Civil Society Reconsidered: The Durable Nature and Community Structure of Collective Civic Action

This article develops a conceptual framework on civil society that shifts the dominant focus on individuals to collective action events— civic and protest alike—that bring people together in public to realize a common purpose. Analyzing over 4,000 events in the Chicago area from 1970 to 2000, the authors ﬁnd that while civic engagement is durable overall, “sixties-style” protest declines, and hybrid events that combine public claims making with civic forms of behavior— what they call “blended social action”—increase. Furthermore,dense social ties, group memberships, and neighborly exchange do not predict community variations in collective action. The density of nonproﬁt organizations matters instead, suggesting that declines in traditional social capital may not be as consequential for civic capacity as commonly thought. The reigning image of American civic life in both the popular

bership in the parent-teacher association, or slipping attendance at public meetings, Americans have been famously said to be "bowling alone" more than ever (Putnam 1995(Putnam , 2000)).
Of course, not all hold to this view.The most common dissent from the narrative of civic decline turns on a quarrel with the numbers.Using one of the same national surveys as Putnam does (2000, esp. chaps. 3, 4, 6, and the appendix), some observers have argued that individuals have not, in fact, declined in certain forms of institutional trust and traditional organizational memberships (Paxton 1999;Rotolo 1999).Others have argued that the organizational locus of civic engagement is what has changed-that Americans have turned to looser but still effective associations in the form of social support or self-help organizations (Wuthnow 1998;Ray 2002).Still other critics contend that American civic life, which was formerly organized around traditional membership-based voluntary associations, has been restructured around membership in advocacy organizations and other professional civic groups (Skocpol 2003(Skocpol , 2004)).
The positions of Putnam and his critics each have merit but tend to reinforce the way in which the debate on civil society has unfolded.Most of the data in dispute turn on trends in the individual-level backdrop to civic society-especially declines in group membership and social-psychological states of trust-rather than collective political action or public civic events.Why should we care about the prevalence of membership in the Elks Club, PTA, or any other group, unless it translates into collective civic action?Is it important that fewer people trust "generalized others" or attend regular meetings?Does this tell us about the capacity of collectivities to act? 2 The alarm bells that Putnam (1995) set off had little to do with bowling club memberships per se or the local Order of the Moose.Relying on Tocqueville's classic insights from the 1800s, political theorists and social critics have been most concerned with the negative consequences of civic disengagement for democratic capacity and the health of the American polity. 3From this concern, it follows that the civil society debate has been waged on the potentially misleading per-ceptions, memberships, and behaviors of individuals as opposed to truly social, or collective, action.
In an attempt to address this theoretical disconnect, our project turns its attention to robust civic action in the form of collective public events.Drawing on insights and lessons learned in the social movements literature, we gathered a sample of over 4,000 collective action events in the Chicago metropolitan area-both civic participation and protests-for selected years from 1970 to 2000.Through detailed coding of these events we constructed what we believe is a novel database that allows us to chart the variable nature and community structure of collective action in a major urban area during the period of rising concern over American civic decline.In shifting the focus from individuals to public events that bring two or more people together to realize a common purpose or specific claim, we find considerable temporal continuity in civic engagement.Namely, the volume of collective civic events is remarkably stable from 1970 to 2000, as are the rank orderings of public claims and forms of events.At the same time, we discover the growth of a previously overlooked phenomenon that combines protest with traditional civic behavior-what we call "blended social action."Our analysis also provides evidence that interindividual social ties and membership in traditional civic groups are not good predictors of collective action.This article thus reframes the civil society debate in terms of collective, rather than individual, action: we specifically argue that collective civic engagement appears to have changed rather than declined, with sources that are organizational rather than interpersonal in nature.

