A MOVEMENT SOCIETY EVALUATED: COLLECTIVE PROTEST IN THE UNITED STATES, *

Transcription

1 A MOVEMENT SOCIETY EVALUATED: COLLECTIVE PROTEST IN THE UNITED STATES, * Sarah A. Soule and Jennifer Earl In an attempt to make sense of shifts in the social movement sector and its relationship to conventional politics over the past forty years, some have proposed that Western nations are increasingly becoming movement societies. Accordingly, there are four key characteristics of the movement society: (1) over time expansion of protest; (2) over time diffusion of protest; (3) over time institutionalization of protest; and (4) over time institutionalization of state responses to protest. Using newly available data on over 19,000 protest events occurring in the U.S. between 1960 and 1986, we evaluate these four claims. Our findings suggest that movement society scholars are correct in some respects: the size of protest events has grown over time, the percentage of events at which at least one social movement organization is present has increased over time, the number of distinct protest claims has increased over time, and violent forms of protest policing have decreased over time. However, our findings call into question other movement society claims: the number of protests has declined over time, fewer organizations were present at each protest event over time, fewer new groups initiated events over time, fewer new claims emerged over time, and there was more significant activity by groups on the right in the 1960s and 1970s than expected. We suggest potential explanations for some of the negative findings in an attempt to refine the movement society arguments. Over the past decade or so, some social movement scholars have advanced the notion of a movement society (e.g., Meyer and Tarrow 1998; Rucht and Neidhardt 2002). Drawing on observations about the ways in which social movement actors make claims against their targets, these researchers have noted that the protest activities of social movements (at least in advanced industrial democracies) are becoming institutionalized as part of the standard repertoire of political participation. At the societal level, this means that the repertoire of political participation has expanded from electoral activities (e.g., voting, working on political campaigns) to include activities typically associated with social movements (e.g., protest, demonstrations). For example, Putman (2000: 165) notes that at the same time electoral participation has declined in the U.S., protest has become almost a standard operating procedure. 1 According to movement society (hereafter, MS) theorists, as the repertoire of political participation expanded to include protest and other such movement activities, a set of rules (e.g., police procedures for protest control) for managing the threats that this expansion engenders became institutionalized (McCarthy and McPhail 1998; della Porta, Fillieule, and Reiter 1998). In short, proponents of the MS hypothesis argue that protest and other activities typically used by social movements have become widespread and normal (McCarthy and McPhail 1998: 84) in advanced industrialized * This research was supported by grants from the National Science Foundation (SBR , SBR and SES ) and from the University of Arizona Vice-President for Research Small Grants Program. We would like to thank Doug McAdam, John McCarthy, and Susan Olzak for their role in collecting the protest data used for this project. We thank Ann Marie Condo, Lynn Greenhut, and Katrina Kimport for research assistance. Sarah A. Soule is Professor of Sociology at The University of Arizona, Tucson, AZ Jennifer Earl is Assistant Professor of Sociology at the University of California, Santa Barbara. Please direct correspondence to soule@u.arizona.edu Mobilization: An International Journal 10(3):

