Madison’s Constitution Under Stress: A Developmental Analysis of Political Polarization

We present a “developmental” approach to understanding why rising polarization in the United States has not been self-correcting but instead contin-ues to intensify. Under specified conditions, initial increases in polarization may change the meso-environment, including such features as state parties, the structure of media, and the configuration of interest groups. These shifts can in turn influence other aspects of politics, leading to a further intensification of polarization.This analysis has four important benefits: ( a ) It directs our attention to the meso-institutional environment of the American polity; ( b ) it clarifies the features of the polity that have traditionally limited the extent and duration of polarization, and the reasons why their contemporary impact may be attenuated; ( c ) it helps us analyze asymmetrical, or party-specific, aspects of polarization; and ( d ) it provides an analytic foundation that connects discussions of American politics to the comparative politics literature on democratic backsliding.


INTRODUCTION
Much of modern political analysis has, explicitly or implicitly, taken an approach to understanding the American polity that emphasizes its tendencies toward moderation and stability. Some analyses emphasized the stabilizing impact of Madisonian institutions of fragmented and overlapping authority within a highly diverse society (Dahl 1961, Polsby 1997. Others drew on the seminal work of Anthony Downs-which Fiorina & Abrams (2009, p. vii) rightly described as a sort of "master theory" for a generation of Americanists. That framework suggested that American institutions of representation and electoral competition not only created powerful incentives for a two-party system but also induced the two parties to operate close to the middle of the distribution of preferences within the electorate or face electoral retribution. Both these approaches anticipated that efforts by major parties to stake out political projects unable to command broad support would face strong resistance and growing backlash. The center would hold.
The rise of durable polarization between the parties-a major preoccupation of Americanists for the past two decades-has raised obvious issues for these frameworks. Americans grumble about polarization, yet it persists. The absence of any correction has forced acknowledgment that we have shifted from a system marked by low polarization to one of high polarization. American politics scholars have worked, with some success, to advance understanding of the characteristics of this new high-polarization setting (e.g., Lee 2015, Hopkins 2018, McCarty 2019 as well as to understand the forces that moved the system from one arrangement to the other (McCarty et al. 2006, Schickler 2016.
In many respects, the literature on today's polarized system is actually reassuring. The year 2020 marks the twenty-fifth anniversary of Newt Gingrich's ascendance to the Speakership, so intense polarization is at least a quarter-century old. If our politics has been polarized for that long, one could argue that perhaps we have adapted. From this vantage point, a long period of polarization reaffirms the flexibility of our Madisonian framework. A number of observations bolster this position, ranging from the assertion that American politics has often been equally polarized in the past to the claim that partisan divisions in the United States are often more symbolic or tactical than substantive (Lee 2009(Lee , 2016.
This article outlines a different possibility. Rather than describing a simple shift from a nonpolarized system to a polarized but nonetheless stable one, we should consider the conditions under which polarization might feed on itself. Our concern is that with polarization come new developments in the polity that encourage further polarization. Moreover, the self-correcting mechanisms of the Madisonian polity so often celebrated in the past may have either weakened or themselves been transformed into engines of polarization.
Our starting point is to adopt a developmental approach to polarization. Rather than treating polarization as a static point on some continuum, we treat it as an ongoing developmental process. Pursuing a truly developmental approach requires grappling with the ways in which high levels of polarization might become not just self-reinforcing but susceptible to intensification. This in turn requires careful attention to the potential for institutional configurations to either dampen or exacerbate the process of polarization once it begins. 1 In past eras, what we term meso-institutions acted as countervailing mechanisms, fostering deep factional divisions in the parties that undermined polarization. Today's polarization, by contrast, has fostered the rise of new organizations and transformed existing ones, creating new relationships, balances of political power, and incentives. These changes, in turn, have intensified divisions between the parties, their supporting coalitions, and voters.
Our goals are to identify these new relationships and incentives and to explore how they might lead to qualitative changes in the character of politics. We argue that, in the contemporary party system, polarization has indeed become self-reinforcing. Instead of mirroring (and thus bolstering) the fragmentation that Madison famously argued was encouraged by federalism and separation of powers, parties now operate in precisely the opposite fashion. This change in party behavior is itself a reflection of very significant changes in civil society. Today, many different actors across institutions see their interests as dependent on the success of their party. Put more ominously, they see the costs of the other party's success as unacceptable. Interest groups, state and local parties, rank-and-file politicians, campaign donors, and media outlets that in the past had exercised-at least in part-a centrifugal influence on party politics increasingly contribute to a single, nationalized, polarized politics. Rather than introducing cross-cutting cleavages and a diversity of concerns, these actors behave in ways that reinforce or intensify partisan divides.
Our approach stands in contrast to the standard Madisonian story told about American politics: one of stability induced by the configuration of political institutions and reinforced by a broadly supportive political culture. If convincing, this developmental analysis of polarization has at least four important benefits: (a) It usefully directs our attention to the meso-institutional environment of the American polity; (b) it clarifies the self-correcting or countervailing features of the American polity that have traditionally limited the extent and duration of polarization, and the reasons why their contemporary impact may be attenuated; (c) it helps us analyze asymmetrical, or party-specific, aspects of polarization; and (d) it provides an analytic foundation that connects discussions of American politics to traditions in comparative politics that seek to assess a polity's vulnerability to democratic backsliding.
Our discussion is presented in four parts. First, we ask what historical evidence has already taught us about contemporary polarization. Second, we review the characteristics of our Madisonian system that were widely believed to make polarization difficult to produce and even harder to sustain. Third, we outline a developmental account of contemporary polarization, focusing on some of the major changes in what we call the meso-institutional environment (interest groups, state parties, media) that we believe were first encouraged by polarization and now serve to intensify it. 2 Fourth, we outline some of the significant implications of this account.

