How Sociology Lost Public Opinion

Political Sociology at Midcentury

The organization of the field of political sociology largely after World War II built on some of the foundations from the classical heritage, even while it significantly reshaped many of its key insights. One central figure was Seymour Martin Lipset, who once famously described his own standpoint as one of “apolitical Marxism” (Lipset [1960] 1981:521). Lipset argued that the rise of the educated middle class, and its distinctive values and policy preferences, were key to understanding the emergence and historical robustness of Western democratic capitalism.

Marxian-inspired questions about the consciousness of the working class (in an era of rising affluence) and other sources of socioeconomic influence on political orientation and attitudes provided another set of questions (e.g., Bell 1960; Centers 1949; Lipset [1960] 1981; see Rossi 1959 for a review of work in that era) that motivated scholarly research and debate. During this same time, other scholars focused less on class and status and more on how social networks and workplace organizations shaped the attitudes and political behaviors of individuals and groups (e.g., Berelson et al. 1954; Goldthorpe et al. 1969; Lazarsfeld et al. 1948). Yet in both class-centered and social network traditions, questions of how politically relevant opinions and identities were shaped (and occasionally remade) were fundamental concerns.

Important traces of the Durkheimian notion of social solidarity and the collective conscience carried forward in midcentury work that focused on the role of “national values” as key to understanding sociopolitical variation. Almond and Verba (1963), in their The Civic Culture, for example, suggested in their five-nation study that polities have distinctive political cultures that grow out of underlying civic values and orientations of citizens. The idea that a dominant set of national values can be identified in one strand of scholarship on the welfare state (e.g., Rimlinger 1971) was very much alive in midcentury theories of “American exceptionalism” (Hartz 1955; Hofstadter 1965; Lipset 1963; see also Bellah et al. 1985). Similarly, Inglehart’s (1977) model of political change emphasized cross-national variation in political outcomes based on the relative importance of “materialist” and “postmaterialist” values.

Tilly’s Analysis of How Citizens Speak to Their Rulers

The most prolific and wide-ranging political sociologist of the contemporary period is arguably the late Charles Tilly. A key figure in establishing the larger social history movement of the 1960s (see Tilly 1961), Tilly’s work (and that of his numerous students and collaborators) expanded the analytical focus of political sociology to consider how social movements and collective action—disruptive politics occurring outside of institutional channels—could become critical sources of political change (see Tilly 1978, 1986; see also McAdam, Tarrow, and Tilly 2001). In his numerous case studies of collective action in France, Britain, and elsewhere, Tilly developed methods for systematic counts of protest events and analyzing them to identify “repertoires” of action. He also made foundational contributions to the political sociology of the state by developing theoretical models of state formation and development rooted in an analysis of war making and revenue collection (see Tilly 1985, 1990) and the social foundations of democracy (Tilly 2007).

Tilly’s work provided, however, a very limited account of the place of public opinion, a significant omission for a body of work that seeks to focus attention on the voices of ordinary people through social movements and other forms of collective behavior. He did, however, once address the question of public opinion (Tilly 1983), arguing that it was through collective action that citizens found ways to express their opinions to political elites long before polling and other formal ways of measuring public opinion came into existence. While not claiming that opinion polls are irrelevant in the contemporary world, Tilly (1983:477) asserted that “even today we can look to the language of popular collective action as a complement to the knowledge offered us by elections and surveys.” But no further account of how the knowledge offered by elections and surveys might matter is provided. This is especially striking given the large number of historical instances in which mass opinion stands at odds with the demands expressed through protests. Yet throughout his vast body of work, there is no further clarification or detailed discussion of the relationship between public opinion as measured in elections and surveys and the public opinion expressed through collective action.

Tilly’s contributions to political sociology (and the importance of his neglect of public opinion) were not limited to his work on collective action and his later work on democracy. In his influential scholarship on state formation through war making, Tilly (1985, 1990, 1994) focuses on macrocomparative dynamics in the creation and maintenance of the modern nation-state. At the heart of his account is the role of military capacity in generating state structures and state capacities. A key example is the effects of military conflicts in promoting the development of taxation capacity (see, e.g., Tilly 2007, 2009). Regimes that could not develop these institutional capacities fell behind or would be absorbed by stronger units. Here too, however, public opinion is absent from the account. The possibility that citizens more readily assent to paying higher taxes in wartime is not, for example, engaged as an important piece of the puzzle of why states are able to improve or increase revenue collection in the first place.