2 346 Mobilization democracies and that this has been accompanied by the institutionalization of the methods of managing social movement activities. The concept of the MS is intuitively appealing and seems to be entirely plausible based on early and descriptive accounts of the frequency of protest, of the various groups who use protest, and of the ways in which protest is managed by the state (see contributions in Meyer and Tarrow 1998). However, the social movement literature lacks an empirical assessment of the extent to which the U.S. has become an MS. 2 This is due to the fact that the comprehensive data needed to make such an assessment have not, until recently, been available. With the recent completion of a much-awaited dataset compiled by Doug McAdam, John McCarthy, Susan Olzak, and Sarah Soule, this article begins to examine the extent to which the U.S. has become an MS, joining Rucht s (1998) assessment in West Germany using a companion set of data. In what follows, we first review the key claims of the MS argument and then use these newly available data to evaluate these claims. THE MOVEMENT SOCIETY: KEY ARGUMENTS In the 1990s, scholars of social movements began to suggest that protest activities associated with social movements in advanced democracies were becoming institutionalized (Tarrow 1998; Meyer and Tarrow 1998). By institutionalized, it is generally meant that protest activities have developed a taken-for-granted nature and have become part of the conventional repertoire of activities that citizens in such contexts use to express opinions about their polities (and the activities and policies thereof). On the one hand, the institutionalization and legitimization of protest activity is heralded by scholars as useful or beneficial for social movements; that is, because such activities are becoming more modular, they can be employed by citizens in any number of different contexts (Tarrow 1998). Modular activities are powerful because they are inherently usable and understandable and, as such, become part of the standard repertoire of political participation. For an example, see Tarrow s (1998: 37-38) discussion of the boycott, which became a basic routine for rebellious colonists in the late 1700s and was later imported to England by anti-slavery associations. But, on the other hand, some scholars note that the institutionalization of protest activities does not benefit social movements and the probability of attaining their goals; if activities such as protest and demonstrations are to become part of the standard or conventional repertoire of political participation (in other words, if they become legitimated in the eyes of the state and/or the broader public), then they risk losing the very elements that gave them power in the past. Specifically, they risk losing their ability to elicit special response and impress their targets (Meyer and Tarrow 1998). Before scholars can adequately assess debates about whether the MS is positive or negative for social movements and the ultimate attainment of their stated goals, it is important to begin to understand the extent to which the U.S. (as well as other advanced industrial nations) has become a MS. To this end, it is important to understand what precisely characterizes a MS. We argue that there are four common characteristics of the MS as the concept has been used and discussed in the literature: the over time expansion of protest, the over time diffusion of protest, the over time institutionalization of protest, and the over time institutionalization of state responses. While few scholars address all of these dimensions, these are nonetheless a reasonable place to start. 3 The Expansion of Protest The first characteristic of a MS as it has been used in the literature is that it is characterized by fairly continuous protest (Meyer and Tarrow 1998; Rucht 1998), as contrasted with the more

3 A Movement Society Evaluated 347 sporadic nature of protest in the past. 4 This has been empirically examined in a variety of different ways. For example, some have argued that more U.S. citizens report that they have participated in the activities of social movements in recent years than was the case two or three decades ago (e.g., Dalton 2002; Putnam 2000; Verba, Schlozman, and Brady 1995). 5 Another strategy for examining arguments about the expansion of protest is to look at the sheer volume of protest events and the number of participants reported at such events, as Rucht (1998) does for West Germany, using protest event data gathered from newspapers. Rucht (1998) shows that, while there is yearly variation in the number of protest events reported in West Germany, there appears to be an upward trend between 1950 and 1992, especially in demonstrative types of protest events. Similarly, Rucht (1998) reports that while average protest size shrank in the 1970s (relative to the 1950s and 1960s), there has been a general upward trend in the number of participants reported at protest events in the 1980s and 1990s. The Diffusion of Protest A second characteristic of the MS is that it is marked by diffusion of protest to new types of constituents. Groups that at one point in time shied away from using protest are no longer reluctant to do so in movement societies. For example, consider the contrast offered by Meyer and Tarrow (1998) about the differences between the liberal, student activists in the Free Speech Movement in Berkeley in 1964 and the conservative, anti-tax activists in Rome in Their discussion highlights the evolution of protest, from a form of political expression that was once used primarily by left-leaning individuals interested in issues that benefit broad classes of people to the professionalized and organized form of political expression now also used by actors on the right end of the political spectrum (e.g., Blanchard 1994; Blee 2002; Koopmans 1997; Van Dyke and Soule 2002; McVeigh 1999; Meyer and Tarrow 1998). But the MS argument is not simply that right-wing groups are now more likely to use protest, rather the argument holds that many groups (regardless of left-right ideology) who once eschewed protest tactics now turn to these to press their claims. In addition to the diffusion of protest to new groups, others argue that the number of claims articulated by social movements over time has expanded (Rucht 1998). This is intricately related to the diffusion of protest to new groups, but is a different argument. According to this, over time there has been an expansion of different claims used by protesters, such that protesters now (as a result of the institutionalization of protest) make many more claims than they once did. No longer are protest claims all associated with large, national movements of the past (e.g., civil rights, peace, women s); instead, there has been an expansion in claims to include many different and unrelated issue areas. The Institutionalization of Protest Alongside the increase in the sheer volume of social movement activity that accompanies the expansion of protest is the argument that not all forms of social movement activity have increased over the past several decades. Instead, some have argued that less contentious forms of activity (e.g., petitioning) have increased in frequency, while more contentious forms (e.g., building takeovers) have not (Meyer and Tarrow 1998; Kreisi 1995; Everett 1992). For example, Dalton (2002: 65) examines World Values Survey and Political Action Study data to show that in the U.S., respondents report an increase in the use of the petition but no increase in the use of building takeovers between 1975 and As such, if protest has become more tame over time, then the level of violence and property damage associated with protest ought to have also declined over time. Potentially related to the moderation of protest tactics is the argument that there has been an increase in the number of organizations purporting to support the public interest through the