FROM PROTEST TO COLLECTIVE CIVIC PARTICIPATION
There are discernible biases in the literature on social movements that have undercut its contribution to the civic society debate.From our perspective, movement scholars have tended to privilege a rather narrow, stylized form of contention-featuring disruptive protest linked to broad national struggles waged by disadvantaged minorities-over far more numerous, if less visible, kinds of collective engagement (McAdam et al. 2005).One of the goals of this article is to grant greater empirical attention to "mundane" but no less important forms of collective civic action.
Fortunately, social movement scholarship offers important building blocks that can be used to transcend its traditional focus.One of the strengths of research in social movements is the decisive empirical shift from individual civic participation to a focus on collective action events, which logically, we argue, is essential to the underlying phenomenon at the heart of the civil society debate.A second contribution concerns the honing of a methodology consistent with an event-based focus, a point we elaborate further below.A third contribution, more theoretical in nature, concerns the central importance assigned by movement analysts to an understanding of the social processes that give rise to and help sustain collective mobilization and action (e.g., Emirbayer and Goodwin 1994;Gamson, Fireman, and Rytina 1982;Gould 1995;Tarrow 1998;McAdam [1982McAdam [ ] 1999McAdam [ , 2003;;Mische 2003), an intellectual move that rejects the idea that collective action results simply from the aggregation of individual civic behavior.
Recent work on the differential social organization of urban communities also bears on our theoretical understanding of the conditions under which collective capacity emerges and is built.Focusing on neighborhood residents, Sampson, Raudenbush, and Earls (1997) argue that the linkage of a working trust with shared expectations for intervening on behalf of the common good defines the spatial context of what they term collective efficacy.Just as self-efficacy is situated rather than global-one has selfefficacy relative to a particular task-a neighborhood's efficacy exists relative to specific tasks such as maintaining public order and providing local services.Whether garbage removal, the siting of a fire station, school improvements, or police response, a continuous stream of challenges faces residents of modern communities.No longer can such challenges be met by relying on strong ties among neighbors, for the evidence is clear that friends and social support networks are decreasingly organized in a parochial, local fashion (e.g., Fischer 1982).Moving away from a narrow focus on private ties and personal memberships, the concept of collective efficacy is meant to signify an emphasis on conjoint capability for action to achieve an intended effect, and hence an active sense of collective engagement on the part of residents to solve problems.Collective efficacy is best observed under conditions of challenge, reinforcing the idea that resolving conflict is an important part of civic engagement.
Social movements and collective efficacy theory thus share a common orienting framework-a focus on the mobilization of action for an intended purpose.We draw on this framework to argue for a direct focus on civic events that are collective in nature and that bring together members of the community.Events such as blood drives, community festivals, fund-raisers, and community watches against crime go straight to the heart of civic capacity.Like protest events, such civic events may seek to procure resources.However, civic events expressing communityoriented or collective interests typically do not represent a challenge to the existing system.Rather, many such events can be said to "celebrate community"-whether pancake breakfasts at the local fire hall, fundraisers for cultural causes, ethnic festivals, or neighborhood block parties.
The collective and spatial nature of such phenomena has been sur-prisingly overlooked in prior research.The social capital debate has turned on national trends in things like group membership, social ties, trust, leisure, voting, routine meeting attendance, giving, and television viewing (e.g., Putnam 2000, pp. 426-39), rather than directly on collective civic events in community context.The collective efficacy literature is sensitive to neighborhood context and the idea of collective events, but it, too, omits events and relies on survey questions about trust and expectations (Sampson et al. 1997, pp. 919-20).The social movements literature focuses on nonroutine events but eschews much concern for neighborhood variation and concentrates on explicit "protest" agendas such as the antiwar, civil rights, labor, and environmental movements.Although these agendas are each important in their own right, we seek to leverage a more general conception of collective civic action.We do so by borrowing different elements of each theoretical approach, first by replacing the individuallevel focus of most social capital arguments with an event-based approach to examining collective civic behavior.Next, we adopt the task-oriented and geographically bounded framework of collective efficacy.Finally, we expand the protest agenda of traditional social movement research by including an explicit focus on collective civic events.We accomplish these objectives by examining temporal and communitylevel variations in both collective civic action and protest in a major metropolitan region.Because this is the first such effort of which we are aware to directly measure collective civic action events and situate them simultaneously in time and space along with protest, our empirical assessment highlights basic trends, patterns, and predictors.We ask three broad sets of questions to orient our analysis: 1. What is the nature of collective civic behavior over time?Is it declining, stable, or possibly even increasing?What is the relationship of protest to civic engagement? 2. What are the claims or purposes of collective civic action and protest?
Are they aimed at producing public goods?Have collective claims changed over time, and if so, how? 3. Perhaps most important, what are the predictors of collective civic action at the community level?Does membership in traditional social capital organizations predict civic action events, or is institutional structure more important?
SOCIAL APPROPRIATION AND ITS ORGANIZATIONAL BASIS So far we have not made an explicit theoretical link between the social organization of communities and political action, but neither has much past work.Social capital theory, in particular, tends to gloss over the mechanisms by which bowling leagues and group membership lead to generalized civic capacity.Reasonable stories can be told about how social exchange within groups leads to political awareness, building community capacity, and political mobilization, but inferences are usually indirect, and the evidence sometimes negative (Kaufman 2003).By combining one strand of social movement research with recent work on collective efficacy, we derive testable hypotheses about the conditions that facilitate collective civic action.We propose that these conditions go beyond the kinds of social participation stressed to date.We start with one consistent finding from the social movement literature: episodes of contention tend to develop within established institutions or organizations (for a summary of this literature, see McCarthy 1996).Besides well-known studies that helped establish the fact (e.g., Orum 1972;Curtis and Zurcher 1973;Tilly, Tilly, and Tilly 1975;Morris 1984;McAdam 1999), a host of more recent works have confirmed it in new and creative ways.For example, in his work on the 1989 Chinese student movement, Zhao (1998) shows how the dense ecology of college campuses in Beijing served as the locus of initial mobilization.Glenn (2001), among others, documents the role that a network of independent theater companies played in the origins of the Civic Forum Movement in Czechoslovakia.Osa (1997) highlights the central structural importance of the Catholic Church to Solidarity and the larger dissident movement in Poland.
Consistent with these studies, proponents of the political process model of social movement emergence have long emphasized the role of established institutions or organizations in the onset of contention."Absent any such 'mobilizing structure,' incipient movements [are] thought to lack the capacity to act even if afforded the opportunity to do so" (McAdam 2003, p. 289).As straightforward and seemingly self-evident as this proposition is, it is worth reiterating that the emphasis is conceptually different from claims that turn on rates of individual participation in voluntary associations and related civic activities as emphasized in the social capital literature.The main difference of note is that movements and related protest events are not just aggregations of individual participants; rather, they are social products born of complex interactive dynamics played out within established social settings.Recently, movement theorists have sought to soften the structurally determinist tone of the earliest work in this tradition and to move instead to identify the contingent social and cultural processes that serve to transform established institutions into sites of emergent contention (McAdam 1999;McAdam, Tarrow, and Tilly 2001).Termed "social appropriation" by movement theorists, this transformative process is thought to depend on, among other factors, "attri-butions of threat or opportunity," the adoption of innovative action forms, and/or a shared sense of efficacy among group members.The last factor speaks directly to collective efficacy theory and, more generally, to recent work on community-level processes and dynamics (Sampson, Morenoff, and Gannon 2002).
Drawing on these different strands of work, we hypothesize that collective action events in the modern city are highly concentrated geographically and explained by systematic variations in community-level characteristics.Wuthnow (1998, p. 112), like Wilson (1987), argues that the new urban poverty is defined increasingly in terms of its geographic concentration, and therefore that "the character of civic involvement must be understood in terms of the social ecology of entire neighborhoods, rather than as an attribute of individuals or families alone."We agree, but note that restricting the focus to poor neighborhoods and concentrated poverty as the main contributor to declining mobilization is an analytically narrow approach.As noted, the dominant counterperspective on civic decline poses a different kind of thesis, whereby the density of local ties and memberships in local voluntary associations and organized groups constitute the main force generating civic action, a force that has systematically declined (e.g., Putnam 2000).In many ways, this idea harkens back to classic themes in the literature on urban voluntary associations and personal networks (e.g., Komarovsky 1946;Fischer 1982) as a source of collective capacity.
Rather than poverty or civic memberships, we emphasize the proposition that the capacity for sustained collective action is conditioned mainly by the presence of established institutions and organizations that may be appropriated in the service of emergent action.High rates of individual participation and dense personal ties may well be related to this kind of infrastructure, but conceptually, they are not the same thing.Imagine, for example, an "urban village" with intense participation by its residents in only one local organization, compared to another community with a lower prevalence of memberships spread across a multiplicity of institutions.Moreover, when membership and institutional density are coincident, it would seem that high rates of individual participation are an outgrowth of the existing organizational structure, not the reverse.In the same way, we see emergent collective action more as a product of extant institutions and organizations (or, more accurately, specified processes that take place within them) than of the levels of individual participation that attach to these structures.Adjusting for key confounders that may also bear on mobilization capacity, such as the concentration of economic resources and racial composition, we therefore compare the predictive power of aggregated civic memberships with a more institu-tionally based explanation rooted in the density of local organizational structures.

Emergence of Blended Forms of Action
Our integration of the social movements and urban community literature suggests a further twist to the social capital-civic society debate.The insight we glean is to view forms of protest not as inherently problematic, but as potentially part of building community.Protest, after all, is a form of collective action that Tocqueville (2000) himself might have admired for its democratic underpinnings.Putnam seems to agree: "Whether among gays marching in San Francisco or evangelicals praying on the Mall or, in an earlier era, autoworkers downing tools in Flint, the act of collective protest itself creates enduring bonds of solidarity" (2000, p. 153). 4 Along these lines, a recent analysis of democracy by the political scientist Eric Oliver claims that social conflict and civic engagement have a symbiotic relationship in a well-functioning democracy (2001, p. 202).
We thus take the step of conceptualizing a largely unrecognized but potentially transformative type of activity.Quite simply, we marry the metaphorical bowling league to civic action by examining events that combine traditional community "togetherness" (e.g., community festivals) with claims for social change.Consider, for example, an event reported in the Chicago Tribune under the headline: "Spearheads Playground Battle: Community Group Demands Playground Facilities for Youth at Park District."This article describes a collective event in which a community action group called Let's All Get Together and Work attended a meeting of the Park District to demand that a vacant lot be converted to a playground for neighborhood children.Or consider another event reported under the headline: "Neighbors' Library Plea Is Granted: Community Residents and Students Lobby for Temporary Library Facility."In this event, community residents and a group of middle-school students attended a board meeting of the public library to request that a temporary library site be found to service the community during the two years it would take to restore the neighborhood library destroyed by fire.
These two hybrid events represent examples of what we call "blended social action."This form of collective action blurs traditional boundaries by combining common types of civic participation, such as festivals or neighborhood association meetings, with a stated claim and an organized public event that seeks change.In other words, hybrid collective events typically combine protestlike "claims" for change with civil society "forms." 5 Consider that the successful library protest noted above originated in the context of a neighborhood association.Rather than assuming that membership in such a neighborhood form of organization triggers social action, we look instead to concrete public claims and collective action itself.Such blended events, neither wholly civic nor wholly protest in nature, provide a potential key to understanding the seeming paradox of decline in traditional civic memberships amidst durability in collective civic engagement.
Our hypothesis that blended forms of social action constitute an increasing form of collective civic action in the United States is further motivated by the fact that community-based nonprofit organizations have grown over time and have become increasingly embedded in the political process through the mediation of publicly funded services (Marwell 2004).An intriguing body of research also suggests the potential of local nonprofit organizations to foster collective action (Small 2004;Warren 2004) and to correct for market or government "failures" in the production of public goods (Hansmann 1987;Weisbrod 1988;Berry 2003). 6 In short, a focus on community-based organizational structures, especially nonprofit organizations, unifies our dual concern with temporal trends and cross-community variations.Temporal change is seen from the lens of a changing organizational form, and the concentration of the events themselves is theorized as rooted in the density of organizational infrastructures.