Skocpol’s Historical Institutionalism

Theda Skocpol’s work, alongside that of Tilly, has probably done more than any other both to reinvigorate the field of political sociology from the late 1970s onward and to push it toward macrostructural concerns. At the center of Skocpol’s theoretical contributions has been the project of “bringing the state back in,” that is, directing attention to the importance of state institutions and state managers in the making of social and political change (see Skocpol 1979, 1980, 1992, 1995; Weir, Orloff, and Skocpol 1988). In her landmark States and Social Revolutions, Skocpol (1979) argued that analysts of revolutions had paid insufficient attention to the role of crises in state administrative capacity and state-elite relations in understanding the conditions under which successful revolutions occur. Skocpol insisted that postrevolutionary state building influences the long-run fate of revolutionary breakthroughs. Although peasant consciousness (and peasant revolts) was a necessary condition for successful revolutions in France, Russia, and China, structural features of the prerevolutionary and postrevolutionary state were decisive.

Skocpol’s scholarship between 1980 and the late 1990s centered largely on the formation and development of the American welfare state. In particular, Skocpol examined how structural factors, rooted in the organization of democratic political institutions, pushed the American welfare state down paths very different from those of European or even other Anglo-American welfare states. Skocpol and other historical institutionalists identified the constitutional design of governments as a critical factor in accounting for comparative difference, with the U.S. welfare state’s being particularly disadvantaged by a number of features of American political institutions that distinguish it from its European counterparts. For example, the United States is characterized by political decentralization, which lowers the likelihood of generous welfare programs. The multiple “veto points” found in American political institutions provide opportunities for business organizations and their conservative political allies to more frequently block egalitarian policy initiatives than occurs in more centralized polities (see Immergut 1992; Skocpol 1995; Swank 2002). Another example can be found in the party system itself, where first-past-the-post electoral rules did not favor the rise of social democratic parties and also created fragmented parties with limited mobilization capacities (Alesina and Glaeser 2004; Shefter 1994).

Historical institutionalists have paid special attention to the idea of “path dependency” in understanding policy development (see Hacker 2002; Pierson 1994, 1996, 2000; Skocpol 1992). Formulated as a thesis, path dependency anticipates that early developments in the history of a policy establish a trajectory or subsequent “path” that, once adopted, is difficult to reverse (see Pierson 2000 for a systematic statement). But why? The key causal mechanism is policy feedback, through which the establishment of a new government policy exerts influence over sources of support and the strategies of actors seeking to shape policy. In elaborating the idea that “policy causes politics,” Pierson (1994, 1996, 2000) has argued that constituencies and expectations are created around policies that make them difficult to eliminate once they have been established. Furthermore, new policy ideas and proposals for reform start from the existing policy framework; to be taken seriously, reform proposals must operate from within a field or set of understandings that quickly becomes entrenched. In this way, institutional developments at time₁ almost always constrain developments at time₂, irrespective of the demands of social movements from below or the activities of elites from above.

Skocpol and other institutionalist scholars have frequently distinguished and defended their causal arguments through simultaneous and vigorous rejections of a possible role for public opinion (often relabeled “cultural values”), or what they see as “behavioralist” approaches (see Heclo 1974:288–93; Immergut 1998; Skocpol 1992:15–23; Steinmo 1994). What are the grounds for this conclusion? First, there are arguments that the hypothesized liberalism of the American political order did not preclude certain kinds of expensive and sometimes even precocious welfare state building in the late nineteenth century (civil war pensions), in the early twentieth century (mother’s pensions, worker’s compensation), and during the New Deal (e.g., relatively high levels of social spending) (see Amenta 1998; Orloff 1993; Skocpol 1992). In other words, whatever Americans’ underlying preferences for social provisions might have been, it is primarily proposals having an affinity with the logic of American institutions and pre-existing policies are most likely to succeed. More generally, the demands for policy change that do emerge are frequently motivated by, or filtered through, an institutional framework favoring some ideas over others. While policy experts and policy ideas (see, e.g., Orloff 1993; Skocpol and Rueschemeyer 1986; Skocpol and Weir 1985), social movements (Amenta 1998, 2006; Clemens 1997), and state managers (Heclo 1974; Skocpol 1995; Skocpol and Finegold 1995) all have a place in the struggle over policy, public opinion itself is discounted as a potential causal force.