4 348 Mobilization use of extra-institutional means (Minkoff 1994; Walker 1991; Everett 1992). Many scholars argue that increasing organizational involvement in social movement activities is associated with the use of less confrontational tactics (Piven and Cloward 1977; Staggenborg 1988), thus any taming of tactical use over time is said to be caused by increasing organizational involvement. 6 The Institutionalization of Response Finally, those promoting the MS ideas have also argued that the way in which states deal with protesters has changed in rather dramatic ways over the past few decades. Consider the contrast offered by McCarthy and McPhail (1998) of the 1968 Democratic National Convention, where police treated protesters harshly, and the 1996 Democratic National Convention, where police and authorities opted instead to manage the threat posed by protesters by defining specific times and spaces in which protest could occur. McCarthy and McPhail (1998: 84) note: Citizen protest has now become a normal part of the political process, its messages seen as a legitimate supplement to voting, petitioning, and lobbying efforts to influence government policy and practice. At the same time, the recurring behavioral repertoires of both protesters and police, and their interactions with one another, have become institutionalized and therefore routinized, predictable, and, perhaps as a result, of diminishing impact. The observation is that the state and its agencies have over time devised ways to deal with protesters that assure that their First Amendment rights are not violated, while also assuring that order is maintained at most protest events. From the issuing of permits for protesters planning a demonstration, to the specification of potential protest spaces (e.g., public parks, sidewalks) as traditional public forums, which are more open for protest, the state has devised ways of facilitating order in protest (McCarthy and McPhail 1998: 87). As noted earlier, one key issue in the literature on the MS is that it has been impossible to adequately examine most of the four arguments specified above. In large part, this is due to the lack of data on protest frequency over time, as well as some key information on these protest events (e.g., organizational involvement, claims made or issues articulated, state response, and so on). The data described below allow just such an assessment. Following a description of our data, we move to the examination of specific hypotheses. DATA: THE DYNAMICS OF COLLECTIVE PROTEST, The collective action event data analyzed in this article were drawn from daily editions of the New York Times (NYT) between 1960 and For a particular event to be included in our dataset, it must meet three basic criteria. First, since we are interested in collective action, there must be more than one participant at the event. This means that we do not code individual acts of protest, such as self-immolation. Second, the participants must articulate some claim, whether this be a grievance against some target or an expression of support of some target. 7 Finally, the event must have happened in the public sphere or have been open to the public for us to include it in our dataset. Thus, we do not include private meetings by social movement actors that occur behind closed doors, but we do include events that happen in churches and other private organizations, as long as they were open to the public. 8 Events in our dataset include both insider tactics (e.g., petitions, letter-writing campaigns, lawsuits) and outsider tactics (e.g., protest, demonstrations, civil disobedience, violence). Insider tactics are situated within the institutional political system, while outsider tactics are situated externally to that system. We do not code such collective events as block parties, annual parades, and fundraising campaigns, however, as our focus is on events that articulate