THE CHICAGO COLLECTIVE CIVIC PARTICIPATION STUDY (CCCP)
Our data consists of detailed information collected over four years on collective events occurring in the Chicago metropolitan area between 1970 and 2000.Chicago was chosen not only because of its status as a large 5 Maney and Oliver (2001, p. 148) argue something similar with respect to form: "Some protest messages are delivered through nonprotest forms such as ceremonies. . . .As the standard protest forms become legal and normative, they can carry nonprotest educational and awareness content."At the national level, Skocpol's (2004) recent work also points to the increasingly hybridized form of social action that stems from the professionalization of large-scale voluntary, nonprofit, and interest group organizations (see also Berry 2003). 6Marwell (2004) further argues that nonprofit community-based organizations are embedded in a network of political and economic exchange that can leverage the interests and voting power of their clients.Although she focuses on individual voting, from our perspective, this networking process in principle has the capacity to generate a variety of collective events, creating an "externality" of collective civic engagement.and important metropolis, but also because it afforded a unique opportunity to leverage an ongoing study of the city's neighborhoods (described below).We collected event data by extending a well-established methodology of social movement research-gathering and coding reports of protest events from newspaper archives (Earl et al. 2004).We selected the Chicago Tribune as the main newspaper of record because it has the largest circulation and is by consensus accounts the most influential newspaper in the Chicago area (see also Suttles 1990).
Collection of newspaper data has typically relied on keyword searches of newspaper indices (e.g., Lieberson and Silverman 1965;Gurr 1968;Olzak 1989).Another approach to event identification involves reading the entire newspaper, page by page, searching for candidate events (e.g., McAdam and Su 2002).Although more time consuming, we chose the latter approach because we sought to collect all forms of collective civic action, not just the highly visible protest events.Pretesting suggested that selecting events through a keyword search of the newspaper index failed to yield the broad types of collective action events of theoretical interest.To maximize resources and cover as long a temporal frame as possible, we sampled every third day of a given year rather than reading the newspaper every day of the week (see also Kriesi et al. 1995).With this strategy, no one day is privileged, and we were able to cover three decades' worth of events. 7The event identification process was identical for each day that events were collected.For example, there was no sampling within newspaper sections across days.This identical process ensured that data were collected from a representative sample of newspaper sections, days of the week, and days of the month across all the years of our study.

Collective Public Events
Our event identification relied on five primary criteria to distinguish articles that reported collective action events from other news stories, articles, and listings.First, prospective events had to be public.Second, they had to involve two or more individuals (though usually more).Third, we only collected events with a discrete time occurrence that could be identified within a two-week time window on either side of the newspaper date.8Fourth, we excluded routine political activity initiated by the state or formal political parties (e.g., political party dinners, speeches, and rallies).Such regular or ongoing political activity does not arise from emergent claims making or civic capacity of citizens, but develops instead from the initiative of political actors and professionals.Typical examples would be a speech by a mayoral candidate, a regular city council meeting, or a party-sponsored rally held for a state senator.9Furthermore, routine political lobbying by groups like Mothers Against Drunk Driving or the Sierra Club, while important, is continuous in nature and does not fit with our definition of discrete events of civic participation.Finally, we excluded profit-oriented events and regularly scheduled gatherings that are typical fare in any large city, such as professional sports games, live entertainment (e.g., rock concerts, theater), school swim meets, church services, university classes, and meetings of self-help groups such as Alcoholics Anonymous or weight-loss groups.10Self-help groups, classes, and regular public meetings initiated by professionals or as part of an organization's mandate-like routine political activity-do not provide what we argue is a direct indicator of collective civic action or capacity.11

Event Coding
Trained project personnel systematically read each page of the Tribune, collecting all articles for coding that contained a collective public event per the above criteria.Once event identification was completed, the articles were examined to exclude "double counting" of the same event.In addition, the articles were scanned to determine if there were multiple events reported in one article.We systematically coded each event within an article based on a coding scheme that was pretested on independent years of data.12Interrater reliability was established at both stages of the data collection process: event identification and event coding.Across persons and sites, interrater reliability averaged 90%. 13We coded nine distinct categories of information: 1. documentary information (e.g., date of article, event date), 2. event type (e.g., protest event, civic, or hybrid type), 3. frame of reference (e.g., national, state, city, neighborhood issue), 4. claims/purpose (specific nature and intent of event), 5. forms (e.g., a sit-in, march, community breakfast, fund-raiser), 6. location of event (by address, neighborhood, and/or municipality), 7. intensity (e.g., no. of participants, size, arrests, injuries, damages, deaths), 8. event initiator information (e.g., community location and organizational type), and 9. event target information.

Classification Rules
Based on our theoretical framework we classified each event as protest, civic, or hybrid.This classification was accomplished by examining in detail the claims/purpose of events, as well as their functional forms.We define claims, for this project, as a demand for either a change in society or an avowed desire to resist a proposed change.Forms, for this study, are defined as the manner in which action is undertaken by event initiators (e.g., rally, sit-in).
We defined protest events as any event "in which individuals collectively make a claim or express a grievance on behalf of a social movement organization or social category" (Uhrig and Van Dyke 1996).A protest event explicitly states a claim that includes a desire to bring about or prevent a change in policy or services (e.g., civil rights and gender equality).Protest events often take forms that are disruptive and contentious in nature, but they are not limited to these forms.Examples of protest forms include rallies, sit-ins, and marches, as well as petitioning, letterwriting campaigns, and class-action lawsuits.Protest events present a challenge to the existing social order and could entail violence (by protestors or responding civil authorities).However, protests can also be orderly and peaceful, as in symbolic displays or civil disobedience.
We draw the distinction between protest and civic events by noting differences in their forms and claims/purposes.Examples of civic forms include a rummage sale for a local church, a community breakfast, a local cleanup day, or a charity ball.These are the more traditional forms of civic life that Putnam claims have decreased in recent decades.Civic events, in contrast to protest, do not have claims as much as purposes: to celebrate the community (e.g., festivals), to procure resources (e.g., fundraisers), or to accomplish collective goals (e.g., cleanups, preservation).Civic events neither desire to bring about (or prevent) a change in policy, nor are they the expression of a specific grievance, as is often the case of protest events.We can also think of the difference between protest and civic claims/purposes as follows: protest events have explicit claims while civic events have implicit (or latent) purposes.
Perhaps most interesting to us theoretically are the hybrid events, which represent a blend of civic and protest forms of action.More precisely, such events typically combine civic forms with protest claims.They do so by exhibiting a clear protest claim and/or grievance; however, instead of a protest form (such as a march or rally), hybrid events exhibit a form that is typically associated with civic action.An example of this sort of event is a neighborhood art fair that doubles as a protest regarding current AIDS policy.In other words, claims of this sort grow out of traditional forms of civic participation, not from social movement organizations or from other organizations defined in terms of their claims.Figure 1 depicts our theoretical typology, or protest-civic-hybrid event framework, along with summary definitions and concrete examples.