HISTORICAL PERSPECTIVES ON POLARIZATION
Our developmental framework could easily be mistaken for a simple call for a turn to political history. Briefly considering how historical evidence has informed recent discussions of contemporary polarization can clarify the distinctiveness of this analysis. Comparisons with earlier periods can offer a useful counterpoint to the tendency to see today's politics as without precedent. In our view, however, rather than providing reassuring evidence that polarization is the norm, the historical evidence more often highlights the distinctiveness of recent developments-particularly how polarization has fed on itself rather than dissipating in the face of the countervailing forces operating within much more decentralized political configurations.
A key focus of historical work on polarization has been to ask whether the party divisions observed today are unprecedented, or even unusual. Many such analyses focus on measures of polarization derived from congressional roll call voting, such as Poole and Rosenthal's Nominate methodology (Poole & Rosenthal 1997). These measures, for example, compare the difference in the estimated ideology of the median Democrat and median Republican in Congress, or consider the percentage of partisans who are closer to the other party's center than to their own party. Scholars working in this tradition have observed that "Congress is now more polarized than at any time since the end of Reconstruction" (Hare et al. 2014; see also McCarty et al. 2006, pp. 23-29;Schaffner 2011;McAdam & Kloos 2014). This understanding has become a working assumption underlying a range of studies analyzing the impact of polarization, such as Levitsky & Ziblatt's How Democracies Die (2018, p. 204), which notes that contemporary "polarization, deeper than at any time since the end of Reconstruction, has triggered the epidemic of norm breaking that now challenges our democracy" (see also Mason 2018, p. 137).
Yet even as Nominate scores suggest that today's polarization is high by historical standards, a second lesson often drawn from these measures is that polarization is the normal state of affairs in American politics. For example, Aldrich et al. (2002, p. 20) characterize the 1930s-1970s "as a lengthy, singular exception to what had typically been essentially constant before World War II.. . . It thus appears that a Congress divided more or less sharply, but almost invariably divided, by parties is the historical norm." Brady & Han (2007, p. 506) concur, noting that "the recent period of polarization mirrors patterns of polarization that have prevailed throughout most of congressional history. In fact, the truly unusual historical period is the bipartisan era immediately following the Second World War" (see also Hetherington 2009;Schaffner 2011;Disalvo 2012, p. 185;McAdam & Kloos 2014, ch. 2;Tucker et al. 2018). Although skeptical of Nominate-based measures, Lee (2015, p. 262) suggests that other indicators-such as party platforms-also suggest that polarization has been common in American history; intense party conflict "may well be the normal state of affairs, and the long postwar period of muted party conflict may constitute a mere exception" (see also Karol 2015).
Yet historical time series tracking polarization levels only take us so far. Nominate scores reveal how differently Democrats and Republicans vote in Congress but not whether the parties are fighting about big or small policy issues (Lee 2009(Lee , 2016. They offer a snapshot of behavioral differences at a given point in time, but much more needs to be done to understand the developmental aspects of polarization. Moving beyond such time series indicators, scholars have considered whether specific historical periods offer a useful analogy for today's politics. The most common analogies have been to the lead-up to the Civil War and the turn of the twentieth century. When the Civil War is the point of comparison, most observers conclude that contemporary divisions are far less serious. For example, Hetherington (2009, p. 417) writes that "the substance and intensity of the conflicts today are certainly nothing like those in the years leading up to the Civil War, which ranged from the unpleasant to the dangerous" (see also Brady & Han 2006;McAdam & Kloos 2014, p. 253). The historian Joanne Freeman (2018, p. xiii), however, notes that while "there are worlds of difference between the pre-Civil War Congress and the Congress of today. . .the similarities have much to tell us about the many ways in which the People's Branch can help or hurt the nation." More specifically, Freeman's (2018, p. 229) analysis reveals the dangers posed when party divisions are overlaid on an intense sectional cleavage. Under such circumstances, compromise can be seen as a sacrifice of honor, and the logic of conflict can escalate far beyond previously accepted bounds of political contestation (Freeman 2018, p. 248).
Given that contemporary polarization has not generated anything resembling an actual civil war, the period of high party voting and centralized leadership in Congress spanning the late nineteenth and early twentieth century is more often seen as the best analog for today. Azari & Hetherington (2016, p. 92) note that the politics of the post-Reconstruction era are "strikingly similar to what we have in the present day. Elections were consistently close; race, culture, immigration, and populism were salient issues; and states almost always voted for the same party in election after election." Furthermore, congressional polarization "reaches its peak in the late Gilded Age and the contemporary period as well" (p. 93). Azari & Hetherington (2016) suggest that the close party balance nationally and the salience of racial concerns were critical sources of polarization both in the late nineteenth century and today (see also Disalvo 2012, Karol 2015, Hopkins 2017, Hall 2019 on the analogy between these two periods).
While specific periods, such as the 1850s or the 1890s-1900s, provide valuable lessons about what polarized politics look like in practice, any given analogy breaks down when pushed too far. The Civil War era represents an obvious extreme point in the intensity of divisions, yet the period of partisan polarization was remarkably brief: The major American parties featured deep internal divisions on slavery up until the mid-to-late 1850s, and the new Republican majority became deeply divided over Reconstruction and key economic questions soon after the war ended. By the 1870s, the within-party divisions among Republicans on Reconstruction and key economic questions rivaled the interparty cleavage in importance. The 1890-1910 analogy also has evident limitations. These years of supposed peak polarization actually featured serious regionally based intraparty divisions over economic development and regulatory policies (Sundquist 1983, Bensel 2001. These divisions became unmanageable with the rise of progressive insurgency after 1906 and the Taft-Roosevelt split in 1910-1912(Mayhew 2002. Intense partisan polarization, such as that existing just before and after the Civil War and at the turn of the twentieth century, proved short-lived and deeply vulnerable to regionally based factional interests. This vulnerability of previous episodes of polarization is what has led some of the most astute observers of our politics to conclude that polarization will ultimately falter. After highlighting the intensity of today's divisions, Azari & Hetherington (2016, p. 108) note that "the fate of previous eras of division suggests that this brand of politics is rarely sustainable in the long term. If not in 2016, it seems change is likely to come soon." We are far less confident that polarization is likely to recede, or even stabilize. Our main point, however, is that any such claim cannot rest on the observation that "this is what has happened before." Instead, it requires a careful identification and assessment of the mechanisms that either attenuate or intensify polarization. This permits consideration of the prospect that within contemporary party politics the institutional configurations that previously acted as countervailing mechanisms have been displaced by alternative structures that instead reinforce or even intensify polarization. We can sharpen this point before considering the specific changes we have in mind by briefly discussing why so many have seen our Madisonian institutions as a strong protection against intense polarization.