5 A Movement Society Evaluated 349 some claim. 9 In all, we have collected data on 19,143 events occurring in the U.S. between 1960 and 1986 covered in the NYT. These events are associated with any claim or issue area articulated by protesters (in other words, these are not specific to a particular movement or set of movements). And, these events occurred all over the U.S., as we did not limit the events to those only occurring in a specific locale. Information on collective action events reported in newspapers and newsources is probably the most widely used form of data used by social movement scholars (Earl et al. 2004). As such, there is a vast literature that has attempted to uncover the possible biases of this data source (see review in Earl et al. 2004). Our research design has attempted to avoid some of the potential problems with this type of data by employing several innovations over prior research. Our first innovation is that we, unlike most past scholars, have not used the index to the NYT as a data source, nor have we used it to identify possible events. Instead, we have read each daily edition of the newspaper for our entire period ( ) and we have identified all collective action events reported therein. 10 While far more time consuming than using an index to identify candidate events, this technique allows us to find events that are not obvious from the title of the article (thus may not be caught by looking at the index). For example, an article that primarily discusses problems with housing conditions in some locale may mention a rent strike or other protest event related to such conditions. Also, this technique allows us to find events that are embedded in an article on a separate event. For example, an article on a series of school desegregation events may make mention of boycotts and other events that are also taking place, sometimes in other locales. Finally, this technique allows us to avoid any possible bias that might be introduced by changes in indexing procedures used by the newspaper (Earl et al. 2004). A second innovation over prior research (e.g., Kriesi et al. 1995; Wisler and Giugni 1999) is that we have not sampled from published newspapers within a given period. While a useful method for obtaining a rough indication of the landscape of protest events, sampling over time may introduce biases. For example, while coding only the Monday editions of newspapers may yield a high number of events that took place over weekends, certain types of events that are unlikely to take place on weekends may be underestimated. Thus, as noted above, we code events from the daily editions of NYT during this period. As should be clear from this discussion, the data collection and coding took place in two, separate stages. The first stage involved research assistants reading the daily editions of the newspaper and photocopying articles on all events located. The second stage involved content coding of these events. At both stages, periodic inter-assistant reliability checks were performed to assure consistency across our research assistants. Reliability estimates were consistently at, or above, 90% agreement. A MOVEMENT SOCIETY EVALUATED As noted earlier, a chief problem in the literature on the MS has been the lack of data to assess the key arguments discussed above. Using the data described above, it is possible to evaluate the four central claims proffered by the MS argument. There are three facets of the data that make it particularly appropriate for this task. First, it is critical to examine protest associated with all social movements active in the U.S. Because the data described above are not limited to a particular movement, or limited set of movements, or only movements on the left (or right) side of the political spectrum, we are able to use them to evaluate the claims of the MS argument that are not sensitive to a single movement or set of movements. That is, the MS arguments are fundamentally about the entirety of protest events at the societal level. Second, and related to this, the data we use are aggregated to the national level. While the locale of each protest event is