Event Methodology and Issues Unique to Our Study
A large literature examines the methodological challenges of using newspaper event data, and many studies have investigated the reliability of such data.Several rules of thumb have emerged with respect to protest events: (1) national newspapers (e.g., the New York Times, Washington Post) have a geographical bias toward coverage of national versus local events, (2) newspapers are more likely to report events with greater size, duration, or conflict, and (3) compared to official records such as protest permits, newspapers underreport the number of events, but do report larger events in which conflict occurred in central locations or that enjoyed business sponsorship (see Snyder and Kelly 1977;McCarthy, McPhail, and Smith 1996;Oliver and Myers 1999;Myers and Caniglia 2004;Earl et al. 2004).Some researchers have attempted to model explicitly and subsequently "correct" for any perceived bias in newspaper data (Hug and Wisler 1998;Meuller 1997).Others have countered that a formal equation of sample selection bias cannot be estimated (Franzosi 1987, p. 8).
Our study addresses the challenges associated with the collection of newspaper data by using the local newspaper of widest coverage as our main source for events, employing a uniform sampling method across the time period of the study, and following a strict protocol for event identification and coding of information.Our study nonetheless introduces at least two new issues to the methodology of newspaper event coding.The first, anticipated by Oliver and Maney (2000), is that we expand event analyses to include not only less standard forms of protest but civic forms such as fund-raisers, petition drives, and celebrations-exactly the type of events that are systematically excluded from research on traditional social movements.By including both civic and hybrid events we are able to capture more fully the range of collective civic action.
A second issue raised by our study is that portions of the analysis rely on coding the location of events in geographic space.Although scholars have begun to champion the understudied spatial aspects of protest (e.g., Pile and Keith 1997;Miller 2000;Sewell 2001), previous research has not explicitly modeled community-level variations in collective action coded from newspapers.The question of geographical bias is related to the more traditional issue of temporal consistency in coverage.Does the Tribune cover certain neighborhoods more than others?Does the newspaper consistently report events over time?There is evidence that newspapers choose stories about protest based on their "newsworthiness" (McCarthy et al. 1996;Oliver and Maney 2000;Oliver and Myers 1999;Myers and Caniglia 2004), but many civic events are not that newsworthy in the traditional sense, and so the role of temporal and geographic bias is unknown.Based on interviews with Tribune staff, along with an extensive search for changes in newspaper policies toward civic reporting, we found no evidence or reason to believe that particular times or particular neighborhoods were systematically favored over others. 14Because of the importance of these issues, however, we devote a separate section of this article to independent validation, especially the question of differential coverage by the racial composition of readership and community.Our claim is not that our systematic sampling and coding scheme produced an exact count, without error, but rather that we have achieved a credible sample and representative picture of collective civic and protest events for metropolitan Chicago during the period in question.

TRENDS AND PATTERNS
Figure 2 displays basic trends in collective action events in metropolitan Chicago for the decade years of our study (1970, 1980, 1990, and 2000), disaggregated into the three broad categories of events.These categoriesthe definition and measurement of which we described in the previous section-are protest, civic, and hybrid collective events.
The first pattern worthy of note is that the general trend in protest and civic action does not accord with claims regarding the monotonic decline of social capital and civic life in the United States over the past three decades.While there appears to have been a sharp decline in protest and civic activity between 1970 and 1980, overall the picture for 1990 and 2000 is higher than in 1970, at or near the height of the New Left "protest cycle," suggesting a continuing vitality to civic/political life in metropolitan Chicago that is at odds with the sobering account of many social capital theorists. 15he second general observation we wish to make concerns the distribution of activity across our three event categories.Perhaps to the surprise of many steeped in the social movement tradition, protest-the "classic" type of collective action-comprises but a small segment in our data.In fact, civic events comprise nearly 80% of all events in Chicago over the years in question, and protest comprises just 15%.Located between the highly visible worlds of electoral politics (studied primarily by political scientists) and sociological studies of social movements, the vibrant arena of collective civic life lies largely invisible and yet accounts for the vast majority of our events.
A third pattern concerns the increase in hybrid forms of social action that combine traditional civic pursuits with claims making.Although the base prevalence of such events is low, it is notable that the percentage of hybrid events increased from under 4% of the total counts in 1970 to over 12% in 2000, a more than threefold increase.Figure 2 clearly shows that protest and hybrid are virtually dead even in 2000, whereas in 1970, protests were relatively much higher.
Finally, it is important to our argument that the basic pattern in these trends is not explained by variations in population change.Using census data on population size for metropolitan Chicago by decade, we calculated the rate of civic engagement per 100,000 to be 13, 10, 14, and 12 in 1970, 1980, 1990, and 2000, respectively, again a pattern of recovering stability by century's end.And the event rate of hybrid action more than doubled from one per 100,000 persons in 1970 to over two in 2000.The data thus seem reasonably clear that while the total volume and rate of collective civic action events is more or less stable from 1970 to 1990-2000, the proportion and rate of events that combine civic engagement with movementlike claims has increased.We have no way of knowing, but this merging or "blurring" of protest and traditional civic activity may portend a new kind of social action that will become increasingly important.
In table 1, we turn to the question of continuity and change in the claims that inform collective action events and the forms they take.Whereas figure 2 shows evidence of both continuity and change in the overall structure of action, a careful disaggregation by specific claims will allow us to take stock of changes in the extent to which claims are predominantly other regarding and aimed at the collective good.We are also able to assess whether there have been shifts over time in forms of collective action-for example, are community festivals and public meetings relatively less common now than they were 30 years ago?
Overall, the claims data in table 1 paint a picture of considerable continuity in collective action events for ostensibly public-goods purposes.The Spearman's rank-order correlation from 1970 to 2000 is extraordinarily high, at .994.Across the decades, the predominant focus is charity, which never drops lower than fifth in the claim rankings.Education also ranks consistently high, as do claims related to children, youth, and the arts.Interestingly, the kinds of claims traditionally associated with social movements, such as the environment, women's rights, civil rights, and housing, are not necessarily the top-ranked issues.With the exception of the environment, these movement-type claims do not even rank in the top ten.
What about forms of action?Although protest marches and rallies dominate our image of collective action, the social capital narrative of decline bemoans the loss of the simple public meeting.The data on forms in table 1 demonstrate that the public meeting is alive and well.Ranked second in 1970 and first in 2000, the public meeting as a form of collective action consistently rises to the top along with charity events.Overall the continuity in type of event form is considerable, with the Spearman rankorder correlation coming in at .987.
The community festival is an interesting standout, however, climbing steadily from ninth place in 1970 to the third-most-common form of col-lective action in 2000.In reading the many narratives of the events on which we collected data, it became clear that the community festival is a broad-based form that allows a plethora of activities to flourishincluding protest.One example of the versatility of the community festival form was a Fourth of July community celebration that included a special time set aside for a "human chain" composed of residents displaying signs, t-shirts, buttons, and the like, advertising the causes they support.This event would presumably raise awareness of the different issues supported by members of the community and perhaps attract others to support the same issues.Our data suggest that the community festival embodies many of the other-regarding qualities that Putnam (2000) argues are foundational.