THE MADISONIAN SYSTEM: A SHIELD AGAINST POLARIZATION?
The dynamics of contemporary polarization raise important questions about the standard, Madisonian account of American politics, which emphasizes the ways in which core political institutions encourage compromise and stability. The Founders were, of course, preoccupied with the question of how to create a stable republic. Understanding that factional divisions are inevitable, Madison famously argued that American political institutions could prevent all-out conflict between competing camps. Critical mechanisms that would tend to attenuate or countervail against polarization, rather than reinforce it, were built directly into the constitutional system. Others, such as the development of what were by comparative standards highly decentralized and geographically factionalized political parties, were crucial (if unintended) outgrowths of the constitutional framework.
Most obviously, separation of powers, checks and balances, and federalism divide power, making it less likely that any single group will gain control of the entire government. This, in turn, means that governance will routinely require accommodating a range of group interests. The rules structuring elections also require building different kinds of coalitions for different offices, discouraging the emergence of a single coherent and dominant cleavage.
The creation of an extended republic with immense geographic diversity reinforced the institutional obstacles to polarization. Madison lays out the logic of this argument in Federalist 10: The scope of the new nation ensured a heterogeneity of viewpoints, which would make the emergence of a majority faction unlikely (Kernell 2003). While a small republic might be characterized by a single intense cleavage-over religion, wealth, or some other characteristic-an extended republic would tend to give rise to cross-cutting cleavages. Creating a majority would require broad appeals to widely shared interests, rather than narrow, parochial appeals to a particular faction. As Dahl & Lindblom (1953, p. 307) argue, social pluralism, when combined with America's fragmented constitutional structure, forces bargaining among diverse groups in order to achieve policy success (see also Truman 1951, p. 514;Gunnell 2004, p. 224).
Federalism, from this perspective, interacts with the extended republic in critical ways. It is not just that the national government shares power with 50 separate state governments. The diversity of state circumstances and the relative autonomy of state political institutions promote carefully brokered compromises that are mindful of an array of distinctive interests (Anderson 1955, pp. 135-36;Elazar 1966, pp. 6, 203;Truman 1966).
These core institutions of American government tend to frustrate efforts to consolidate the power of a particular individual or coalition; each puts a premium on finding ways to accommodate opposing interests. Under many conditions, we can think of them as functionally equivalent mechanisms for attenuating polarization. Even if one mechanism fails to operate in a particular context-such as when unified government reduces Congress's incentive to check the presidentthere are built-in redundancies that reinforce the overall tendency toward stability and moderation. Moreover, these institutions have a homeostatic quality. Given the independence and diversity of political settings and roles, politicians unwilling to engage in compromise are likely to trigger an increasingly powerful backlash.
The Framers, of course, did not anticipate several major developments that might have undermined the Madisonian formula. Most notably, political parties were not part of the Constitution, yet developed soon after the Founding. Even so, the constitutional system profoundly shaped these new parties in ways that made them unlikely vessels for intensely and durably polarized politics. From the start, American political institutions helped produce parties that were federal in character and decentralized in many of their operations (Disalvo 2012). In the words of V.O. Key (1964, p. 315), American parties were "confederative," consisting "of a working coalition of state and local parties" that provided pluralistic representation of diverse interests (see also Epstein 1982, Schattschneider 1942. A critical source of power and independence for state parties has been their control of nominations and, more generally, their role in shaping career paths for ambitious politicians. Truman (1966, p. 92) observes that "the basic political fact of federalism is that it creates separate, selfsustaining centers of power, privilege, and profit. . . [and] bases from which individuals may move to places of greater influence and prestige in and out of government." In addition to the formal leverage created by state parties' control of nominations, the need to compete for power across a wide span of very different states forced American parties to take the form of catch-all organizations that accommodated a range of ideologies and social groups.
While there also was an important national aspect to these party organizations-particularly rooted in competition for the presidency-they retained substantial autonomy throughout much of their history. Polsby (1997, p. 40), echoing an earlier comment from Dwight Eisenhower, argues that "one may be justified in referring to the American two-party system as masking something more like a hundred-party system." A Massachusetts Democrat and an Alabama Democrat might belong to the same formal organization at the national level, but they need not agree on much of anything when it comes to policy. Hershey (2017, p. 26), in her textbook on American parties, concludes that federalism and separation of powers mean that American legislative parties "can rarely achieve the degree of party discipline that is common in parliamentary systems." This feature of party politics has historically lowered the stakes of political conflict-an effect that comparativists have long stressed is conducive to democratic stability (Acemoglu & Robinson 2006). Even if one party wins power, it is forced to accommodate a diverse array of interests that likely will make its ultimate policies broadly acceptable. Furthermore, the cross-cutting cleavages and fluidity of alliances ensure that even if one's side loses today, the outcome could easily change soon (Latham 1952).
This account of the American political system always had critical blind spots. It tended to overlook the systematic biases in representation and inequalities in social and economic power that ensured that even if there were no permanent winners, there were plenty of permanent losers. Institutionalized white supremacy was the most glaring refutation of this pluralist faith, but not the only one. The Madisonian account also did not easily accommodate social movements that, in important moments, challenged the legitimacy and stability of the American political system. The political violence of southern "redeemers" during and after Reconstruction, the rise and suppression of labor militance in the late nineteenth century, and the civil rights-and later, black power and antiwar-movements of the 1960s each presupposed a kind of politics that is not comfortably categorized in terms of pluralist stability. And, of course, the Civil War constitutes a crucial moment when the Madisonian system catastrophically failed to contain conflict, as a single, overarching division broke the polity apart.
But for all of these failures, the Madisonian system was, for much of American history, a robust obstacle to the consolidation of power. The operations of the constitutional system might be remade on the ground over time by assertive presidents (see Skowronek 1997 on "Reconstructive" leadership) or new ideological formations (e.g., the New Deal), but the core features that gave rise to pluralism and fragmented power remained: separation of powers, checks and balances, territorially grounded representation, and the extended republic. The modern presidency might be a much more powerful office than Madison anticipated, yet modern presidents continued to be frustrated by the need to deal with contending power centers in Congress, the Courts, the bureaucracy, and the states (Neustadt 1960, Moe 1985, Skowronek 1997. The New Deal remade the role of the national government, yet also had to confront fundamental limitations imposed by separation of powers, federalism, and the Democrats' north-south regional coalition (Weir 2005).
In sum, while political parties might bridge the differences across branches, institutions, or localities in a way that the Framers had not anticipated, sustained, intense policy polarization at the national level has been rare. Even in periods of high party voting in Congress, there remained substantial intraparty divisions that limited the scope of partisan battles. A fragmented party and interest group system meant that national party lines failed to capture or contain many of the critical disputes animating politics-and these disputes countered the force of national party polarization.
American politics scholars are not the only ones to highlight the impact of these features on the stability of American democracy. The comparative literature on presidentialism also showed appreciation for the apparent robustness of the Madisonian system. In pathbreaking work, Linz (1990) argues that presidential systems tend to be less stable due to dueling bases of legitimacy. Viewing the United States as an apparent exception, Linz suggests that our weak and fragmented parties have prevented this kind of all-or-nothing showdown between branches under the control of competing parties. We argue below that this confidence in the moderating influence of American political institutions may no longer be justified.