Esping-Andersen and the Power Resources Model of Welfare States

Research on, and theorizing about, the rise and consolidation of welfare states in the rich democracies have occasioned some of the most vigorous and far-reaching debates in political sociology in the past 25 years (see Arts and Gliessen 2001; Korpi 2003 for reviews). Although there have been a wide variety of theories proposed to explain the rise of the modern welfare state—including historical institutionalism, discussed above—perhaps the most widely influential model has been the power resources approach. Associated first and foremost with the influential writings of Gosta Esping-Andersen and Walter Korpi, and later with the work of Evelyne Huber and John Stephens, Alexander Hicks, and Duane Swank, power resources analyses begin from the premise that enduring class inequalities facilitate the formation of social groups with distinct and competing interests in relation to government policy (see Esping-Andersen 1985, 1990; Hicks 1999; Huber and Stephens 2001; Korpi 1983, 1989; Stephens 1979; Swank 2002). These classes and groups develop organizational capacities through parties, unions, and interest group organizations. Organizations representing working-class, agrarian, and middle-class interests, working separately and sometimes together, eventually gained enough strength to build welfare states.

How do welfare states evolve? For power resources theory, the structure of class compromise is central. Elections in welfare capitalist democracies provide classes and class-related organizations such as unions with a recurring set of opportunities to extract benefits from national government or to block those reforms that would lead to concessions and relative losses. Class conflict through the ballot box—the “democratic class struggle” in Korpi’s (1983) adoption of the Lipset phrase—is thus a central mechanism in the development of welfare states (see also Hicks 1999; Huber, Ragin, and Stephens 1993; and Huber and Stephens 2001). The second set of factors relates to the (varying) strength of unions and corporatist bargaining structures. Here, the capacities of organized groups to project power—even if they do not actually use it, as in the case of strong unions threatening to strike—can influence policy makers.

The key to understanding welfare state outcomes is the development of what Esping-Andersen (1990) famously described as the “decommodification” of labor (in which the welfare or life chances of individuals and classes on labor market activity decline in proportion to the extent of income support programs and public service provision). Differences in the degree of decommodification and the forms through which social provision is provided (targeted versus universalistic), Esping-Andersen (1990) argued, produce a division of the rich democracies into three broad clusters of countries: a “social democratic” cluster in Nordic Europe, a “conservative/Christian democratic” cluster in continental western Europe, and a “liberal democratic” cluster in the Anglo-American democracies. Social democratic welfare states are characterized by high degrees of generosity, universalism in social provision, and a significant degree of labor market decommodification. Christian democratic regimes couple high generosity with targeting aimed at reinforcing traditional gender roles. Liberal regimes have comparatively lower generosity and more extensive labor market commodification. Scholars following in Esping-Andersen’s footsteps have long debated the nature of variation in welfare state regimes and whether there are additional ideal types of welfare states (Castles and Mitchell 1993; Kangas 1994; Kersbergen 1999). But the basic framework emphasizing these broad sets of distinctions has been widely and robustly applied in comparative research.

Power resources theory comes perhaps the closest of the theories under consideration to an endorsement of a role for public attitudes in contributing to welfare state making. Critical pathways to each of the three welfare state regime types is evident in the patterning of cross-class coalitions and in the timing and sequencing of these coalitions. Examples of the importance of these coalitions abound. For example, when a highly organized working class is successfully allied with farmers, this tends to facilitate the emergence of welfare states with extensive social provisions, as in the case of Sweden (Korpi 1983). By contrast, when the working class has high levels of regional or racial/ethnic fragmentation that inhibit class alliances, the interests of employers will tend to figure more centrally in regime development, as in the case of the United States (Esping-Andersen 1990).

Furthermore, because they directly benefit from many welfare state programs, middle-class citizens may join with the labor movement and its allies to support generous welfare programs. According to Korpi and Palme (1998), the encompassing scope of public services within Nordic welfare states contributes to higher levels of legitimacy in comparison with regimes organized around means testing (or more limited benefits for middle-class households). Universalistic polities, by reducing zero-sum conflict between welfare recipients and nonrecipients, are expected to secure high levels of both middle-class and working-class support (Rothstein 1998; Svallfors 1995, 1997).