COMMUNITY-LEVEL SOURCES OF COLLECTIVE CIVIC ENGAGEMENT
We now shift gears and consider the second of our key motivating questions.What predicts variation in collective action events across communities?We juxtapose competing perspectives on the social-organizational drivers of collective action events by capitalizing on a community survey in Chicago that allows us to construct direct measures of local social ties, organizational density, and membership in voluntary associations, which we then integrate with our newspaper event data.The Community Survey of the Project on Human Development in Chicago Neighborhoods (PHDCN) was conducted in 1995, when 8,782 Chicago residents were personally interviewed in their homes.The basic design for the survey had three stages: at stage 1, city blocks were sampled within neighborhood clusters; at stage 2, dwelling units were sampled within blocks; and at stage 3, one adult resident (18 years of age or older) was sampled within each selected dwelling unit.Abt Associates carried out the screening and data collection in cooperation with PHDCN, achieving an overall response rate of 75% (see Sampson et al. 1997).
For the purposes of this study we link survey responses to one of the 77 community areas in Chicago, which each average about 38,000 in population.Community areas are well known (e.g., the Loop, Lincoln Park, the Near North Side, Hyde Park), widely recognized politically, and often serve as boundaries for service delivery and allocation of resources (Taub, Taylor, and Dunham 1987;Suttles 1990).Theoretically, then, community areas are more appropriate as sites of collective action than census tracts that are much smaller and often characterized by arbitrarily defined boundaries.The mean sample sizes within community areas are also large (1 100), which yields the crucial advantage of our being able to reliably tap parameter variance between communities in institutional and social interactional processes.More specifically, the community-level reliabilities of all survey-based measures described below are .70or higher. 16 Organizations is a multi-item index that taps the total number of surveyreported local organizations and programs in the neighborhood-community newspaper, neighborhood watch, block group or tenant association, crime prevention program, alcohol/drug treatment program, family planning clinic, mental health center, youth center, after-school recreational programs for youth, counseling or mentoring services (e.g., Big Brothers), crisis intervention center, and mental health clinics for children.Although a single survey errs on the conservative side in capturing the volume of organizations, we believe that the systematic survey methodology, coupled with its broad range of coverage of community-based organizations that are typical in large cities (e.g., Warren 2004;Marwell 2004), produces a reliable and valid measure of between-community variations in the density of institutional structures that are available for social appropriation. 17 Civic membership in organizations was also recorded in the survey, allowing us to create a scale of community-level associations and memberships.Residents were asked whether they or a household member belonged to (1) religious organizations, (2) neighborhood watch programs, (3) block groups, tenant associations, or community councils, (4) business or civic groups such as the Masons, the Elks, or the Rotary Club, (5) ethnic or nationality clubs, and (6) political organizations such as a neighborhood ward group.These are just the sorts of voluntary associations and memberships emphasized by Putnam (2000); interestingly, they correlate positively but very modestly ( , not significant) with orga-r p .17 16 Community-level reliability is defined as: , calibrating the preci-2 S[t /(t ϩ j /nj)]/J 00 00 sion of measures, averaged across the set of J areas, as a function of (1) the sample size (n) in each of the j communities and (2) the proportion of the total variance that is between communities (t 00 ) relative to the amount that is within communities (j 2 ).
For further discussion of "ecometric" measurement strategies see Raudenbush and Sampson (1999). 17Our measure also comports with external evidence and independent "local knowledge" of the social organization of Chicago communities.On the latter, e.g., Hyde Park, the neighborhood surrounding the University of Chicago, is usually considered one of the most, if not the most, organizationally rich areas in the city.In our data, Hyde Park is in the ninety-ninth percentile of the organizations scale.Only one community-Beverly-scores (slightly) higher, and it, too, is known for its strong organizational base (see Taub et al. 1987, pp. 184-85).
nizational density, permitting us to assess their independent contributions to predicting collective action events. 18 We use the survey to assess two other key dimensions typically posited as sources of local social capital-density of friend/kinship networks and reciprocated exchange among neighbors.The measure of friend/kin ties is based on the combined average of two measures capturing the number of friends and relatives (each coded 0, 1-2, 3-5, 6-9, 10 or more) that respondents reported living in their neighborhood.Reciprocated exchange is measured by a five-item scale tapping the relative frequency of social exchange within the neighborhood.The items used were: "About how often do you and people in your neighborhood do favors for each other?By favors we mean such things as watching each other's children, helping with shopping, lending garden or house tools, and other small acts of kindness."(never, rarely, sometimes, or often)."How often do you and other people in this neighborhood visit in each other's homes or on the street?""How often do you and people in this neighborhood have parties or other get-togethers where other people in the neighborhood are invited?""When a neighbor is not at home, how often do you and other neighbors watch over their property?""How often do you and other people in the neighborhood ask each other advice about personal things such as child rearing or job openings?"Taken together, these items tap the sorts of personal networks and exchanges traditionally thought to underlie mobilization capacity.
As control variables, we include predictive (1995) survey measures of community-level socioeconomic status (SES, or scale of income, education, and occupational prestige), racial composition (% non-Hispanic black), and the aggregate level of violent victimization in the community.Population density (persons per kilometer) and population size from the 2000 census are also accounted for when modeling 2000 events.In assessing social-organizational sources of collective action events we thus adjust for key aspects of exposure (population), place stratification, resources, and social problems (see also Cohen and Dawson 1993).In separate validation analysis we examined a number of other specifications (e.g., % Hispanic, 18 We believe the systematic methodology provides a good barometer of relative between-community variations in the intensity of membership in civic/voluntary associations.To provide the broadest test of social-capital theory, our main results do not restrict membership to groups located in, or holding meetings in, the immediate neighborhood of the respondent because we are interested in variations at the larger community-area level (on average there are about 10 census tracts per community area).Nevertheless, total memberships correlated with the more restricted measure of local neighborhood memberships at ( ) across the 77 communities, suggesting r p .83 P !.001both measures tap the same propensity for civic membership.Not surprisingly, both measures produced identical results.measures of police-recorded crime, and both poverty and racial composition measured from the 2000 census), but the results were virtually identical, in large part because the 1995 survey accurately forecasts census 2000 demographics.For example, in evidence of the survey's representative design, the measure of % black in 1995 correlates with the independent 2000 census measure at ( ).Because of the rela-r p .97 P !.001tively small sample size ( ) and lack of statistical power in the ratio N p 77 of cases to predictors in multivariate analysis, we prefer a parsimonious model that is motivated by a small set of theoretically salient constructs.

Statistical Model for Event Counts
Three features of collective engagement in Chicago influenced our choice of an appropriate statistical model: (1) the phenomenon is a count of rare events, (2) we tested a Poisson model and found significant heterogeneity between communities in the latent event rates, and (3) the events are spatially clustered by community areas.Although we cannot identify the source of heterogeneity in our count data, we control for its presence through an overdispersed Poisson (or a negative binomial) model (Barron 1992, p. 189).Specifically, we model the event count Y i for a given community as sampled from an overdispersed Poisson distribution with mean , where n i is the population size in 100,000s of community i, and n l l is the latent or "true" collective action rate for community i per 100,000 people.We view the log-event rates as normally distributed across neighborhoods, using a hierarchical generalized linear model approach to set the natural log link equal to a mixed linear model that in-h p log (l ) i i cludes relevant community covariates and a random effect for each community to reflect unobserved heterogeneity.Furthermore, we account for the spatial clustering or nonindependence of events within communities by estimating robust (Huber-White estimator) standard errors.Our approach thus conforms to the features of the data through estimation of the log-event rate by a negative binomial count model (King 1988;Barron 1992) with adjustments for spatial clustering.