6 350 Mobilization recorded, for our purposes we examine national-level figures; this is appropriate, since the MS arguments are cast at the national level. Third, the data we use roughly cover the period that most scholars argue is critical to the development of the MS in advanced, industrialized nations. For example, Meyer and Tarrow (1998) talk of changes between the 1960s and 1990s, Rucht (1998) talks about changes between 1950 and 1992, Crozat (1998) examines changes between 1974 and 1990, McCarthy and McPhail (1998) look at 1968 to the 1990s, and Katzenstein (1998) contrasts the 1960s with the early 1990s. Thus, the time period that our data covers is appropriate for examining the MS claims. We proceed by examining each of the four core claims outlined above. The Expansion of Protest If it is true that social protest has moved from being a sporadic, if recurring feature of democratic politics, to a perpetual element in modern life (Meyer and Tarrow 1998: 4), then it ought to be the case that, over time, there is more protest. We can begin to evaluate this claim by examining the aggregate number of protest events that occurred in the U.S., associated with all issue areas, as reported in the daily editions of the NYT. Figure 1 shows little support for this hypothesis. While there certainly were peaks in the number of protest events over the period, there appears to be a downward trend in the amount of protest. In 1960, there were 839 events, a number that increased to 1052 in However, following the late 1960s, the trend began to decline (with the exception of ) to a low of 409 in However, if we consider the size of protest events, we find more support for the claim that protest has expanded over time. Like Rucht (1998), in figure 2 we also consider the number of participants reported at protest events, since it is important to discern whether the pattern of protest frequency shown in figure 1 is the same as, or different from, the trend in the size of these events. In figure 2, there are three categories of size of protest events: small (2-99 participants), moderate to large (100-9,999 participants), and mass (10,000 or more participants). Figure 2 shows that in the U.S. during this period there was an overall upward trend in the size of protest events, much as Everett (1992) has shown for events in Washington D.C. over roughly this same period. For example, in 1960, 77% of all events were classified by our coders as small, having between 2 and 99 participants, but by 1985, 56% of the events were this small. On the other end of the size spectrum, we see that very few events had 10,000 or more protesters present in 1960 (.84%), but in 1985 nearly 4% of events had more than 10,000 protesters present. 11 When we consider the moderate to large protest events (100-9,999 participants), we see that in 1960, 22% of the events fit this description. This figure increases to 55% in 1967 and then begins to decline slightly to 30% in 1982 (which is still higher than the percentage in 1960). Toward the end of our period, we see another increase in moderate to large-sized events, to 44% in Another way of examining the size of protest events over this period is to consider the mean number of protesters present at events each year. In our dataset, a specific number of protesters was reported in the news article for 9,743 of the events (or 51% of the 19,143 events coded). In the remaining 49% of the events, coders were asked to estimate the number of protesters based on verbal cues in the article (e.g., small, few, or handful of protesters were estimated to be in category 1, fewer than 10 protesters). Table 1 shows the average size of protest events, by each decade in our dataset. The figures in the second column use data on only the events in which a specific number of protesters was reported. The figures in the third column are based on all events; for events with estimates of size, we imputed a value by choosing the mid-point of the estimated size category for that particular event. Much like the story we told with figure 2, table 1 shows that over this period, the average size of protest events has increased. In the 1960s, the average size of protest events was 1,365, when we use information on only those events in which the number of protesters was reported.

7 A Movement Society Evaluated 351 Figure 1. Yearly Number of Protest Events, Percent Number of Events Year Figure 2. Percentage of Protest Events in Each of Three Size Categories, Small (2-99) Moderate to Large (100-9,999) Year 60 Mass (10,000 or more) Table 1. Mean Size of Protest Events, by Decade Mean Number of Participants in Events Year When Number Was Reported (n = 9743) Mean Number of Participants in All Events a (n = 19,143) ,365 1, ,771 1, b 9,330 4,350 a Mid-point of estimated categories used when exact number was not reported. b Decade is not complete for the 1980s