THE DEVELOPMENT OF MODERN POLITICAL POLARIZATION
Standard accounts of polarization emphasize the sorting of the parties, at both the elite and mass levels, which flowed from realignment of the political parties around issues of race. This itself was a very long-term process (Schickler 2016), but in national politics the critical events surrounding the parties' repositioning on civil rights occurred in the 1960s and early 1970s. By the end of this period, national party leaders had clearly placed themselves on the conservative and liberal sides of racial issues. In doing so, they made the Republicans and Democrats, respectively, more clearly conservative and liberal parties. This new clarity in turn helped trigger the well-known processes of elite and mass sorting, and elite replacement, that analysts associate with modern partisan polarization (McCarty et al. 2006).
Although it has received less discussion in the analysis of polarization, a second development in the 1960s and early 1970s-what Skocpol (2003, p. 135) has termed the "long 1960s"-was also critical: a dramatic expansion and centralization of public policy (Melnick 1994, Jones et al. 2019. Civil rights legislation was only the entering wedge. During the long 1960s, liberal Congresses enacted, often on a bipartisan basis, major new domestic spending programs (especially Medicaid and Medicare, which now account for roughly a quarter of federal spending as well as, in the case of Medicaid, a big share of state spending). They greatly enlarged the regulatory state, creating powerful new federal agencies (such as the Environmental Protection Agency) and enacting extensive rules covering environmental and consumer protection as well as workplace safety. Finally, federal courts introduced or expanded a range of rights (most dramatically, reproductive rights enshrined in Roe v. Wade), which essentially nationalized policy making with respect to a host of controversial social issues. By the late 1970s, Washington had become a much more prominent force, across a much wider range of issues, than it had been two decades earlier.
The expanded role of Washington became a critical issue dividing the parties, contributing to polarization both directly and indirectly (Hopkins 2018). Directly, it reinforced the process of sorting between the consolidating liberal and conservative parties. Increasingly, the two parties diverged on fundamental questions regarding this emerging activist federal state. Indirectly, this new government activism encouraged the mobilization and nationalization of interest groups in response to the growing stakes in national-level political contestation (Leech et al. 2005). As we shall see, over time these two dynamics merged. These extensive and now nationalized groups faced growing incentives to align themselves with one party or the other.

THE TRANSFORMATION OF MESO-INSTITUTIONS
The initial development of polarization and of nationalized policy contestation had profound consequences for the American polity. In this necessarily abbreviated account, we wish to stress the changes it encouraged in what we are calling meso-institutions: interest groups, state parties, and the media. In earlier eras, these arrangements had been crucial bulwarks of the formal institutions of our Madisonian system, tending to attenuate partisan polarization. Today, they have changed in ways that instead encourage further national party polarization. As we briefly note, these changes help to account for an increasingly tribal mass politics marked by negative partisanship, which further intensifies polarization.

The Shifting Relationship Between Interest Groups and Parties
The expanding role of the national government led to a proliferation and nationalization of interest group activity. Over time, the new environment produced a second major shift: a growing inclination of powerful groups to draw closer to one or the other of the major parties. In an increasingly polarized environment, these groups faced incentives to pick a party-to try to achieve their policy goals by working with, and in support of, a durable political coalition. Rather than being a source of incentives and action that cross-cut parties and thus restricted polarization, interest groups became another factor reinforcing the divide between them.
Of course, some groups have always aligned closely with parties-organized labor's attachment to the Democrats offers a clear mid-twentieth-century example (Karol 2009, Schlozman 2015. Yet core understandings of politics emphasized the strong incentives for most groups to avoid close alignment with parties (see, e.g., Hansen 1991).
There is broad consensus today that the interest group environment has become more polarized and party-aligned (Krimmel 2017). The list of prominent groups that have moved into tighter alliance with a single political party as the nation has polarized is long. For Republicans it includes such influential organizations as the National Rifle Association (NRA) (Karol 2009), the Koch brothers network (Skocpol & Hertel-Fernandez 2016), the oil and gas industries, and major conservative Christian organizations (Schlozman 2015). The powerful US Chamber of Commerce provides a striking illustration of the broader trend. Traditionally conservative but studiously nonaligned, it now carefully coordinates its extensive electoral activities with the Republican Party, and its political director (a former GOP operative) can refer unselfconsciously to Republican Senate candidates as "our ticket" (Hacker & Pierson 2016).
The Democratic coalition has, since the New Deal, been aptly characterized as a coalition of distinct policy-demanding groups (Grossmann & Hopkins 2016). However, we contend that several key groups that had maintained substantial ties to both parties prior to the 1960s and 1970s are now more firmly tied to the Democratic Party. The list would include major civil rights, environmental (Karol 2019), and reproductive rights groups.
These growing attachments partly reflect the fact that as the parties polarized and offered clearer choices, more groups found one party's policy preferences to be a much better fit than the other's. A second factor inducing groups to ally with one party is that the previously dominant strategy of maintaining a collection of friends in both parties became far less effective. There used to be major bipartisan roads to policy making in Washington that did not run through the party leadership. Indeed, a "middle-out" strategy was often the best road to political success-a political reality that underpinned Mayhew's (1991) finding that divided government was uncorrelated with the production of major laws in the 1940s-1980s. While middle-out coalitions are not entirely obsolete (Lee & Curry 2019), party leaders now have much greater agenda control, and leaders look at which groups are solidly in their party's corner. The same incentives influence appointments to courts and regulatory bodies. This tightening party control makes a party-based alliance far more valuable for groups. The rise of negative partisanship further changes interest group incentives. For a party's core supporters, it has become increasingly suspect to work with politicians of the other party (Lee 2009). In such a setting, many groups will calculate that achieving their goals depends on helping their team win rather than reaching out to members of the other party.
There is a powerful self-reinforcing logic at play. The deeper and more intense the partisan divide, the stronger the incentives for most interest groups to join a team, and the more closely aligned groups are with parties, the stronger the incentives for them to do all they can to help their team win. As an interest group's party moves closer to their preferred policy positions (and the other party moves in the opposite direction), the stakes in the outcome of interparty conflict increase. As groups join teams, see increasing benefits of victory by their team, and thus work to insure those victories while punishing defectors, interest group political behavior can intensify polarization rather than moderating it. Indeed, the transformed interplay between groups and parties does more than just remove one of the traditional mechanisms that limit polarization. Many of these contemporary groups are national in scope and invested in an ambitious policy agenda. They eagerly push their partisan allies to advance that agenda wherever possible (Grumbach 2018, Hertel-Fernandez 2019. Groups may also depend on member mobilization strategies that rely on the intensification of conflict. These tendencies have generated concern among analysts about a possible "hollowing out" of the parties (Schlozman & Rosenfeld 2019). Parties have contracted out mobilizing voters to groups, who may also have considerable influence over fundraising and candidate recruitment. The fixation on winning elections that characterized many traditional party elites encouraged moderation. Party networks, however, increasingly lack the kind of robust organizational infrastructure that might limit extremism. Under conditions of polarization, they may cede power to groups that are more accepting of electoral risk to achieve potentially extreme ends (Hacker & Pierson 2014, Schlozman & Rosenfeld 2019.