These lines of thinking have clear implications for understanding the underlying nature of welfare states across polities. Yet scholars within the power resources tradition appear generally to be reluctant to bring into focus or empirically buttress these implications with detailed analysis of how mass opinion fits into their larger theoretical account. At best, it leaves public opinion as a residual rather than a central factor in the policy process.

Mann’s History of Power

Michael Mann’s large-scale, multivolume social history of power offers fresh insights into the central role of what he calls “ideological power,” one of four types of power he analyzes across time and space (Mann 1986, 1993). In this sense, Mann’s approach to thinking about historical dynamics departs sharply from Tilly’s. Perhaps not coincidentally, some of Mann’s earliest sociological work directly engaged questions of attitudes and public opinion in liberal democracies and in relation to working-class consciousness (Mann 1970, 1973).

With his turn to large-scale historical analyses of power, Mann embeds both popular and elite consciousness in the “I” of his “IEMP” (ideological, economic, military, political) model of power. Mann identifies ideological power (and the capacity of ruling authorities to exert “ideological power” and “ideological organization” on governed peoples) as one critical resource of power (see Mann 1993:22–24). Perhaps not surprisingly, given the overall direction of political sociology during this era, Shroeder notes that of the four components of power, “ideological” power is the one that has been the most controversial (2006:6).

Mann’s focus in the first volume of his history (through 1760) centers primarily on religious organization and beliefs as a source of elite cohesion and, in many cases, popular consciousness. For example, at the end of a chapter about “the global emergence of civilizations,” Mann asserts that “in all cases ideological power had a privileged role in solidifying regional organizations. … The religious cultures were socially transcendent, providing organized solutions to problems affecting an area more extensive than existing authoritative institutions could regulate” (1993:126). Elsewhere, ideology and religion play critical roles in the normative integration of state officials spread over large swaths of lands (e.g., Mann 1993:301–25 on Christianity in the later Roman Empire).

While emphasizing the impact of beliefs grounded in religion prior to the nineteenth century, Mann’s examination of the dynamics of power turns more explicitly on structural sources of power (economic and political), albeit with some attention to socialism and nationalism as motivating sources of social change through social movements (but not ordinary citizens’ opinions). In examining the role of nationalism in the late nineteenth and early twentieth centuries, Mann suggests in places that public attitudes play a key role in promoting the move toward violent conflict culminating in World War I:

The middle class … identified its own economic, political, and cultural progress with the achievement of national citizenship. It became intensely proud of being “British,” “French,” “German,” etc. … In Germany, Britain, and France, the middle class dominated jingoist, chauvinist, militarist and even racist pressure groups. In the last decades before 1914, states found themselves confronted by a new problem. When boxed into geo-political crises, they were urged on to further aggression by this middle class “public opinion.” (Mann 1988:156)

Elsewhere Mann argues that modern states—especially the United States—have become militarized in ways that create what he calls “spectator-sport militarism,” with the masses capable of being mobilized for war making by manipulative elites (Mann 1988: 183-186). Mann’s analysis thus captures the well-known propensity of publics to support military interventions, at least in their early stages.

But his more recent analysis of the Geoge W. Bush administration’s war on terror and imperial ambitions seems largely to neglect the flip side of public opinion as a constraint, for instance, or the possibility that growing public opposition to the war (and not imperial overreach) undermined some of the initial long-range goals of the administration. That is precisely the expectation of the famous “rally round the flag” model of public opinion during wartime (Mueller 1994, 2005), one that Mann himself has embraced (see Mann 2004:101–104). Understanding the social and ideological forces relating to military adventures in the modern world is substantially enhanced by, and hardly can proceed without, more detailed investigation into the interrelationships between opinion formation and military policies and campaigns.

Domhoff’s Class-dominance Synthesis

The final influential theory in political sociology that we consider is the “power elite” model of American politics, especially as developed by G. William Domhoff (1967, 1990, 2006; see also Gordon 1994, 2004; Quadagno 1988). ⁴ The core ideas of the power elite model—the role of common backgrounds and network linkages among the upper class, the corporate community, and policy-planning organizations and the disproportionate influence of the members of this core group over policy and politics—have been brought together into a multidimensional model of power and its reproduction in the United States in Domhoff’s various writings.

