Party conflict is pervasive in the House.” – Stonecash

““(We) used to be able to do more together on a bipartisan basis than seems possible these days. I’m not sure exactly why.” – Vice President Dick Cheney

The puzzle of partisan polarization is one of the more interesting and mysterious developments of the late Twentieth and early Twenty-First centuries. How is it that in an increasingly independent, split-ticket voting, valence and symbolic politics-oriented environment where the days of the party machine are long past (as a consequence of a series of Progressive era reforms), the behavior of legislators in Congress has become increasingly polarized and partisan? As Stonecash et al. note, “Members of each party are more likely to join together and vote against the other party. The parties increasingly adopt sharply different policy positions” (Stonecash, Brewer, and Mariani 2003). This argument is consistent with the trends in DW-Nominate scores of Congress over the past 40 years (Poole and Howard 1997; Poole and Rosenthal 2001). In the House, the 105^th Congress was sharply divided in comparison to House nominate scores two decades previous. Contrary to Darcy’s earlier finding that congressional parties were no more constrained than the public (Darcy 1980), the parties have become more ideologically coherent and ‘constrained,’ in Conversian terms, and further distant from one another (Ono 2005). This trend is apparent in Figure 12.1, tracking the change in the average 1^st dimension D-W Nominate score for congressional Republicans and Democrats in each chamber over the last 50 years. Jacobson finds a growing disparity in congressional support for presidential initiatives, arguing that the degree to which the president and members of the opposing party in Congress share constituencies has declined (Jacobson 2002). The natural consequence of this is a decline in bi-partisan legislative compromise, the heightening of political conflict over legislative action, and increased combativeness with a president from the opposition party. Jones suggests that party polarization is a more significant factor in policy agenda stalemates (gridlock) than divided government (Jones 2001). Ono argues that the two party caucuses in Congress have become more ideologically unified, as evidenced in the increase in party unity scores, and more divergent from one another (Ono 2005).

One explanation for partisan polarization in Congress is that the political party apparatuses have been “hijacked” by amateur, ideologically extreme activists (Fiorina, Abrams, and Pope 2004). The ‘storming of the gates’ to the Republican party nominating conventions by religious conservatives is a significant factor in the partisan polarization identified by Layman (Layman 2001). Others have attributed it to the increasing influence of party organizations on candidate selection (Fleisher and Bond 2004; Fleisher and Bond 2000). Stonecash and his colleagues argue that partisan polarization in the Congress is a consequence of a long-term secular realignment and social change. Constituencies in both parties have become more homogenous and more distinct from one another. “Realignment has brought the Democrats an electoral base that is less affluent, urban, and non-white. Republicans have acquired an electoral base that is more affluent, suburban and rural, and primarily white” (Stonecash, Brewer, and Mariani 2003). Stonecash makes a district-based argument, positing that the districts from which congresspersons are elected and reelected have become more homogenous and thus contributing to the realignment, partisan polarization in the House, and the differing issue positions the parties have taken up in this

Figure 12.1: Average Partisan House & Senate 1^st Dimension DW-Nominate Scores 1954-2004

newly aligned era (Stonecash 2005; Stonecash, Brewer, and Mariani 2003; Stonecash et al. 2000; Stonecash 2000). However, as is apparent in Figure 12.1, this trend has also occurred in the Senate (Poole and Rosenthal 1984), suggesting the effect of partisan polarization may not be dependent on changes in the district but may be a function of factors operating at a higher level of aggregation (Gelman et al. 2005). Ono’s argument that the growing incumbent advantage and uncompetitive districts has little to with redistricting but rather has an origin in factors that have operated on a national basis (party affiliation) is consistent with this view (Ono 2005).

On a contrarian note, while Brady and Han recognize that party elites have clear ideological differences and that parties have become more cohesive, they argue there is more bipartisan action in Congress now than in the past, perhaps due to the ambiguous ‘polarization’ of the masses. They argue “the difficulty of interpreting polarization in the electorate ultimately constrains elites” (Brady and Han 2006). Brady and Han define bipartisan unity as the “percentage of partisans who vote with the majority of their party when it is not a party vote” (Brady and Han 2006). So, when majorities in both parties pass legislation together, this is an act of bipartisanship. They conclude that “today’s level of bipartisan unity is close to as high as it has ever been.” Campbell and Cannon argue this inflates ‘bipartisanship’ as most of these votes are “noncontroversial,” such as reprimanding Iran for threatening behavior and designating the birthplace of President Clinton a National Historic Site. Campbell, in response, makes an argument in every way the opposite of that of Fiorina’s and, to a lesser extent, Brady and Han. While most scholars have reached a consensus that partisan polarization has occurred while opinion polarization, to the extent it exists, is much murkier, Campbell argues just the contrary. He asserts that polarization “may not always be well represented by the parties” (Campbell and Cannon 2006).