8 352 Mobilization This figure increased to 2,771 in the 1970s, and to 9,330 for the six years on which we have data in the 1980s. Thus, when thinking about the expansion of protest over time in the U.S., there are fewer protest events over this period, but more recent events have drawn more participants. This may, at least in part, be due to technological advances (e.g., expansion of telephone service, direct mail, and so on) that make contacting potential participants easier, since many scholars note that being asked to protest is one of the most significant predictors of actually protesting (Verba et al. 1995; Klandermans 2004; Schussman and Soule forthcoming). If this is the case, then we ought to see the expansion in protest size continue into the 1990s and 2000s, with the growth of the internet as a means of communicating with potential activists. 12 This finding also resonates with what individual-level surveys have shown; that is, over time the percentage of respondents reporting that they have participated in a protest event has increased (Dalton 2002). It is important to note, however, that these findings are different from those that Rucht (1998) reports for West Germany. In that country, protest events in the 1950s were relatively rare, but when they occurred, they attracted generally far more people than in later periods (Rucht 1998: 38). In the 1970s in West Germany, there were more protests but they tended to be much smaller than in the 1950s. And, in the 1980s and 1990s, there appears to be an increase in both the frequency and size of protest events. The discrepancies between the U.S. and West Germany may be due to different political opportunity structures in different national contexts and they offer an important corrective to the MS hypothesis; that is, the extent to which protest has expanded over the past several decades may vary considerably by national context and should not be assumed to have occurred uniformly in all advanced, industrialized democracies. The Institutionalization of Protest Another hypothesis implied by the MS argument is that, over time, there has been a shift in tactics from those that occur primarily outside of institutional channels to those that occur within institutionalized channels. One way to examine this is to look at the proportion of events using different types of tactics over time. To do this, we created two different variables designed to differentiate between insider and outsider protest tactics. An event was coded as using insider tactics if it used any of the following tactics: lawsuits, legal maneuvers, press conferences, press statements, setting up information tables, petition gathering, lobbying, or letterwriting campaigns. An event was coded as using outsider tactics if it used any of the following tactics: rallies, demonstrations, marches, vigils, pickets, civil disobedience, physical or verbal attacks, riots, melees, or boycotts. Figure 3 shows how each of these very broad classes of tactics varies over time. At the most basic level, it appears that the percentage of events using outsider tactics has decreased over time, while the percentage of events using insider tactics has increased over time, offering support to the MS argument. In 1960, about 79% of events used outsider tactics while about 18% used insider tactics. 13 These figures can be contrasted with those from 1986, where 64% of events involved the use of outsider tactics, and just over 28% involved the use of insider tactics. However, there is some variation between these years, with the use of insider tactics surpassing the use of outsider tactics in 1974 (48% of events used insider tactics, while 42% used outsider tactics in that year). Following 1974, outsider tactical use begins to climb, but it does not reach the levels used in the 1960s. Similarly, following 1974, insider tactical use begins to decline, but does not fall below the levels used in the 1960s. Related to tactical use is property damage and violence at protest events (Rucht 1998). If it is true that protest has become more institutionalized over time, we should see less property damage and less violence at protest events. In the dataset, property damage is measured by whether or not the NYT article reported any property damage at the event. Violence is measured by whether or not the NYT article reported that protesters used violence at the event. Figure 4 shows that there has been an overall downward trend in both violence and property damage at protest events following 1967, per the prediction of the MS hypothesis. Following a peak in

9 A Movement Society Evaluated 353 Figure 3. Percentage of Events Using Insider and Outsider Tactics, Outsider Tactic 60 Percent Insider Tactic 0 Year Year Figure 4. Percentage of Events with Protester Violence or Property Damage, With Violence 25 Percent Year With Property Damage violence and property damage in 1967 when 33% of protest events involved violence by protesters and 21% of events involved property damage, there was a fairly steady decline in both of these such that by 1986, less than 10% of the events involved protester violence and less than 2% of the events involved property damage. However, it is important to note that these percentages for 1986 are less strikingly different from those for 1960, when 12% of the events involved protester violence and just under 7% involved property damage. It is clear, then, that violence and property damage have declined in the post 1967 era, but that this may reflect more of a return to the early-1960s status quo than a decline over the entire period. Either way we conceptualize these findings, they should be contrasted with those reported by Rucht (1998:39) for West Germany. While violence and property damage in that country