Analysis of the role of business interests in the development of social policy has been a fertile and wide-ranging area of empirical research. Studies of the origins of the Social Security Act (Domhoff 1996; Gordon 1994; Klein 2000; Quadagno 1988) and the repeated failures of proposals for national health insurance (Gordon 2004) are representative examples. While emphasizing different groups and/or timing and sequencing, business-centered interpretations of social policy making attribute welfare state innovation to the perennial influence of important segments of the business community (or, in the case of national health insurance, the failure to craft a policy approach capable of bringing important business groups behind it). These accounts provide notably little connection to mass policy attitudes. The adoption of a comparatively limited social safety net in the United States, for example, is traced not to Americans’ long-standing unwillingness to support more universal and generous programs (see Newman and Jacobs 2010) or to activists’ capacity to mobilize such latent opposition. Instead, the political alignments of state and business in the earlier formation of US social policy as such are key.

Domhoff’s scholarship has, however, pushed much harder in exploring (if ultimately rejecting) the possible impact of public opinion than most other political sociological scholarship. He has taken seriously the challenge of unpacking how dominant actors seek to shape and influence public attitudes (see also Ginsburg 1986; Mills 1956; Parenti 1995). Domhoff argues that on policy issues of central importance to the power elite—taxes, governmental regulation of economic activity, spending on social provision—there are a variety of elaborate and well-funded “opinion-formation” activities that elites pursue in an attempt to shape and control public preferences. These include “public relations” and “public affairs” initiatives and efforts to influence the mass media both directly (through corporate ownership) and indirectly (through support for think tanks and idea promotion and the publication of op-eds and policy advertisements) (see Domhoff 2006). Domhoff also has focused novel attention on the dissemination role of “policy discussion groups,” settings where middle-class professionals meet to discuss policy questions with opinion leaders (2006: 90-97).

Taken as a whole, such efforts provide startling evidence of attempts to manipulate public opinion and by implication the idea that public opinion must be important enough to try to manipulate in the first place. Indeed, the paradox of Domhoff’s (1990, 2006) analysis of elite attempts to remake public opinion is the implication that in the absence of such efforts, public opinion might propel government policy in a different direction. But the actual consequences of elites’ opinion-shaping efforts are themselves not well documented. Domhoff’s (1967, 1998, 2006) classic Who Rules America? has (across multiple editions) left entirely unexamined the actual degree to which mass opinion can be, and was in fact, altered by even the most expensive and well-orchestrated corporate campaigns.

This vagueness is striking given that a large social science literature suggests the considerable difficulty of moving public opinion on any salient issue, especially in directions that are substantially at odds with baseline opinions (Brooks and Manza 2007; Page and Shapiro 1992). To be sure, Domhoff and others have noted the capacity of political elites to achieve short-term framing effects and/or changes in salience in situations such as proposals for national health insurance in the United States (cf. Jacobs and Shapiro 2000). But such stories of successful corporate influence may be rarer than Domhoff’s model would suggest. Indeed, counterframing by policy challengers in most cases appears at times sufficient to leave mass opinion relatively intact (Chong and Druckman 2007a, 2007b). This research suggests that in the absence of truly one-sided information flows (as, e.g., in the immediate aftermath of the terrorist attacks of 9/11), counterframes can push aggregate opinion back to the baseline (Chong and Druckman 2007a).

Understanding the Analytical Move Away from Public Opinion

The structural turn in political sociology, combined with a renewed attention to social movements from below, fit in two ways with a larger movement within sociology away from the functionalist approaches of the 1950s. Once public opinion was cast as a variant of “national values,” it could be easily read (mistakenly) as part of the functionalist canon (cf. Mills 1959). Some leading public opinion scholarship contributed to such a reading. For example, in their major and widely cited 1967 essay on social cleavages in voting behavior, Lipset and Rokkan embraced a systems and functionalist logic to understand how national and industrial revolutions produced remarkably “stable” class-based political alignments (said to be “frozen” since the 1920s). While their timing could not have been worse (on the eve of a major global thaw), the bigger problem with the Lipset and Rokkan (1967) thesis, and other models of public opinion that emphasized shared “values,” was that it ran directly counter to the impulses of the 1960s generation of scholars (cf. Sica and Turner 2005).