Mass partisan polarization is a consequence of the degree to which segments of the electorate, elites, or sub groups within the electorate have increasingly identified with one party rather than the other. This increasing identification may be recursively related to partisan politics. The groups themselves become more political and partisan in their outlook, and the politics of the parties and the elites as a consequence cater more to the interests of these increasingly loyal constituents. Layman and Carsey argue that the traditional theory of conflict displacement (as one dimension of partisan conflict emerges, the previous dimension must decline in salience and relevance) is inaccurate. Their “conflict extension” model suggests that multiple dimensions of issue conflict can coexist (what could be termed ‘ideological realignment’ as I will argue later). As such one need not argue that economic class is no longer relevant to suggest that partisans have aligned on one or more other issue dimensions (Layman 1999; Layman and Carsey 2002, 2003). Layman in particular argues that the social issues, or “Culture War” dimension has become more salient and that partisan conflict at the mass and elite level has become increasingly structured along this dimension (Layman 1996, 1997, 1998, 1998, 1999, 2001; Layman and Carmines 1997). Bolce and De Maio concur, arguing antipathy towards Christian fundamentalists has been on the rise in a society that increasingly divides itself along the lines that Hunter first theorized (Bolce and De Maio 1998, 1999, 1999). Resolving some methodological issues in previous studies, Brooks and Manza provide evidence of limited but significant changes in group-specific voting coupled with much larger changes in religion-based partisanship and party coalitions (Brooks and Manza 2004; Brooks 2002).

What direction does the arrow point in the mass / elite relationship? The consensus in the literature is that the elites wear the pants in the family (Dalton 1987; Herrera 1992; Hetherington 2001; Layman 1996; Miller 1986; Aldrich 1995; Poole and Howard 1997; Collie and Mason 2000; Alverez 1997; Alverez and Nagler 1995; McClosky, Hoffmann, and O'Hara 1960). Elites signal to a largely uninformed and unconstrained public what the relevant and salient issues of the day are and structure the political conflict around those issues. In response, the attentive public shifts its positions to conform with the elites they identify with (Arnold 1990). As noted earlier, Darcy argues that consensus among congressional elites and the mass public is similar, though the issues on which consensus exists differs. And while the congressional elites exhibited greater internal constraint, when external constraints were taken into account, this difference washed away given that the congressional elites were drawn from a narrower band of society (Darcy 1980). Though more recent trends in congressional polarization suggest otherwise (Poole and Rosenthal 1984, 2001). Zaller’s seminal work on public opinion addresses the relationship between mass and elite behavior, examining the process by which individuals express political opinions and the relationship between those opinions at the mass and elite levels. Zaller explicitly rejects Converse’s model of a ‘constrained’ public with a structured belief system, arguing instead that the expressed opinions found in public opinion surveys reflect the receptiveness of those individuals to the elite ‘message’ on the issues. Attentive publics are exposed to a great deal of information, but filter that information through their prior commitments and beliefs (i.e. conservatives pay attention to Rush Limbaugh and Sean Hannity while ignoring and discounting Keith Olberman and Jon Stewart). Thus politically aware individuals exhibit greater knowledge but stable preferences, while the masses are the weather vane of politics: exposed to few political messages but more inclined to accept the ones they receive or are ‘primed’ to accept at that particular moment in time (Zaller 1992). Other research has looked to the nature of issues themselves, identifying ‘easy’ and ‘hard’ issues with the easier issues more likely to be employed by elites to move the mass public in their direction (Conover, Gray, and Coombs 1982).

Fiorina expressly argues that elites have taken up increasingly polarized positions as a result of interest group and activist pressure while the mass electorate has increasingly moved toward a consensus on moderation and tolerance on these social and cultural issues. He goes so far as describing this elite polarization coupled with mass moderation as the “Hijacking of American Democracy” (Fiorina, Abrams, and Pope 2004). This view, however, is incongruent with two major arguments on the logic of partisan behavior and the relationship between elites and the masses in terms of shaping political opinion. Zaller argues that the adoption of opposing issues by political elites can have a “polarization” effect where the masses follow the cues of their party’s elite and adopt the position of their party (truest among the most attentive individuals in the electorate). However, if public sentiment strongly moves towards a consensus position, as Fiorina argues is the case with many of the culture issues in the culture wars, Zaller asserts that political elites of both parties will adopt that position and bring the public towards that position, thus producing a “mainstream effect” (Zaller 1992). Elite position-taking in this circumstance should reflect no partisan division and produce greater consensus within the electorate not, as Fiorina suggests, a greater consensus occurring as parties adopt increasingly divergent positions on the issue.