10 354 Mobilization have always been fairly rare, they have both increased in the 1980s and 1990s. Rucht attributes this rise to a general radicalization of protest, with the extraordinary rise in violence in the 1990s associated with the wave of right-wing radicalism in Germany (Rucht 1998; Koopmans 1997). Some researchers have suggested that decreases in violence, property damage, and the prevalence of outsider tactics may be attributable to the increasingly important role organizations may hold in the mobilization of individuals over the course of this period (McCarthy and Zald 1973; Rucht 1998). That is, some scholars have argued that organizational involvement in social movements may reduce the contentiousness of protest tactics (Piven and Cloward 1979; Staggenborg 1988). Thus, it is important to examine not only shifts in insider and outsider tactics and the use of property damage and violence, but also possible changes in organizational involvement in the activities of social movements over time. If organizational involvement increased in this period, that would support the possible connection between formal organization and less disruptive protest. On the other hand, if organizations were less involved in organizing protest during this period, even while tactics became less radicalized, then organizations may play less of a role in institutionalization and deradicalization than previously anticipated. To fully understand this relationship, we turn now to an examination of organizational involvement in protest in the U.S. during this period. Rucht (1998: 42) reports that in West Germany, there has been an overall increase in the average number of groups involved in protest events, from 1.4 groups per protest event in 1950 to 9.6 in the 1980s, perhaps as a result of an increase in coordination and coalition formation during these later years. In the U.S., however, we find just the opposite. In 1960, the mean number of social movement organizations at an event was 1.8. This increased to close to 2.5 social movement organizations in 1972, but then began to decline to under 1.5 organizations in the mid-1980s. Counter to the MS hypothesis, then, we find that over time there are fewer organizations present at protest events in the U.S. We also fail to find the magnitude of change that Rucht found. He found an 8.1 increase in the mean number of groups per protest event, whereas our data show a much smaller change. Another way to look at organizational involvement at protest events is to look at the percentage of events that had at least one social movement organization present each year. Figure 5 shows that in the period, there appears to be an overall upward trend in the number of events that had at least one organization present. In 1960, about a quarter of all events had at least one organization present and by 1986 over one third of events did. But, there is a great deal of variation across the years in this time series, with 1960 being the low point, but with several years (including 1961) having over 50% of the events with some organizational involvement. In sum, then, over time there is an increase in the percentage of events that had at least one organization present, however there is a decrease in the average number of organizations at protest events. How can we reconcile these seemingly divergent findings regarding movement organizational involvement at protest events? It is possible that there has been consolidation in the social movement organization sector over time. That is, it is possible that movement organiations have grown and merged, creating fewer, but larger and more active, social movement organizations. 14 This would explain why it appears that a higher proportion of events have organizations present, but that there are fewer organizations present at each event. Alternatively, it could be that cooperation between movement organizations has actually decreased over time, perhaps as a result of increasing competition between a greater number of social movement organizations. It is beyond the scope of this article to try to understand the reasons for these divergent findings, but future research should certainly investigate why this pattern exists.

11 A Movement Society Evaluated 355 Figure 5. Percentage of Events with SMOs Present, Percent Year Figure 6. Number of Unique or Distinct Claims Articulated Each Year, Number of Claims Year The Diffusion of Protest Meyer and Tarrow (1998: 4) argue that over time, protest will be used to represent a wider range of claims than ever before. Thus, we hypothesize that the total number of claims articlated at protest events in each year will increase over this period. To examine this hypothesis, we count the number of distinct claims articulated across all protest events in a given year. For example, if there was at least one pro-choice protest event in a given year, the count of unique claims increases by 1. If there are both pro-choice and anti-abortion protest events in a given year, the count increases by 2. This count is not sensitive to the volume of protest events assocated with a given claim or issue area. Rather, it is simply the count of different or unique claims used each year. Figure 6 shows that, over time, there has clearly been an expansion of claims used by protesters. In 1960, there were 89 different claims articulated by protesters. This figure peaked in 1977, with 165 distinct claims articulated and, in 1986, there were 104 different claims articulated.