Shifts in State Parties and Federalism
For much of its history, America's federal party system tended to act as a countervailing mechanism limiting partisan polarization. Even when the national parties were relatively polarized on a given set of issues-such as the tariff in the late nineteenth century-state and local parties provided a partially independent, geographically rooted power base to represent competing interests that cross-cut that division. For example, in the 1870s and early 1880s, Pennsylvania Democrats representing industrial districts constituted an important protariff bloc within the mostly antiprotectionist party. This bloc repeatedly allied with Republicans to frustrate their own party's efforts to lower rates, undercutting an otherwise intense cleavage (Peters 1990).
Perhaps even more importantly, the geographically decentralized party system attenuated polarization by providing a mechanism to incorporate new interests that fit uncomfortably with existing national party coalitions. From the start, American mass parties were premised as much on attempting to suppress issues that the two national parties preferred not to fight about as they were on highlighting the issues that neatly divided the parties. Martin Van Buren, for example, believed that the Democrats' cross-regional coalition would keep slavery off the political agenda (Shade 1998). Yet even as Democratic and Whig national leaders emphasized economic development issues, rank-and-file politicians attentive to local constituencies repeatedly brought the slavery issue to the fore. Much of this political work was done through third parties with distinct geographic bases (Brooks 2016), though these parties' efforts also pressured members of the two major parties to adapt. Fights over the gag rule, Wilmot Proviso, and Speakership elections, for example, repeatedly provided opportunities for the slavery issue to surface, with locally rooted party politicians often taking positions that departed from national party lines ( Jenkins & Stewart 2013).
In the post-Civil War era, national party polarization on economic issues was limited by regionally based intraparty cleavages that reflected sectionally distinctive political economies. Currency policy had very different implications depending on the economic development of a particular region. State parties provided a mechanism to represent these particular interests within each party. Similarly, midwestern progressives emerged as a distinctive faction in the early-twentieth-century Republican Party. Although Nominate scores would mark this as a time of high polarization, this progressive faction capitalized on its partially independent electoral base to pursue regulatory policies that conservative leaders opposed. Put simply, the national parties lacked a veto over state party positions and over the entry of new groups into a party coalition, even when those positions and groups undermined an existing line of cleavage. This process of geographically rooted factional entry repeatedly undercut partisan polarization (Schickler 2016).
Today's state parties no longer are well situated to play this countervailing role. Instead, they are far more integrated into national party networks in which key resources are outside the control of state party leaders. Part of this story has involved strengthening formal coordinating mechanisms at the national level, such as the Democratic National Committee, Republican National Committee, and congressional campaign committees. These organizations have become more active as a source of funds and professional services for local candidates, encouraging greater coordination across states (Lunch 1987, Paddock 2005. Nomination process reforms that empower ordinary voters have also shifted influence within states from professional, locally rooted politicians to policy-oriented activists who have often focused on hot-button issues that divide the parties nationally (Paddock 2005, Layman et al. 2006, La Raja & Schaffner 2015. Meanwhile, fundraising has been nationalized. Drawing on Federal Election Commission data, Hopkins (2018) finds that the share of itemized campaign contributions that cross state lines went from 31% in 1990 to 68% in 2012. This, again, undercuts the connection between politicians' ambitions and geographically specific interests.
More generally, the nationalization of politics, including of communication networks, has made it harder for state parties and politicians to tailor their identity to local conditions. As Hopkins (2018, p. 15) notes, "state parties themselves. . .especially as voters perceive them, have increasingly come to mirror their national counterparts" (see also Caughey et al. 2018). Three empirical indicators point to this nationalization pattern.
First, state party platforms are more similar across states and more distinctive across parties than in earlier eras (Paddock 2005(Paddock , 2014Hopkins & Schickler 2016). Even when a state party finds a national position to be disadvantageous locally-such as abortion for northeastern Republicans or guns for Democrats in rural states-they are more likely to avoid the issue than articulate a stance that departs from their national party.
Second, these platform differences are finding their way into state policy. Grumbach (2018) shows that state-level policies are increasingly polarized by party. He argues that these differences reflect responsiveness to intense demands from national groups allied with each party. Hertel-Fernandez (2019) documents how a national network of conservative interests has played an increasingly prominent role in shaping state policy agendas and legislation when Republicans gain power.
Third, the decline in state party autonomy is reflected in the increased alignment of state and national election outcomes (Hopkins 2018). Although there are occasions when a Democratic governor is elected in a red state or vice versa, the correlation between vote share across state and national offices has increased substantially in recent decades (Hopkins 2018). Studies have also shown that national-level factors are now having a bigger impact on state legislative elections than does the state party's own positioning (Zingher & Richman 2018).
As with the transformed role of interest groups, these changes in state party politics help make polarization self-reinforcing. When it became harder for state politicians to distinguish themselves from the national party brand in the eyes of voters, their incentives began to change. As Hopkins (2018, p. 6) puts it, state politicians "may well come to see their ambitions as tethered more closely to their status in the national party than their ability to cater to the state's median voter." When an issue potentially separates the state's median voter from the position of the national party, politicians' incentives to toe the national party line are likely to be stronger as voters prove less attentive to state-level differences and as the relevant audience for their behavior (interest groups, donors, etc.) becomes more nationalized. Indeed, state-level politicians will have an incentive to highlight national cleavages where their party has an advantage in their state, again furthering the polarization spiral (Hopkins 2017).
These changes have resulted in a more integrated party system. What were once relatively autonomous state and local party organizations that provided a basis for dissident factions to form and challenge national party lines now appear to "be rather small cogs" in a nationally oriented network (Paddock 2014, p. 165). Within this new more integrated system, state-level politicians find it in their interest to reinforce or even intensify existing national alignments.