One of the models of mass politics assumes that elites drive and determine the nature of polarization. Polarization scholars have concurred, either siding with the belief that elites are pulling the electorate (or that portion of the electorate responsive to politics) along for the ride, or with the conclusion that elite politics occurs independent of the mass electorate. Is mass polarization the result or the cause of polarization among the elites? Is it that public opinion has become polarized, forcing the parties to move with the public in order to maintain electoral support? Or has the electorate become more coherently aligned along partisan dimensions because of a polarization of the parties themselves? Or both?

Some scholars have strongly argued that polarization in the electorate is the direct result of polarization among elites (Carmines and Layman 1997; Carmines and Gopoian 1981; Adams 1997; Abramowitz and Saunders 1998; Carmines and Woods 2002; Layman and Carsey 2002; McCloskey 1960; Hetherington 2001). Whether it is a matter of some issues falling out of favor while new issues gain salience or an “expansion” of conflict across a host of issues, these scholars has argued that elites lead and masses follow. Certainly there is strong evidence of partisan polarization in Congress (see Figure 12.2). The ideological distance between the Democrat and Republican parties has increased substantially over the last 50 years, as measured by the first dimension D-W Nominate scores. Layman and Carmine argue that when Democrats and Republican elites are polarized on an issue, and party identifiers become cognizant of those differences, then a significant portion of those individuals respond by adjusting their party ties to conform to their issue positions (conversion) while others respond by adjusting their issue positions to conform with the party identification (adaptation). Thus the mass public responds to the polarization at the elite level (Layman and Carsey 2002). Hetherington finds that greater partisan polarization in Congress has clarified the parties’ ideological positions as perceived by the American public, and thus has increased party importance and salience on the mass level (Hetherington 2001). For Fiorina, elite polarization occurs in spite of a largely centrist and

moderate mass electorate and thus is a significant breakdown in the institution of electoral democracy in America. Elites have certainly polarized (see Figure 12.2). Furthermore, they have done so independent of gerrymandering. The polarization trend in Figure 12.2 is apparent for both the House (subject to gerrymandering) and the Senate (not subject to gerrymandering).

For Fiorina, this elite polarization occurs independent of the masses, hence the mass electorate is largely unresponsive to shifts in the alignment of elites (Fiorina and Levendusky 2006). The roots of this orientation of the polarization literature can be found in the decades of work on mass public opinion. Converse first posited an uninformed and unstable electorate–essentially unequipped to participate in elections as traditionally conceived—and a variety of scholars have provided arguments to that effect since (Converse 1964; Blau 1977). This aspect of mass behavior provides the foundation for elite-driven mass politics and, consequently, polarization. This stands in contrast to the literature purporting to demonstrate that mass electorates are responsive and rational (Page and Shapiro 1992; Alverez 1997; Bartels 1986; Conover and Feldman 1989; Gerber and Green 1998; Popkin 1994). Whether it is a matter of responsive electorates, rational publics, or low-information rational voters, polarization in spite of the disposition of the voting public is inherently problematic. In any electoral environment where voters–however imperfectly or incompletely–perceive exogenous shocks, economic and political indicators, and the behavior of elected officials, polarization without the sanction of constituents spells political suicide to elites so inclined.

Furthermore, the most significant study of elites can be found in the decades of literature on the U.S. Congress where hyper-responsiveness to constituent concerns in service, position-taking, political advertisements, etc. has been the norm rather than the exception (Jacobson 2000, 2002; Hill and Hurley 1979; Collie and Mason 2000; Mayhew 1974). Without a responsive electorate, there is no electoral connection that elites are obligated to respect, and hence the distributive rationale for congressional action would be non-operative.

Fiorina’s elite polarization story (see Figure 12.3) essentially doesn’t make sense in a world where electorates are rational and responsive, as a unilateral shift to the poles of the electoral distribution would lead to Downsian party marginalization. Rationalizations offered to explain elite non-responsiveness, such as Fiorina’s theory that elites anticipate ideological challenges from the poles in primaries, fundamentally detach the elected official from his or her interest in re-election (as opposed to standing for election, or receiving interest group benefits, or rewarding partisans, etc.). The argument does not merely suffer from theoretical problems, but also an unresponsive electorate is incongruent with apparent aggregate rationality in terms of short-term effects on elections, increasing homogenization of districts tracking with increased elite polarization, and the effectiveness of political advertising campaigns. The evidence suggests that voters are paying attention to some signals from the political environment and that they hold officials accountable in elections. And while district polarization