Because research on public opinion, at least as construed in some versions of 1950s/1960s sociology, seemingly had an elective affinity with the functionalist paradigm, it was a small step to abandonment by a younger generation of scholars promoting alternative and radical theories within the discipline from the 1960s onward (cf. McAdam 2007). Burgeoning interest in, and scholarly optimism toward, the civil rights, New Left, and women’s movements was a powerful lever for viewing dynamism and dissent in this new era as poorly captured by static conceptions of social order (e.g., Gamson [1975] 1990; Gitlin [1980] 2003; Piven and Cloward 1978). Closely informed by the growing influence of comparative/historical and macro-oriented work, a successive generation of scholars found deep and largely a priori fault with theory and research on public opinion (Eliasoph 1998; Ginsberg 1986; Immergut 1998).

Bringing All the Citizens In

Among the most lasting legacies of Tilly’s varied body of work has been reexamination of the sources and consequences of collective action and how these reflect the underlying opinions of ordinary citizens who are not part of the governing elite. But even though Tilly never considers public opinion (directly or indirectly), there are places where Tilly hints at, or implicitly assumes, a degree of relevance. Consider his landmark theoretical book on social movements, From Mobilization to Revolution (Tilly 1978). Here, Tilly develops a political conflict model of collective action in which movements arise not out of grievances or explosions of frustration (as in the classical collective action literature such as Smelser 1963) but where there are resources (organizational and political) for challengers to draw on in seeking to mobilize supporters. (This issue frequently arises in the context of formal models of collective behavior as well; see, e.g., Baldassarri 2009.)

While it is mobilization rather than grievance that is the explicit focus of his work, Tilly’s organizational model does require (or assume) the existence of some degree of citizen grievances for mobilization to take place (as we suggested in the case of the tax revolt studies discussed in the previous section). But where do these grievances come from? In expanding beyond Tilly’s initial organizational model, subsequent work has probed the ways in which social movements both take advantage of and seek to reshape or extend popular grievances through “framing” processes (see McAdam, McCarthy, and Zald 1996; Snow and Benford 1988).

Social movements develop frames that at times draw on existing grievances and attitudes. When communicated successfully to audiences, they raise the likelihood that individuals will tend to connect these prior dispositions and experiences to broader struggles or organizations. Civil rights leader Martin Luther King Jr., for example, did not have an impact simply because of effective mobilization or organizational resources; King also masterfully employed rhetorics of equal opportunity and human rights that brought individuals and groups into the movement and changed the salience of civil rights issues for northern whites (Burstein 1998b; McAdam 1996). In turn, this expanded the movement’s resource base and strategically reoriented the alignment of a number of government officials and other elites (Morris 1985).

Following in Tilly’s footsteps, analysts of social movements also have grappled with questions about how, when, and why social movements have success. One important way that protests or social movements matter is their influence in the making of public policy. The pathways of influence may be direct (politicians changing their policy stances in the face of civil unrest) as well as indirect (through a movement’s impact on elections, media coverage and frames, public opinion) or even entail cultural tastes and trends (Burstein 1999; Gitlin 1987). The empirical question then becomes how to measure and assess the degree and type of movement impacts. Public opinion is one key pathway through which such framing impacts occur (cf. McAdam and Su 2002).

Consider another example. Tilly’s model of democracy, as we noted, shifts attention to noninstitutional sources of democratic practice or consolidation such as trust, equality, and the reigning in of powerful forces capable of undermining democracy. But as his model does so, Tilly tends to move away from topics having connections to public opinion and electoral politics, ones seen as important by an earlier generation of political scholars (universal suffrage, a free press, elections, and the operation and popular dynamics of political parties). If we seek, however, to fully assess democratic performance without making a priori assumptions about which factors matter (and which don’t), some consideration of policy responsiveness to public opinion is also appropriate and ultimately necessary. Do citizens get the policies they want? And when citizens disagree substantially (or diverge from the endorsements of social movements) on specific policy issues, whose preferences ultimately register with politicians? This suggests an important, positive research agenda, one that can be carried out in conjunction with, or otherwise complementing, a focus on noninstitutional sources of democratic performance and legitimacy.