Transformation of the Media
The presence of news outlets that are allied with a party or ideological cause is nothing new in American history. Nineteenth-century newspapers were, in many cases, clearly associated with parties, and often embraced sensationalistic material attacking the other party. These press outlets at times behaved in ways that exacerbated partisan and sectional divisions. Freeman (2018) shows, for example, how the press amplified the crisis of the 1850s, promoting conspiracy theories, surfacing evidence of pro-or antislavery plots, and spreading stories of sectional violence on the House floor.
Even so, the party press of the nineteenth century was not nationalized. Although more research is required on this topic, case study evidence suggests that voters in different regions who belonged to the same party did not necessarily receive the same messages about key issues. For example, as the fifty-first Congress debated the tariff and currency, GOP newspapers were divided regionally and thus provided a cross-cutting set of cues for many voters (Schickler 2001).
In addition to media outlets being more likely to follow the national party line than in the past, today's media landscape is more dominated by national news. This is primarily a matter of audiences shifting away from print newspapers and local television news sources, which, due to their geographic boundaries, continue to provide more state and local coverage than do other kinds of media sources (Hopkins 2018). The net result, however, is reduced information and engagement with state and local politics on the part of voters, which reinforces politicians' incentives to focus on national issues and cleavages. These changes may contribute to increasing the salience of nationally oriented political and social identities, in contrast to geographically rooted identities, furthering the trends toward nationalization and polarization (Darr et al. 2018, Hopkins 2018. Beyond these shifts in audience and emphasis, technological and commercial developments, such as the rise of cable news, talk radio, and social media, encourage the growth of an "outrage industry" that appeals to partisans (Berry & Sobieraj 2013). This industry has powerful incentives to intensify polarization in two respects. First, it attracts an audience by inflaming negative views about political opponents and making exaggerated claims about the political stakes involved. Second, it holds its audience by delegitimating other sources of information (Benkler et al. 2018). To the extent that a party's voters come to rely on media outlets with incentives to polarize, and increasingly treat alternative sources of information as illegitimate, polarization is likely to become more intense and durable.

The Landscape of Modern Polarization
A developmental perspective emphasizes that processes of polarization and nationalization have been deeply intertwined over the past half century. To understand polarization, many have, understandably, focused on southern realignment. Among other effects, this created more of a 50-50 nation, increasing incentives for elite partisan behavior (roll call voting, procedural hardball, etc.) to gain an advantage in the tightly contested battle for majority control (Lee 2016). The southern realignment can be viewed as the starting point for many, if not all, of the linked transformations we have discussed. It turned the GOP into a party whose base of rural evangelical whites tended to take the conservative position on a wide range of social and cultural issues, in addition to race (Schickler 2016, O'Brian 2019. We have also stressed the growth of the role of the federal government, which happened at about the same time. That growing role meant that more was at stake in politics. In response, groups got stronger and became more focused on controlling outcomes at the federal level. The mechanisms for self-reinforcement are clear. The more the parties diverged over increasingly consequential things, the more it mattered which side won. As new issues came on the agenda, the push for Republicans to take the conservative side and Democrats the liberal side became stronger. Due to the changing demographic/geographic bases of the parties and the growing incentives for both voters and groups to stay aligned with parties, guns, abortion, gay rights, and feminism all got absorbed into the party system in the same direction, with southern evangelicals who were becoming a key part of the GOP base on one side and coastal social liberals, who were increasingly likely to be Democrats, on the other (O'Brian 2019). All of this induced yet further changes in the organizational landscape and in the incentives for interest groups, state parties, and the media.
These coevolving forces have, in part through their impact on these meso-institutions, fundamentally changed the way the American polity fits together. In many cases, they did not just weaken the traditional generators of Madisonian pluralism; they transformed them into generators of intensified polarization. Interest groups and issues do not cross-cut; they stack, one on top of the other, along partisan lines. When new issues arise, existing groups, as well as politically aligned (and increasingly national) media, have incentives to push them into existing lines of cleavage (Layman et al. 2010).
Geography no longer encourages pluralism, as it often did even during what are typically characterized as highly partisan eras. Some of this may be just bad luck: Domains that used to run counter to one another or orthogonal to one another geographically (with the South conservative on race and culture, but more liberal or at least moderate on economics) no longer cross-cut. But much of it, we argue, reflects the growing forces of nationalization at work in our polity. If state party competition focuses on intrastate dynamics, it will tend to be multidimensional, distinctive across states, and a source of moderation and plausible bipartisanship at the national level. However, where media and interest groups are nationalized (and help create incentives for local parties to exploit even modest geographically based partisan inequalities), the role of geography may reverse. Nationalization puts the focus of state politics on the main national dimension, which means that even modest geographically based partisan inequalities may intensify over time (D.A. Hopkins 2017, D.J. Hopkins 2018. In time, the strong role of territorially grounded representation in the American system may come to act as an engine of polarization rather than a brake.

The Link to Mass Behavior
The meso-level changes in the organizational landscape described above are, in our view, critical to understanding how polarization could become more intense over time. These organizational changes have important ramifications not only for elite behavior in Washington and state houses across America but also for mass opinion and behavior, which-despite the well-known observation that most voters are not particularly "ideological" (in the conventional sense) or politically attentive-have also become contributors to increasing polarization.
Beyond the much-studied roll call record, arguably the facet of polarization that has been studied in greatest detail is mass-level changes in attitudes and behavior. Research in this area began with an extended debate concerning whether the mass public was, in fact, polarized, or instead, whether fierce divisions over policy were confined to a narrow segment of highly engaged activists and elites (see Abramowitz 2010, Fiorina & Abrams 2009). More recent work, however, has emphasized how even if most ordinary voters are not consistent liberals or conservatives with sharply polarized policy views, changes at the mass level have interacted with elite-level polarization in critical ways.
Indeed, growing polarization produced several dynamics at the level of voters that reinforce the initial polarization. One critical aspect has been the stacking of cleavages in a manner that encourages tribalism. In contrast to the cross-cutting cleavages that (often) attenuated the intensity of mass-level divisions in the past, social identities now line up more crisply with partisan divisions, including race, ideology, geography, religion, and education (Mason 2018). Mason argues persuasively that this alignment-and the associated degradation of cross-cutting social ties-has turned partisanship into a "mega identity." Mason suggests that social sorting is self-reinforcing. For example, as Republicans became firmly affiliated with conservative Christianity, individuals increasingly defined their partisanship in terms of religious-linked issues and images. Some even shifted their religious identity to line up with their partisan commitments (Margolis 2018). When a range of social identities all push in a single direction, it becomes much easier to see one's opponents as socially distant and deserving of hatred. Numerous studies have documented the increase in "negative partisanship" and "affective polarization" that is evident as more voters associate the other party's adherents with social groups that they dislike (Iyengar et al. 2012).
There are many interconnections between these developing mass attitudes and the changed institutional environment we have described. For now, we mention just one example of how mass polarization may short-circuit the system's traditional Madisonian features: This stacking of cleavages and growth in affective polarization may weaken the role of elections in moderating polarization. The pull of the median voter in incentivizing politicians to avoid extreme positions depends on citizens penalizing candidates when they move away from the center. But if there are fewer swing voters, this pull toward the center will weaken. 3 When an increasing share of voters see the other party as an alien force hostile to their core values, the willingness to punish one's own party's politicians for taking an extreme position will weaken accordingly. Bawn et al. (2012) argue that parties have always sought to capitalize on an electoral "blind spot" that allows them to serve intense policy demanders without alienating voters. When voters are more clearly sorted into enemy camps defined by stacked social identities, the size of this blind spot grows, weakening one of the most important self-correcting mechanisms to high polarization.