Figure 12.3: Fiorina’s Elite Polarization with Centrist Electorate*

T: Functioning Electoral System with Parties Located Near Voters

T+1: Increasing Polarization of Partisan Elites Away from Voters

T+2: “Hijacked” Political System with Elites Polarized from Voters

*Points on line reflect the hypothetical mean position of the voters and partisan elites

I argue that elite polarization is recursively related to mass polarization (see Figure 12.4). It is true that elites attempt to influence the mass public through the framing and emphasis of certain issues, but it is just as true that elites strive to detect those issues which have become salient to the electorate and shape their position-taking accordingly. Scholars examining the factors of polarization in American politics miss half of the story when ignoring the influence that shifts in the mass electorate have on elite behavior. Electorates change, they change in comprehensible ways, the issues and opinions of aggregate electorates shift sometimes independent of elite preferences, and these shifts produce shifts in electoral behavior. As a result, the optimum party platform in any given election changes as well. Why don’t elites merely determine the optimum political positions on the relevant set of issues and adopt those as their party platform? There are numerous factors that prevent this simple strategy. Elites operate in a world of uncertainty. They are uncertain as to what issues have become politically salient among the public. They are uncertain as to what portion of the electorate will turnout and in what proportions, hence making the nature of the responsive public unclear. They are uncertain as to what exogenous shocks will influence the public and what impact those will have on public opinion (Alverez and Franklin 1994; Bartels 1986; Calvert 1985; Enelow and Hinich 1981; Ferejohn and Noll 1978; Shepsle 1972; Wright and Goldberg 1985; Hinich and Munger 1994).

Given this, elites attempt to gain and influence information. They make efforts to gather signals from the public as to what issues they care about through polls and electoral results…hence reducing uncertainty as to the electoral calculus. However, even if they manage to determine what the optimum position is, they may not be in a position to adopt it. Position-taking does not occur in a vacuum. The positions that parties and officials can take are bounded by the positions they have taken in the past,

Figure 12.4: Rational Polarization Given Shifts in Constituencies

Type 1: Partisan Voting Constituencies Signal Divergence on Salient Issues*

DemV_T+1 DemV_T RepV_T RepV_T+1

Type 1: Elites Diverge in Response to Voter Polarization

D-Elite_T+2 D-Elite_T+1 R-Elite_T+1 R-Elite_T+2

Type 2: Elites Signal Divergence on Salient Issues

D-Elite_T+1 D-Elite_T R-Elite_T R-Elite_T+1

DemV_T+2 DemV_T+1 RepV_T+1 RepV_T+2

*Points on line reflect the mean position of the voters and partisan elites

and hence reputations are relatively sticky. As such, elites attempt to influence publics through party platforms, campaign messages, statements, press conferences, etc…in an effort to guide public opinion to regard as salient those issues and policies where the party has strength and on which the electorate could be decisive (Alverez 1997; Calvert 1986; Dalton, Beck, and Huckfeldt 1998; DeSart 1995; Feddersen and Pesendorfer 1997, 1999; Huckfeldt and Sprague 1987; Huckfeldt and Sprague 1988; McKelvey and Ordeshook 1985, 1986; Morton 1993; Palfrey and Poole 1987). Emergent issue cleavages are, as Petrocik and others have argued, thus a function of changes in the mass electorate that yield shifts in the party platforms and messages of the two parties and a function of the elites within parties shifting their message so as to expand their base or to take advantage of a perceived avenue of electoral opportunity. Where one party successfully produces a credible platform or policy output that appeals to a sufficiently large portion of the electorate (or conversely, where one party alienates a sufficiently large portion of their political base) then the conditions are set for realignment.

Polarization is a phenomenon where some combination of the above factors serves to push the parties, groups, elites and/or the aggregate electorate apart. These centrifugal forces pull voters towards the poles of political opinion and polarize their beliefs, opinions, and actions relative to other groups in the political space. The predicates of polarization and the consequences of polarization are:

1. 
   Mass Ideological Polarization: where the distribution of the mass electorate along the ideological dimension shifts away from the center and towards one of or both of the poles. This is reflected in greater dispersion of opinion across the ideological dimension.
   

1. 
   Aggregate shifts from the center to one or both of the poles on issue, policy, or ideological dimensions.
   
2. 
   Within-group (or party) homogenization along issue, policy, or ideological dimensions.
   
3. 
   An increase in the distance between the in-group and out-group on issue, policy, or ideological dimensions.
   

2. 
   Elite Ideological Polarization: where the distribution of the elites (politicians, activists, etc.) on the ideological dimension shifts away from the center and towards one of or both of the poles. This is reflected in greater dispersion of opinion across the ideological dimension.
   