Rethinking the Logic of Path Dependence in Historical Institutionalism

Turning to the model of historical institutionalism in the work of Skocpol and later analysts such as Pierson and Hacker, we also see strong possibilities for expanding the analytical reach to engage public opinion. Historical institutionalists have often argued that the structure of political institutions creates a distinct set of opportunities for policy entrepreneurs. But these same institutions or policies shape the views of the mass public and in certain cases themselves evolve in response to public opinion. As policy moves down a particular path, it can take on institutional momentum in part through the accumulation of high levels of public support over time (cf. Brooks and Manza 2007; Campbell 2005). Policy feedback can occur because of such opinion dynamics.

In the scholarly literature on welfare states, there exist plenty of examples of public support for programs as they develop becoming a critical source (alongside key interest groups) of pressure to maintain policy provisions. For example, the social security program in the United States survived the George W. Bush administration’s vigorous plans for privatization in part because powerful interest groups such as the AARP (formerly American Association of Retired Persons) and unions opposed it. But interest group opposition to the plan also drew strength from the fact the plan was deeply unpopular with the U.S. public (Cook and DeSantis 2008). Interest group mobilization in this case appeared to convey meaningful information to members of Congress about mass preferences and the generally high salience of the issue to politicians. Public opinion can play a role in this process because, as Pierson (1996:176) has argued, “politicians are likely to pursue strategies that will not damage their chances for reelection.” ⁶

This is not the end of the story. The influence of institutions on political outcomes can, and in some cases should, be supplemented by theoretical models that pay attention to how public opinion itself can interact with or shape institutions. An instructive example can be seen in the emphasis that political sociologists have often assigned to the overarching powers of the American legal system. The fact that federal courts can and often do overturn legislation or engage in policy making from the bench is an important aspect of the American political system. But scholars who examine changes in Supreme Court rulings find that the Court is in some contexts responsive to public opinion (Erikson et al. 2002; Friedman 2009). These two types of processes coexist and are necessary elements of a full understanding of the politics of law (cf. Fourcade 2011 for a parallel argument in a French-American comparison).

A similar lesson can be discerned with respect to the question of why decentralization matters so much, a point Skocpol and many other historical institutionalists have emphasized. Garland’s (2010) institutional analysis of the persistence of the death penalty and Lieberman’s (1998) analysis of the limited welfare state in the United States both demonstrate how institutional fragmentation and multiple veto points make it possible for public preferences in local contexts to matter so much. For example, Garland shows that it is mostly southern states (with high levels of public support for the death penalty) that continue to execute, while in much of the rest of the country (where support for the death penalty is far more limited), states rarely if ever employ it. This interaction between institutional context and mass opinion highlights how institutional analysis alone can be usefully supplemented by analysis of public opinion.

Extending Mann’s Model of Ideological Power

Michael Mann’s historical sociology, as we noted above, steps at times toward an embrace of popular as well as elite beliefs in understanding the dynamics of social power in his notion of extensive ideological power. Yet Mann’s main focus is on the ideologies of elites across his many cases. To be sure, the lack of good sources of data about mass opinion hinders systematic analysis, but nothing in his theoretical model would preclude taking mass attitudes into account.

Consider, for instance, Mann’s analysis of religion as a source of ideological power. Religion becomes a source of social power when it unifies elites. In a commentary on Mann’s work, Gorski (2006:111) notes that while Mann’s focus on religion as a key source of early state building advances on militarization models of state formation (cf. Tilly 1990), he does not further extend the analysis of religion to thinking about variation in state capacities to pacify citizens and win consent to state policy. Gorski’s (1993, 2003) own historical work provides a significant demonstration of how popular consciousness before the era of modern democratic politics can be examined and theorized. Gorski argues that even before extensive state authority and bureaucratization of state organizations were established, nonauthoritarian governance could be fostered where mass identity with Calvinist religious traditions was present. The Dutch Republic was a successful entrepreneur in establishing modern state institutions in part because the religious beliefs associated with the Calvinist tradition aligned with styles of collective discipline—these would include communal surveillance, respect for legal authority, and bureaucratic state building. These powerfully yoked mass and elite were on a common pathway toward constitutional government that was not found elsewhere to the same degree.