COMMUNITY-LEVEL RESULTS
We begin in table 2 with results for community-level variations in the location of collective action events.We examine two sets of predictors in a panel framework, with event counts in the year 2000 modeled as described above.One set of predictors includes background structural features in the form of SES, population density, race, and the violent crime rate in the community (the latter of which is highly correlated with a number of other compositional factors), all measured in 1995 and temporally prior to collective action events.The second set of predictors derives from our main theoretical model and includes friend/kin ties, reciprocated exchange, civic memberships, and organizational density, also measured in 1995 in the PHDCN community survey.The event counts are normalized by population size.
We proceed by estimating models for the event counts that conform to our theoretical categories of interest-collective civic engagement and hybrid or blended social action.The results in model 1 of table 2 show that organizational infrastructure, SES, low population density, and higher rates of violent crime predict increased rates of both traditional civic engagement and blended social action.Interestingly, however, the indicator of friend/kinship ties is associated with lower rates of blended social action, and reciprocated exchange is associated with lower collective civic engagement.Social and demographic features of communities, which are explicitly controlled in the analyses, cannot explain these patterns.
Our panel model resolves temporal order, but there remains a concern that we have left out important determinants of collective action correlated with organizational infrastructure and the other significant predictors in model 1 (i.e., spuriousness).To assess this concern, we control in model 2 for the volume of collective action events in 1990, a conservative procedure when assessing the estimated effects of social-organizational predictors measured at a later point in time.The reason is that the lagged outcome may partial out previous effects of the predictors, and, hence, a reduction of associations does not necessarily imply a lack of causal salience.
The results of this rather strict control are nonetheless revealing on two counts.First, the pattern of stability in collective action is surprisingly robust across types of events-a consistent and significant predictor of collective engagement intensity in 2000 is the intensity of engagement in 1990, whether traditional civic or the newer hybrid form.This finding serves as an indicator of the predictive validity of our data and as further evidence of the continuity in collective civic engagement.Second, some results change substantially after controlling for the baseline intensity of collective action; for example, in both cases the role of socioeconomic resources and population density diminishes completely, and neither violent crime nor reciprocated exchange now predict the most common event type, collective civic engagement.Once prior activity is adjusted, the only predictor of both traditional civic and hybrid social action is organizational density. 19These results support the idea that organizational infrastructure is the largest and most proximate community-level factor predicting the intensity of collective engagement of diverse types. 20 A critic might wonder whether we adequately tested our main thesis when the outcome, as in table 2, reflects the location of collective action events.Many of these events occurred in the Loop or downtown area and may not reflect the social capital or collective efficacy of the communities that actually initiated the events.For example, a group on the Far North Side of Chicago might regularly initiate protests or civic events in the Loop, and we would be remiss not to reflect on such initiating or generative capacity.We therefore collected additional data on the community location of the initiating organization or group responsible for each event, regardless of where the event itself took place.We were able to identify and geocode the initiating organization in approximately half of all events given the descriptions in the newspaper, compared to an 87% location rate for the events themselves.Assuming that the availability of initiator address information was distributed more or less randomly across areas-which we examined, finding no evidence to the contrarywe can subject our main results to a test of robustness.
We do so in table 3 by predicting events coded according to the community location of the initiating organization rather than by where the events took place (as in table 2).Because of the much lower number of initiator locations that could be geographically coded, and also the potential differential reporting of organization locations by type of event, we focus on the sum of hybrid, protest, and civic events.The model results are crisp-collective engagement is predicted to increase threefold with a one-standard-deviation increase in organizational density, controlling the 1990 rate (t-ratio of 5.80, ; model 2).By contrast, SES, pop-P !.01ulation density, and violent crime fail to predict the intensity of initiating activity once the prior rate of collective engagement is accounted for.Friend/kin ties are irrelevant as well, and reciprocated exchange is associated with a modestly lower prevalence of initiating activity.Local ties may promote the social control of crime (see Sampson et al. 1997) but not to increase civic and hybrid rates of collective action by a factor of 2.7 and 4.07, respectively, controlling for the lagged outcome and all other factors. 20For comparative purposes we also estimated count models of protest.The main predictor of 2000 protest events was lagged 1990 protest-in the full model none of the social-organizational or demographic/compositional predictors were significant at .In a trimmed model focusing on memberships versus organizations and holding P !.05constant lagged protest, organizational density was significant at , but mem-P !.10bership was not.Although this evidence weighs in favor of our theoretical interpretation, we are not inclined to place much weight on community variations in pure protest.From a civic society perspective the main action is elsewhere, or at least to be found in newer forms of hybrid action that draw on the protest narrative in creative but heretofore unappreciated ways.necessarily wider participation in collective events or an institutional readiness to address political problems.Wilson's (1987) concept of social isolation is consistent with this interpretation, where dense local ties and exchange foster a more parochial sense of community and attenuation of collective or larger community-based initiatives. 21 An implication of our results is that protest and traditional civic participation, although typically considered separately and by different scholarly fields, share similar correlates.We probe this intriguing finding more directly by mapping the community location of collective action events according to our threefold typology.Are blended protest/civic events lo-21 None of these results are sensitive to the statistical specification of the overdispersed Poisson model, confirmed by substituting the logged event rate and estimating an ordinary least squares regression for all models.We also checked for influential observations using standard regression diagnostics.Using logged rates, only model 2 in table 4 produced a large outlier-the Loop-which was eliminated and the model reestimated for this one case.The key results were replicated, and no major discrepancies were obtained-the t-ratios for the estimated effects of organizations in model 2 of table 3 (with log-1990 rate controlled) were 3.23 for collective civic events and 2.45 for hybrid events.For table 4, model 2, the t-ratio was 3.34.In no case was civic membership significant.cated in the same geographic areas as collective civic engagement?It is not obvious that this should be the case, even given our results thus far.To aid in further understanding, we map the density of collective action events in 2000 against the backdrop of the organizational infrastructure of communities, which our analysis has uncovered as an important independent predictor.
The results in figure 3 show a consistent and durable structural form to events, whether they are rare protests or more common civic engagements.Collective action events of all kinds, even hybrid action, tend to occur in a relatively narrow band of communities characterized by a dense organizational profile.Communities such as Hyde Park, Lakeview, and the Near North Side (not just the Loop), for example, are rich in organizational life and generate many of the social action events.That the organizational infrastructure of these areas matters above and beyond their population composition, SES, and density of personal ties was revealed in tables 2 and 3.The pattern in figure 3 therefore confirms not only the spatial overlap and concentration of diverse collective action events, but also their common origins in institutional infrastructure.
It is also of interest that communities such as Hyde Park and Uptown, while at or near the top in organizational infrastructure, sit much lower on the scale of traditional civic memberships, further supporting our contention that the prevalence of membership is distinct from the breadth or density of organizations.In fact, Uptown is in the bottom one-third of the civic membership distribution.Other communities (e.g., Avondale) are in or near the middle of the civic membership distribution but near the very bottom of organizational density.The lowest-scoring "civic" neighborhood in membership-Oakland-nonetheless fares rather well in the organizational representation picture (seventy-second percentile).These two dimensions of community social organization are conceptually distinct, and Chicago communities vary considerably in their joint profiles of civic memberships and organizational density.Apparently, some communities are able to achieve an active organizational life absent the dense civic memberships of yesteryear.One of the reasons may be that membership in groups such as the Elks, the Rotary Club, and neighborhood tenant groups are more aimed at instrumental goals tied to self-interest than to the promotion of social goods (see also Kaufman 2003).Another, perhaps more likely, reason is tied to the fact that corporate actors often provide the economic underpinnings of local organizations, a pathway distinct from civic membership (Taub et al. 1987, p. 184).

Frames of Engagement
Another critic might be concerned about our broad conceptualization of collective civic engagement, in the sense that so far we have not limited our attention to strictly neighborhood-specific frames of social action.The idea is that "many instances of protest are not rooted in either the neighborhoods where they occur or in the neighborhood where the sponsoring organization happens to be located.A peace vigil, an anti-nuclear demonstration, many sorts of protests at city hall over a wide range of issues . . .are not meaningfully treated as neighborhood events."22This is true, as by design we coded many events that imply a geographic and/or jurisdictional orientation that transcends the neighborhood in which they occur or even in which the initiating organization is located.Although our focus in this manuscript is not primarily protest, the general point can be logically extended to hybrid and traditional civic engagement.What is the frame of the event, and does it matter to our theoretical approach?
Our response is twofold.First, we are interested at the most fundamental level not in the nature of, or changes in, the geographic/jurisdictional orientation of events but in the variable capacity of neighborhoods to mobilize collective action.In this sense, all events are created equal.That is, regardless of their implied geographic orientation, our events speak to the capacity of local actors to generate emergent action-precisely the lacunae of the social movements field.Whether the goals are primarily local, citywide, national, or even international is less relevant to us than the local time, energy, and effort that must be expended in organizing and carrying out the event.Our data reveal the sources that help produce this social action.
Second, the question is ultimately an empirical one that we can address because of the nature of our coding scheme.For each event, we categorized the "frame" or purpose of the event that in most cases could be classified in terms of its geographical or substantive reach.This analysis revealed that "local neighborhood" frames (e.g., a street cleanup, a fund-raiser for a local school playground) and "city" frames of reference (e.g., a meeting to debate policy on deconcentrating poverty in the city's public housing system, a rally to protest distribution of city services) constitute 25% and 46% of the Chicago events, respectively. 23To gain further insight into the sources of these more local issues, we disaggregate model 2 in table 2 by neighborhood and city frames.Because of the markedly reduced sample size of events, we analyze the sum of hybrid and collective engagement events.Given the more localized framing of city and neighborhood events, we also examine civic memberships specific to local neighborhood associations (see n. 18).
The results in table 4 show remarkable similarity with the overall pattern obtained previously.Whether neighborhood-specific or a citywide framing of events, the density of local and interpersonally produced social capital, in the form of friend/kin ties, neighborly exchange, and civic memberships, proves ineffective (substituting total civic membership produced equivalent results).The structural controls are inconsistent and again largely insignificant, with the exception of a negative association of SES with neighborhood, but not citywide, events.By contrast, other than lagged 1990 events, the density of community-based organizations is the only consistent predictor of collective civic engagement in events with local or citywide frames.