1. 
   Aggregate shifts from the center to one or both of the poles on issue, policy, or ideological dimensions.
   
2. 
   Within-group (or party) homogenization along issue, policy, or ideological dimensions.
   
3. 
   An increase in the distance between the in-group and out-group on issue, policy, or ideological dimensions.
   

3. 
   Electorate-Driven Polarization: where partisan electorates become more homogenized along issue or policy dimensions, reducing the uncertainty parties have as to the policy preferences of their respective electorates and permitting them to construct clear responses in terms of positions and policies that the more homogenized electorate can receive. Aspects:
   
   1. 
      Become more polarized along current policy or ideological dimensions
      
   2. 
      Coalesce around new salient issues or policies on which they are more polarized
      

4. 
   Elite-Driven Polarization: where partisan elites become more homogenized along issue or policy dimensions and hence reduce the signal-to-noise ratio for electorate decisions. Aspects:
   
   1. 
      Have divergent beliefs as to the direction of a move along an issue, policy, or ideological dimension position that will yield an expansion of their partisan electoral base.
      
   2. 
      Have divergent beliefs as to which position to adopt on a newly salient issue that has entered the policy space.
      

5. 
   Party Image Polarization: polarization of or in response to the perceived positions of the political parties.
   
   1. 
      As the electorate or identifiers perceive a shift from the center to one or both of the poles on the issue, policy or ideological dimensions of the party elites, they respond accordingly: shifting party allegiance consistent with their issue positions or changing their positions to conform to their party (i.e. polarize).
      
   2. 
      As elites perceive a shift in how they are perceived by the public, they seek to capitalize on this change by shifting in the polar direction congruent with the public or the partisan electorate’s perception.
      

6. 
   Constituent Polarization: polarization of or in response to the constituents of the elected officials of the political parties.
   
   1. 
      As constituents perceive shifts from the center to one or both of the poles on the issue, policy or ideological dimensions of the party elites, they respond accordingly: shifting party allegiance consistent with their issue positions or changing their positions to conform to their party (i.e. polarize).
      
   2. 
      As elites perceive a shift in their constituent’s position towards one of the poles on issue, policy, or ideological dimensions, they seek to capitalize on this change by shifting in the polar direction congruent with their constituent preferences, or are replaced by someone who will.
      

H_o: No change or centripetal change in the distribution of ideology in the mass public over time.

H_o1: Constant or decreasing variance in the ideology of the mass public over time.

H_o2: Constant or centripetal change in mean ideology of the mass public over time.

H_a: Centrifugal change in the distribution of ideology over time.

H_a1: Increasing variance in the ideology of the mass public over time.

H_a2: Centrifugal change in mean ideology of the mass public over time.

Elite Ideological Polarization over Time

H_o: No change or centripetal change in the distribution of elite ideology over time.

H_o1: Constant or decreasing variance in elite ideology over time.

H_o2: Constant or centripetal change in mean elite ideology over time.

H_a: Centrifugal change in the distribution of ideology.

H_a1: Increasing variance in the ideology of the mass public over time.

H_a2: Centrifugal change in mean ideology of the mass public over time.

Mass Elite Ideological Polarization

H_o: No relationship between mass ideology and elite ideology.

H_a1: Simple: mass ideology determines elite ideology.

H_a2: ME Aftershock: Lagged mass ideology determines elite ideology.

H_o: No relationship between elite ideology and mass ideology.

H_a1: Simple: elite ideology determines mass ideology.

H_a2: EM Aftershock: Lagged elite ideology determines mass ideology.

H_o: No relationship between constituent ideology and elite ideology.

H_a: Squared differences between constituent ideology and elite ideology decline over time

Political Party Hypotheses

Inter-Partisan Identifier Ideological Polarization over Time

H_o: No change or a decrease in the average ideological difference between Republican identifiers and Democratic identifiers over time.

H_a: Increase in the average ideological difference between Republican identifiers and Democratic Identifiers over time.

Intra Partisan Identifier Ideological Polarization over Time

H_o: No change or centripetal change in the distribution of ideology among partisan identifiers over time.

H_o1: Constant or Increasing variance in the ideology of the partisan identifiers over time.

H_o2: Constant or centripetal change in mean ideology of partisan identifiers over time

H_a: Centrifugal change in the distribution of ideology among partisan identifiers over time.

H_a1: Decreasing variance in the ideology of partisan identifiers over time.

H_a2: Centrifugal change in mean ideology of partisan identifiers over time.