IMPLICATIONS
Conceptualizing polarization as a dynamic process has four important implications: (a) It directs our attention to the meso-institutional environment of the American polity; (b) it highlights the potential that these dynamic effects may disrupt the operation of self-correcting or countervailing processes; (c) it helps us account for and analyze asymmetrical, or party-specific, aspects of polarization; and (d) it provides a firmer analytic foundation for exploring the potential for democratic backsliding in the American polity.

The Significance of Meso-Institutions
The initial polarization of American politics triggered broad changes in major interest groups, state parties and governments, and media. These social arrangements are not formal (constitutional) rules, yet they play a crucial and often underappreciated role in mediating interactions among American citizens, among political elites, and between elites and ordinary citizens. Considering these meso-institutions is especially vital when we seek to explain large-scale political change, since the formal institutions of American politics are, for the most part, fixed. Over the past generation, changes in these meso-institutions have generally worked to intensify partisan polarization.

The Weakening of Self-Correcting Mechanisms
A striking feature of earlier periods of polarization is the extent to which meso-institutions operated as countervailing mechanisms that (often quickly) dampened the intensity and breadth of partisan warfare. Within a fragmented, pluralist polity, partisan pushes away from centrist or consensus positions, in support of more distinctive and aggressive policy agendas, have tended to generate a reaction.
Crucially, these reactions did not simply depend-as they do in many traditional (Downsian) models of party competition-on the median voter's political moderation. Instead, they worked through the meso-institutional features described above, often primarily within parties rather than (as Downs postulated) between parties through the constraints directly imposed by electoral competition. In large part because of the incentives created by institutional design, parties have been pluralistic and resistant to central direction. This pluralism has reflected the competing concerns of interest groups, geographically diverse state parties, locally embedded media, and the distinct institutionally derived interests of politicians situated in different positions within our fragmented system of political authority (Schickler 2001(Schickler , 2016.
If polarization helps transform these intermediary institutions and their associated incentives, however, these self-correcting processes may cease to operate. When interest groups have strongly committed to a party and regard the stakes of party defeat as very high, they may find it prohibitively costly to push back against unwanted initiatives. State parties, operating in an increasingly nationalized system of incentives, may cease to produce the political diversity that would generate backlash. The same would be true for highly partisan media. In a transformed polity, all of these forces, which might in the past have generated dissent and signaled to voters that a party had moved to the extreme, may no longer operate in the traditional fashion.
To put the basic issue more formally, a developmental analysis of polarization points to a more complex view of causal relationships in a polity. We may see a causal relationship in which a change in X changes Y. However, we should not assume that if X later returns to a prior value, it will have the same impact on Y, because the initial change in X may have had important effects on other variables as well. A central contribution of developmental analyses is to push us to question simple, symmetrical notions of causality in which the relationship between X and Y is fixed over time and across contexts. In fact, many of the developments that have been triggered or accelerated by polarization may be very difficult to reverse.

The Analysis of Asymmetrical Polarization
In recent years, Americanists have grappled with growing evidence that polarization is asymmetrical: The Republican Party has made a much more pronounced shift toward extremism (Hacker & Pierson 2005Mann & Ornstein 2012;Grossmann & Hopkins 2016;McCarty 2019). For conventional models celebrating the political dominance of the median voter, such asymmetry constitutes a considerable puzzle.
Exploring how polarization develops over time offers traction on this puzzle in two respects. First, it clarifies why a party's movement away from the center might not be self-correcting, even if the opposition party is not moving as far or as fast. In an intensely polarized environment, where party officials, aligned interest groups, and sympathetic media have strong incentives to stay with their team, the countervailing pressures that would generate a homeostatic political process may be weak or absent. In addition, negative partisanship may thrive in this kind of environment, again limiting the force of moderating pressures.
Second, appreciating the developmental elements of polarization may provide leverage for understanding the sources of asymmetry between the parties. Many frameworks for studying American politics are institutionally "thin," focusing on the mass electorate and a set of formal institutions structuring competition between elites. These thin formulations are almost intrinsically symmetrical, since competing teams of elites operate within the same rule structure and face the same electorate. Once we incorporate meso-arrangements, including interest groups, media, and diverse state parties and governments, however, the presumption of symmetry makes less sense. Instead, we can see the possibility that different parties face different structures, and therefore their incentives and behavior may differ as well.
For reasons rooted in differences in party coalitions, ideological orientations, and historical trajectories, these meso-arrangements do in fact look quite different for the two major political parties. This is perhaps clearest with respect to the media. As Grossmann & Hopkins (2018) show, the conservative media ecosystem, which developed partly in response to the perception of mainstream media bias, created new outlets that were explicitly tied to conservative organizations and causes, and focused on discrediting alternative sources of information. Grossmann & Hopkins (2018, p. 11) note that "the strategy was self-reinforcing, as right-leaning citizens came to rely more on conservative media and become less trusting of other news sources." The media ecosystem on the right is far more isolated from the informational mainstream than that of the left (Benkler et al. 2018). Messages from conservative media sources have worked to activate the existing symbolic predispositions of their audience, insulated viewers from countervailing forms of influence, and increased viewers' vulnerability to conspiracy thinking (Muirhead & Rosenblum 2019). Recent evidence suggests that Fox News-itself just one part of the conservative media ecosystem-has in fact pushed viewers' opinions further to the right (Martin & Yurukoglu 2017). As with interest groups, partisan media on the right has become tightly intertwined with the GOP, with increasingly open coordination and exchange of personnel (Skocpol & Williamson 2013).
Although empirical research on the interest group side is less developed, these differences seem likely to exist there as well. Republican networks seem to involve fewer but very powerful groupsespecially the Christian right and economic elites-with ambitious policy agendas that drive the entire party rightward. The Democratic coalition, by contrast, is made up of a wider range of interests, none of which (especially given the decline of organized labor) are large enough to dictate priorities. Thus, the interest group structure of the Democratic coalition makes it more conducive to compromise or log rolling (Grossmann & Hopkins 2016; see also Hertel-Fernandez et al. 2018).
An important contributor to the asymmetry may be the shift of political resources to the wealthy and corporations that has accompanied rising income inequality (Hacker & Pierson 2014). This shift has different effects on the two parties, accelerating polarization in the GOP while introducing an important cross-cutting cleavage for the Democrats. Like the Republicans, Democrats rely on wealthy donors, but those donors are unlikely to push the party to the left on regulatory and budgetary issues (Broockman et al. 2017). Thus, for a left-leaning party, growing inequality in economic and political power exerts a moderating influence on at least some important issues. In the case of the GOP, however, a growing concentration of economic power, mobilized by organizations like the Koch network and the Chamber of Commerce, has pushed a right-leaning party farther to the right (Skocpol & Hertel-Fernandez 2016).
In addition, a number of the Madisonian countervailing mechanisms that limited polarization in the past may remain more relevant on the Democratic side. For instance, the polarizing role of contemporary federalism that we have noted operates more weakly for Democrats, given the unfavorable geographic distribution of the party's voters. The growing concentration of Democratic voters in urban areas is, within the American electoral framework, politically inefficient (Rodden 2019). As a result, a Democratic victory in Congress requires winning red-leaning districts and states, creating an incentive to moderate and/or tolerate heterogeneity within the party. Republicans, by contrast, receive an electoral bonus from this political geography, facilitating a move to the right.