CIVIC ACTION AND THE RACIAL DIVIDE
There are, of course, limitations to our analytic approach.We have addressed methodological challenges the best we could at each step along the way, but acknowledge that the uncharted territory of our data brings with it the potential for error.Franzosi (1987) has argued that researchers assessing newspaper data focus too narrowly on reliability, which we agree is often not the problem.Like Franzosi (1987, p. 5), we believe the bigger issue concerns validity-in our case, potential bias in what the newspaper measures tap across time and neighborhood.In particular, are better-off white neighborhoods more likely to be covered by the Tribune than innercity, disadvantaged areas?Were the trends we discovered driven by changes in Tribune editorial policy rather than by real change?To address these questions and the limitations of our main set of data, we undertook an additional research project with two goals in mind.One was to verify the broad temporal trend in collective civic events found in our Tribunebased analysis, and the second was to examine an independent source of neighborhood collective action events that is sensitive to the racial divides that seem so salient in Chicago.The latter is critical, as we are attempting to address nonspurious variations between communities in collective action.
Our analytic strategy was to maximize differences in newspaper structural organization, ideology, and editorial policies, all while maintaining a focus on coverage across multiple communities in the Chicago metropolitan area.Neighborhood newspapers were thus ruled out, as were purely suburban newspapers.The largest-circulation newspapers in Chi- cago other than the Tribune are the Chicago Sun-Times and the Chicago Defender.The Sun-Times is in many respects a smaller, more tabloidlike version of the Tribune, whereas the Defender is the nation's oldest African-American newspaper, not just a venerable and well-known institution in Chicago, but "the most important black metropolitan newspaper in America" (Grossman, Keating, and Reif 2004, p. 134).The Defender is explicitly designed to serve the needs of an African-American readership, but its coverage is still metropolitan-wide, giving us leverage to see if a different picture emerges of collective action events than in the Tribune.Chicago is a racially divided city in many ways, and in some quarters, the Tribune is seen as a conservative newspaper that reinforces this divide by pandering to elite interests.Yet in principle, all African-American communities (as distinct from events) should be covered in the Tribune, so the Defender offers a strategic advantage over the Sun-Times for examining biased coverage of collective action in the large and predominantly black communities of Chicago's West and South Sides.The key question is whether both newspapers are capturing the common capacity of communities to generate collective action events, initiating organizations, or both-even if contaminated by a racialized lens of reporting.
To ascertain if our main conclusions hold up to this kind of external validation we collected additional data from the Defender using the same basic criteria as we did for the Tribune, but with necessary differences in sampling and temporal coverage.The Daily Defender was not universally available on microfilm for all days and years in our libraries, so rather than sampling every third day we collected each weekend edition.To make the validation project manageable we also collected two types of data.One was a simple aggregate total of Defender-reported events in metro Chicago for 1970Chicago for , 1980Chicago for , and 1990 for all 12 months of each year.Second, we selected a sample of four months (January, March, July, and October) from the years 1970, 1980, 1990, and 2000 to be read in detail and coded.Because of the much smaller number of events examined in the Defender compared to the Tribune, along with the shorter four-month window for city events, we focus on total collective action (hybrid, protest, and civic) for both event and initiator location.
The first question is straightforward.What is the trend in collective action events over time?The number of collective action events as recorded by the Defender is remarkably consistent with the Tribune pattern over the three-decade period.The total number of Defender events in the Chicago metropolitan area in 1970, 1980, and 1990 are 548, 342, and 532, respectively.Thus we see a dip from 1970 to 1980, followed by a recovery in 1990.In fact, the overall tally is approximately the same in 1970 and 1990, very similar to the pattern for the Tribune (cf.fig.2).Because of its larger black readership base, the city of Chicago is the more critical test, however.Yet the patterns are identical for both raw counts and rates.Notably, the Defender event rate in Chicago per 100,000 residents is 5. 06, 3.46, 4.96, and 5.97 in 1970, 1980, 1990, and 2000, respectively.Despite the very different nature of the two newspapers, then, the data are consistent and paint a picture of the relative stability of collective civic engagement when we compare 1970 and 1990/2000, with 1980 the aberrant "down" year.
The second and perhaps more important issue addressed in this study is the distribution of collective action events across communities.To achieve consistency with our Tribune-based analysis, we geocoded all Defender events in Chicago to the city's 77 community areas.As expected, the Defender reports more events in African-American communities than does the Tribune.For example, traditional black communities like Douglas, Grand Boulevard, and Englewood rank in the top 10 communities for the Defender but not the Tribune.Communities like the Near West Side, the Loop, and the Near North Side nonetheless rank highly across both newspapers, indicating some continuity in geographic perspective.To measure this more precisely, we calculated Pearson correlation coefficients of association across communities for each decade.The correla-tions of Tribune and Defender counts were .84, .79, .85, and .67 for 1970, 1980, 1990, and 2000, respectively (all ), revealing a high level of P !.01overall agreement; the per capita rates were also all significantly positive.We learned earlier that event counts are highly skewed, however, with a handful of communities like the Loop dominating the overall count, and most communities recording no events.We thus calculated nonparametric Spearman correlations to detect rank-order agreement for each decade and for the aggregate total.Each was significant, with the aggregate coefficient (r) across the three decades at .40 ( ).We conclude that P !.01 while the Defender reports more events in the black community, more or less by its charge, the overall level of concurrent agreement with the Tribune is surprisingly consistent across time and space.
The question remains, however, as to how well the two newspapers match up at the event level.It could be the case that the correlations are positive because each newspaper is reporting more or less the same pool of events.Or, perhaps more interesting, it could be that the newspapers are tapping into the same capacity of neighborhoods to generate social action, but with a different mix of actual events underlying the pattern.To gain leverage on this issue we matched specific events, with the results confirming both scenarios but with a definite leaning toward the second.In particular, the Defender was much more likely than the Tribune to cover collective civic events in smaller African-American churches on the South and West Sides, whereas the Tribune's coverage in the black community ran more to larger middle-class churches and larger, politically visible events.Still, 18% of Defender events from 1970-2000 were matched exactly in the Tribune. 24e then examined a model in which variations in Defender-based collective events in 2000 were predicted by a function of % black, % poverty, and population density in 1990, along with the rate of total collective engagement as recorded in the Tribune.If racial or economic composition accounts for significant bias in reporting, this model should reveal a weak or insignificant predictive role for the Tribune.By contrast, if the two data sources tap common variance in the rate of collective action, then even when controlling for racial and economic composition of community (and hence readership), the Defender and Tribune should converge.In the resulting equation the coefficient for the 1990 Tribune yielded a t-ratio of 6.52 ( ); specifically, a one-standard-deviation increase in the Trib-P !.01une rate is associated with nearly half a standardized unit increase in the Defender event rate, adjusting for poverty and % black.
For a final set of critical tests we turn to this article's main hypothesis on the organizational sources of collective civic engagement.To assess the generality of findings we estimated an exact replication of the statistical model and variable specification in table 2, model 2, but with the Defender count of collective civic events in 2000 as the outcome.As expected, % black had a significant effect on predicting the event rate as reported in the Defender, whereas it did not for the Tribune.However, in all other substantive respects the results were replicated.The key result was that civic memberships, dense ties, and reciprocal exchange all failed to predict black civil society as measured by the Defender reports of collective action, whereas the estimate for organizational density was positive and significant (t-ratio p 3.49 and 2.19 for initiator and event location rates, respectively).Taking another perspective, we estimated a principal components analysis of the initiator and event location rates for both the Tribune and the Defender.Only one component was extracted from the data, and all loadings were over .80,yet another indicator that the two newspapers are tapping common variance in an underlying dimension of collective action capacity-whether initiator based or event location based.When we then estimated the predictors of this principal component, the now basic result obtained: in the replication of model 2 in table 2, the three traditional social capital measures were null.Organizational density, by contrast, directly predicted the common variance in collective civic action (coefficient p .59,SE p .26, ). P !.05In short, there is broad relative agreement on the ecological concentration and organizational sources of collective civic action when comparing the Tribune and Defender even when the manifest events differ.Given the divergent nature and organization of the two newspapers, we believe this is a sociologically interesting point, one that lends validity to our earlier results and the underlying argument.Note that our analytic strategy was divided into two fundamentally different questions-trends over time for the entire Chicago metropolitan area, and predictors of collective action events across communities of Chicago in 2000.The differential error structures in the design mean that the results taken as a whole are unlikely to have been generated by a common flaw; potential bias in counts over time  is very different than potential biases across communities at the same time (2000), and, moreover, the Defender is very different in nature than the Tribune.We now consider the general implications of these results, which although different in constituent respects, yield a consistent and, we believe, important picture.

