004. 23

John B. Gilmour. Strategic Disagreement: Stalemate in American Politics. University of Pittsburgh Press, Pittsburgh, PA, 1995. Timothy Groseclose and Nolan McCarty. The politics of blame: Bargaining before an audience. American Journal of Political Science, 45(1):100–19, 2001. E. Howell, William, Scott Adler, Charles Cameron, and Charles Reimann. Measuring the institutional performance of congress in the post-war era: Surges and slumps in the production of legislation 1945-1994. Legislative Studies Quarterly, 25(285-312), 2000. Gary C. Jacobson. Party polarization in national politics: The electoral connection. In Jon R. Bond and Richard Fleisher, editors, Polarized Politics: Congress and the President in a Partisan Era, number 9-30. CQ Press, Washington, DC, 2000. R. Jones, David. Party polarization and legislative gridlock. Political Research Quarterly, 54(1):125–141, 2001. Sean Q. Kelly. Divided we govern: A reassessment. Polity, 25(3):475–84, 1993. Gerald H. Kramer. Short-term fluctuations in u.s. voting behavior. American Political Science Review, 65(1):131–43, 1971. Keith. Krehbiel. Pivotal Politics. University of Chicago Press, Chicago., 1998. David R. Mayhew. Divided We Govern: Party Control, Lawmaking, and Investigations, 1946-1990. Yale University Press., New Haven, 1991. Nolan McCarty, Keith T. Poole, and Howard Rosenthal. Polarized America: The Dance of Ideology and Unequal Riches. MIT Press, 2006. 24

Richard E. Neustadt. Presidential Power: the Politics of Leadership From FDR to Carter. John Wiley & Sons, Inc, New York., 1960. Keith T. Poole and Howard Rosenthal. Congress: A Political-Economic History of Roll Call Voting. Oxford University Press, New York, 1997. Keith T. Poole and Howard Rosenthal. The polarization of american politics. The Journal of Politics, 46(4):1061–1079, 1984. David W. Rohde. Parties and Leaders in the Postreform House. University of Chicago Press, Chicago., 1991. Steven A. Shull and James M. Vanderleeuw. What do key votes measure? Legislative Studies Quarterly, 12(4):573–82, 1987. Barbara. Sinclair. Party Wars: Polarization and the Politics of National Policymaking. University of Oklahoma Press. Norman, OK, 2006. Jeffrey M. Stonecash, Mark D. Brewer, and Mariani. Mack D. Diverging Parties: Social Change, Realignment, and Political Polarization. Westview. Boulder, CO., 2003. Sean M. Theriault. The case of the vanishing moderates: Party polarization in the modern congress. Party Politics, 12(4):483–503, 2004. George F. Will. One-house town, Nov 4, 1999. John R. Zaller. Monica lewinsky’s contribution to political science. PS: Political Science and Politics, 31(2):182–189, 1998.

25