The Risks of Democratic Backsliding
Finally, exploring the dynamics of polarization may help equip Americanists to address an issue that comparativists, drawing on the experiences of other democracies, have recently brought to the fore: the prospect that the United States may be vulnerable to "democratic backsliding" (Mickey et al. 2017, Levitsky & Ziblatt 2018, Roberts 2019. Gradual democratic erosion is not a scenario that the approaches traditionally employed by Americanists are well-equipped to analyze. A developmental perspective on polarization provides a basis for seriously considering the comparativists' concerns. 4 Backsliding-the gradual undermining of rules, norms, and pluralistic organizational arrangements that sustain open political contestation-is (as the term implies) a dynamic process. True, American politics has been polarized since the early 1990s, but many of the key features of the polity, as well as the nature of partisan competition, are now quite different. The early 1990s preceded the development of the electoral weaponry that has further nationalized American politics, from Super-PACs to Fox News. In short, the social infrastructure that can deepen animosity between the parties and diminish the prevalence or impact of countervailing pressures is far more developed than it was two decades ago.
Prominent analysts of American politics have questioned the challenges to democratic stability by emphasizing that the parties may not be as far apart as their rhetoric or roll call votes make them appear. As Lee (2009Lee ( , 2016 has forcefully argued, much of what looks like polarization is better characterized as "teamsmanship." The durably close electoral balance between the two parties that distinguishes contemporary politics heightens incentives to engage in zero-sum contestation. From this perspective, conflict between the parties is often intense but not necessarily deep. Fierce jockeying for majority status obscures consensus on basic questions of government, and there is little reason to anticipate that polarization will destabilize political arrangements.
Lee is right to warn that we should not assume that fierce battles necessarily reflect deep disagreements. Yet even acknowledging the roles of close party balance and teamsmanship in accentuating partisan behavior, there is still good reason to be concerned about the intensity of the forces pulling the parties apart, the growing size of the divide between them, and the diminishing effectiveness of many of the mechanisms that would traditionally have pushed back against these developments.
Our analysis supports the view of Roberts (2019) and others that the dangers of backsliding are closely related to developments at work in the contemporary Republican Party. Powerful organized interests tightly aligned with the GOP have pushed the party rightward and have made the party vulnerable to what comparativists call bandwagoning. Bandwagoning is a process in which disparate elites within a coalition face growing incentives to go along with extremist or antidemocratic practices (Levitsky & Ziblatt 2018). A developmental perspective suggests that the prospects for bandwagoning are much greater in today's GOP than they might have been a few decades ago. Formidable, intensely partisan media have blossomed on the right. Along with the growing role of intense organized groups like the Koch network, the NRA, and organized evangelicism, these media forces have fueled negative partisanship. What Roberts (2019) calls the "movementization" of the GOP creates new incentives for political elites to stick with their team on matters-including challenges to established norms of restraint and tolerance, the rule of law, and the integrity and autonomy of core democratic institutions-where previously they might have chosen to dissent.
There has long been a view, common among both comparativists (famously Linz 1990) and Americanists, that the US system of checks and balances with its radical dispersal of political authority constituted a formidable barrier against democratic backsliding. We have already discussed why many of the stabilizing forces that traditionally were linked to these institutions seem much weaker today. In fact, in some cases (as with federalism), these arrangements now introduce new polarizing elements. We close by noting three additional reasons for questioning the extent to which Madisonian institutions-under the conditions operating today-offer effective insurance against democratic backsliding.
First, our electoral arrangements, under current circumstances, provide an unusual gateway to power for a politician with authoritarian inclinations. Presidential elections in a context of high negative partisanship facilitate a "half of a half" strategy (Pierson 2017). Donald Trump's populist, ethnonationalist posture proved highly effective within a "movementized" GOP nomination process. Once he was the nominee, both unenthusiastic elites and skeptical Republican voters felt compelled to come aboard, given their intense dislike of the alternative in the November 2016 election. 5 As Roberts (2019) points out, parliamentary systems are much less vulnerable to this sequence of events.
Second, the Madisonian system of territorial representation may create a powerful incentive for bandwagoning that is absent in systems lacking that feature. The stacking of cleavages in our polarized system has helped to deepen the territorial divide between the electorates of the two parties. Increasingly, elected officials find themselves facing local electorates dominated by a single party, further undercutting the effectiveness of traditional Downsian mechanisms for limiting extremism.
Finally, the emergence of hyperpartisanship means that the check on authoritarian developments in the presidency that the Madisonian system relies on most, Congress, may not work. Instead, GOP members of Congress face multiple incentives to bandwagon rather than resist. Among those incentives are the intense preferences of the party's interest groups, the heavily "red" and negatively partisan electoral bases of these politicians, and the likelihood that influential partisan media will exact a very high price for defection. Given these realities, it is perhaps unsurprising that even the most extreme and disquieting behavior in the White House fails to shake the solidarity of Republican members of Congress. In short, the developmental perspective we offer raises a disturbing prospect: Under conditions of hyperpolarization, with the associated shifts in meso-institutional arrangements and the growth of tribalism, the Madisonian institutions of the United States may make it more vulnerable to democratic backsliding than many other wealthy democracies would be.