Building on similar themes, Kahl (2005) provides an even more expansive examination of how religious beliefs shape different paths toward poor relief during a 500-year period. Kahl (2005) argues that the three dominant religious traditions in western Europe and North America—Catholicism, Lutheranism, and Reform Protestantism—gave rise to distinctive mass and elite models of justifications and policies for poor relief. Kahl argues that these models, although emerging before mass opinion could be systematically measured, have produced enduring patterns of difference that can be traced up to the present. The influence is felt throughout the political order, and she argues that variations in state policy can be traced, in part, back to founding religious traditions.

Public Opinion as a Power Resource

In the case of power resources theory, analytical attention to mass opinion fits, we believe, remarkably well within the theoretical premises of the model, waiting only to be more substantially engaged by practitioners. The question raised by the power resources model’s insistence on the importance of class alliances or the role of Left parties is, Which factors enable those alliances (and resulting Left party electoral successes) in the first place? Focusing on class resources—such as the size of the industrial working class and its political mobilization in the party system—ultimately brings us to the question of how welfare states have endured in the face of globalization pressure and the declining organizational strength of labor unions in virtually all capitalist democracies (Wallerstein and Western 2000). The most likely answer to this puzzle of persistence is that generous welfare states, once established, win mass popular support. For instance, because they directly benefit from many welfare state programs, middle-class citizens may join with the labor movement and its working-class members to support the maintenance of welfare programs. Universalistic policies, by reducing zero-sum conflict between welfare recipients and nonrecipients, secure high levels of both middle-class and working-class support (Korpi and Palme 1998; Rothstein 1998; Svallfors 1995, 1997).

Within the power resources tradition, analysts have yet to employ the language of contemporary political psychology and opinion research, but clear hints of a role for mass opinion as a source of policy persistence can be found in a number of recent statements. For example, Huber and Stephens (2001:3) propose that mass opinion may operate as a “ratchet,” whereby “the rapid growth of support for welfare state policies … that benefit a large portion of the population … then turns these policies into the new point of reference for discussions on further welfare state development.” Korpi (2003:598) likewise suggests that “it can be argued that major welfare-state institutions are likely to be of relevance for the formation of values, attitudes, and interest among citizens in ways that are of relevance for patterns of collective action.” Power resources theorists have also sometimes given an explicit role for public opinion in protecting existing levels of social provision from significant cutbacks (see, e.g., Esping-Andersen 2000:4; Huber and Stephens 2001:322). There are additional and fuller opportunities for extension and integration.

Corporate Power and Public Opinion

Finally, what of the power elite/class dominance model? We have already noted that the importance of shaping public opinion to powerful actors (something Domhoff has empirically documented they undertake at considerable expense) implies that at some level public opinion matters. Put another way, it would be irrational for corporations, parties, and other elite organizations to expend resources in attempts to influence public attitudes unless those attitudes do, in fact, affect election outcomes and public policy.

Furthermore, by placing elections and public opinion at the center of struggles over public policy, Domhoff’s recent formulations of his “class dominance” model productively place political sociology firmly into a dialogue with opinion research and empirical democratic theory (see, e.g., Domhoff 2006). His research on the role of the political money in American elections calls attention to one important set of factors, but it would be implausible (and in conflict with virtually all research on the question) to assert that public opinion does not also influence election outcomes. Any serious account of the policy impact of public opinion would likely test for both direct and indirect paths of influence, the latter through the influence of opinion on election outcomes (cf. Erikson et al. 2002). Paying serious attention to elections calls out for some further analysis of public opinion, where money and mass opinion are simultaneously taken into account (cf. Smith 2000).

The influence of elite attempts at framing and shaping mass opinion is, we would emphasize, very much a question amenable to sustained empirical research. It has been separately explored in political-psychological literatures on framing, priming, and persuasion (cf. Chong and Druckman 2007b). Likewise, the contentious claim that elections don’t matter has also been subjected to careful empirical work, and there is good evidence that even in the United States, Democratic control of the White House is associated with reduced inequality and greater income gains for moderate-income households (e.g., Bartels 2008; Kelly 2009). Such results are notably consistent with comparative scholarship on the political economy of partisan governance (Huber and Stephens 2001), where the degree to which Left, Right, or religious party coalitions dominate parliaments influences policy outputs and the distribution of income and services in a nation.