IMPLICATIONS
The study of politics has changed dramatically over the past 30 years.In the 1960s and 70s, studies of voting and other forms of traditional civic participation were common, whereas research on social movements was a blip on the radar screen.Today the reverse is true.The study of social movements is a large and thriving subfield in the social sciences; the study of voting-with important exceptions like Brooks and Manza (1997) and Manza and Brooks (1997)-has waned, as has research on "low-visibility" civic events.Yet our data show that only 15% of collective action in Chicago during the study period is of the social movement-or protestvariety.Our results thus support a renewed emphasis on the civic sector in the United States, perhaps especially the nexus of voluntary and nonvoluntary organizations (e.g., Ayala 2000).
An even more direct challenge to social movement scholars would appear to come from the changing nature of movement activity itself.Although the literature tends to equate movement activity with disruptive protest in the context of loosely coordinated national movements waged by disadvantaged minorities, our study fails to confirm this stylized picture.Protest and collective civic engagement events tend to be overwhelmingly mundane, local, initiated by relatively advantaged segments of society, and devoid of major conflict.Conditioned to view movements as highly contentious and disruptive national struggles on behalf of the disadvantaged, it appears movement analysts have largely missed this interesting, mostly local, and far more moderate form of social action (see also McAdam et al. 2005).
Of further interest is the trend in our data toward ever-greater numbers of hybrid events, or what we call "blended social action."These are events that essentially combine traditional civic forms of action with movementstyle claims.Indeed, most of the overall rise in events since 1980 appears to be associated with this type of social action.Steeped in studies of the sixties-or "sixties-style"-struggles, movement scholars will need to do a better job of studying all manner of contemporary forms of action if they are to understand the changing face of contention in the United States.This conclusion accords well with the recent-although largely speculative-writings on the growing institutionalization of protest that has created a "movement society" in America (Meyer and Tarrow 1998;Tarrow 1998).To some observers (McAdam 2003), this changing form of protest has simultaneously routinized and tamed the social movement as a form of social change advocacy.Our findings on the sharp rise in "hybrid events," along with the nondisruptive nature of protest more generally (McAdam et al. 2005), are consistent with this view.

From Bowling to Social Action
Finally, what do our results suggest about Putnam's (1995Putnam's ( , 2000) ) work and, more generally, about the limits of the empirical and conceptual terrain on which much of the social capital-civil society debate has been waged?Most obviously, the trend in Chicago-area collective action over the past 30 years does not mirror the pattern adduced by Putnam and others from mostly individual-level trends.Instead of a steady decline over the period, we see a marked drop in collective action events between 1970 and 1980, followed by an impressive "recovery" in 1990, a pattern revealed in both the Chicago Tribune and the very different Defender.The Tribune totals in 1990 and 2000 also exceed those for 1970, which is all the more impressive given the high levels of protest activity-antiwar, women's movement, environment-we associate with 1970.We recognize that our research is limited to metro Chicago and thus cannot strictly bear on national trends (but see Rotolo 1999).Still, the question lingers: Why do the trends not match?
Our answer is clear: these trends don't have to match, and in some respects, the expectation that they should betrays a certain theoretical predilection.Indeed, to infer the health of civic society from group membership, regular meetings, and dense social ties builds on the proposition that collective action is aggregated out of large reservoirs of individuals who are predisposed to social and civic participation.We offer instead a conceptual framework grounded in the institutional origins of movement activity-the prospects for collective action are powerfully conditioned by the presence of established social settings within which emergent mobilization can occur.From this view collective civic action does not, in any simple sense, emerge or grow directly from individual memberships and dense social ties.Our results confirm that in accounting for community-level variations in both civic and hybrid collective action events, the organizational infrastructure of the neighborhood matters a great deal, and civic membership much less, if at all.
Our takeaway on Putnam (2000) is thus that we are looking at a different phenomenon that need not have the same correlates or trends, and that may even be more consequential for civil society than traditional forms of civic membership. 25There is hope in this perspective, for it suggests new organizational forms and strategies that can draw in indi-viduals for collective pursuits.In fact, Putnam's new work (Putnam and Feldstein 2003) might be read in this context, for we find it notable that his examples of successful community building all revolve around institutional or organizational structures, from a community newspaper in Mississippi, to branch libraries in Chicago, to a neighborhood organization in Boston, to a church in Southern California.
Another example is seen in Skogan and Hartnett's (1997) and Fung's (2004) descriptions of community policing in Chicago, where "beat meetings" bring together residents and the police to address local problems.Residents did not spontaneously begin attending beat meetings as a result of local social ties; rather the meetings had their origin in a structural initiative-by organizing, hosting, and supporting beat meetings, the police created an equal opportunity for involvement across the city.In disadvantaged neighborhoods with few indigenous secular organizations, residents typically have fewer opportunities for involvement in anticrime efforts (Skogan 1988), but when they do, residents are at least as likely to be involved as residents of better-endowed neighborhoods.In an interesting way, Chicago's policing program has evened the playing field with regard to opportunities to participate in community governance, with the greatest increase in collective participation seen by African-Americans.As Fung (2004) argues, designed institutions of "participatory democratic governance" such as these can spark citizen involvement that in turn generates innovative problem solving and public action.
On a general level, our interpretation also shares an affinity with Skocpol's (2003,2004) macrohistorical argument that American democracy has undergone a civic reorganization rather than simple decline.Paraphrasing Skocpol (2004, p. 4) for our case, community-based organizations concentrate resources, voice, and clout in collective civic engagementso we should care as much about the organizational as we do about the individual level of memberships and politics.This argument further implies that local initiatives will falter absent state or organizational backing, consistent with our results on the community-based organizational sources of collective action.

CONCLUSION
We believe we have identified a fruitful new theoretical approach and novel empirical strategy for tackling fundamental questions about the nature and changing structure of civic life in the modern city.Our results on collective action in Chicago over the past 30 years justify confidence in an event-focused approach that gives priority to variations across time and space in civic actions that bring people together for a common and specific purpose.In essence we are proposing an "extra-civic" or robust mechanism of collective engagement in the form of nonroutine events that are not initiated by the state or by political professionals, but by collectivities motivated by a particular issue to act together in public (i.e., civic) space.Our demonstration of the ecological concentration and organizational sources of these collective events, along with the related discovery of increasingly blended social action, further suggest not only that robust civic engagement inadequately explained by individual-level processes, but that it may be unfolding differently than in earlier eras.From this view, collective civic action has not declined; rather, what we have looked to as the traditional indicators of civic participation have declined while the nature of participation in collective events has changed.One might think of this as "changing continuity" in collective civic engagement.Combining civic forms with stated claims for social change may be a harbinger of the future, especially in an increasingly organizational society where time constraints on traditional social ties and individual memberships are severe.

Fig. 1 .
Fig. 1.-Theoretical classification and examples of three types of nonroutine collective action events

TABLE 4
Lagged Event-Count Prediction Models of Chicago Collective Civic and Hybrid Social Action in 2000 by Event Frame: 1995 PHDCN Survey of 77 Chicago Community Areas and CCCP Event-Count Locations, 1990 and 2000