12 356 Mobilization Table 2. New Claims Articulated at Protest Events, Percentage of New Claims Used in Subsequent Five Years Percentage of New Claims Not Used Again During Observation Period Number of New Years Claims % 15% % 21% % 39% % 68% a -- a Note: a Since 1986 is the endpoint of our time-series, figures are not available. Another way to think about the expansion of protester claims is to consider the emergence of new claims, rather than the use of different or unique claims (as shown in figure 6). Table 2 presents the number of new claims used every five years. When looking at table 2, it is important to note that the figures for the first five-year block ( ) are upwardly biased; since we begin our observation in 1960, all claims are new in 1960, inflating the number of new claims in this first period. The data in table 2 show, first, that the number of new claims used in each of these fiveyear blocks decreases over time. While, as noted above, the figures from are artificially high, since all claims used in 1960 are new (at least to our data set), there is still a steady decline following this period (column 2 of table 2). As well, the data also show a decrease over time in the percentage of new claims used in the period immediately following a particular period (column 3). Of the 86 new claims used in , 54 (or 62%) of these were observed in the period. And, of the 61 new claims used in , 33 (or 51%) of these were observed in the period. Finally, the data show that over time, a higher percentage of new claims become obsolete; that is, a higher percentage of new claims are not used again during this period of observation. 15 These findings offer little support for the argument that there has been an expansion in new claims made by protesters. Instead, the number of new claims used decreases over time (even though, as shown in figure 6, the overall number of different claims used each year increases). And, these data show that of the new claims used by protesters, over time more and more become obsolete. However, these findings are not necessarily inconsistent with the MS arguments, even though at first glance they may appear to be so. One interpretation of these findings is that in the more recent period, protest has become more localized in nature and may be actually driven by issues and problems specific to a particular locale (see Binder 2002: 11, for a similar argument). If it is true that protest has become more local and more issue-specific in advanced, industrial societies, then the lack of staying power of new claims (both over the entire period of observation and in the immediately subsequent five-year period) adds an important refinement to MS arguments about the expansion of claims. That is, these findings indicate that the claims articulated are likely different in form over this period and are less likely to be used by other groups because they are specific to a particular locale and set of issues with which citizens therein are grappling. But, what can our data tell us about specific claims used by protesters? Many have argued that protest is traditionally a strategy of political participation used mainly by individuals who are on the left-side of the political ideology spectrum to articulate left-wing goals (Dalton 2002). However, if Meyer and Tarrow (1998) are correct, it ought to be the case that over time we see more and more protest articulating right-wing claims. To investigate this, we code the following events as articulating a right wing claim: anti-feminist, pro-war, pro-apartheid, anti-protection of environment, anti-civil rights of race/ethnic groups, anti-gay/lesbian rights, anti-disability rights, anti-farm workers rights, anti-abortion, white-supremacist, and anti-animal rights. We then examine the percentage of all events, in each year, that articulated one or more right-wing

13 A Movement Society Evaluated 357 Figure 7. Percentage of Events that Articulate a Right-Wing Claim, Percent Year claims. These results are shown in figure 7. As this figure shows, there is little support for the claim that, over time, a greater percentage of events articulate a right-wing claim, although there does appear to be an upswing from the mid-1980s on. In 1960, over 18% of events articulated a claim on the right of the political spectrum a figure that declines to less than 15% in It is possible that those who have claimed that there has been an increase in the level of right-wing protest in the U.S. have based this observation on an increase in size of right-wing events. While there is a lot of variation by year in the size of right-wing protest events, there is some support for this interpretation. In the period, just over 31% of right-wing events drew between 10 and 99 people, while in the period, 28% did. On the other end of the size spectrum, in the period, 1.4% of right-wing events drew 10,000 or more people, while in % of right-wing events were this massive. Thus part of the reason why observers may believe that right-wing protest is on the rise is that, despite the fact that there is not a not a proportionally greater amount of such protest, there is some evidence of an increase in the size of such events. We note, though, that any increase in size of a particular type of event (e.g., right-wing) must be considered against the backdrop of the general increase in protest size shown in figure 2. That is, the increase in the size of right-wing events may be part of a general trend in the increase in size of protest events in the U.S. Turning from the issue of the expansion and diffusion of protest claims to the expansion of different groups using protest, Meyer and Tarrow (1998:4) argue that one of the key facets of a MS is that over time the use of protest should spread to groups who formerly eschewed such tactics in favor of more conventional types of political activity. Since we have information on characteristics of the groups responsible for organizing each protest event, we are able to count the number of new groups organizing protest events as a way of evaluating the extent to which there has been an increase in the number of new groups using protest over this period. 16 The second column of table 3 shows the number of new initiating groups active in each of several five-year blocks. Note that, contrary to the MS arguments, these figures decline over time. As was the case in our earlier analysis of new claims, the values of new initiating groups for the period are artificially high, since every initiating group in 1960 is new (at least to our data set). Nonetheless, there is still a steady decline following this first five-year period. As well, the data show a decrease over time in the percentage of new initiating groups active during the period immediately subsequent to a particular period (column 3). For example, of the 87 new initiating groups active in , 53 (or 61%) were also active in the 1970-

14 358 Mobilization Table 3. New Initiating Groups at Protest Events, Years Number of New Groups Percentage of New Groups Active in Subsequent Five Years Percentage of New Groups Not Appearing Again During the Observation Period % 13% % 28% % 46% % 87% a -- a Note: a Since 1986 is the endpoint of our time-series, figures are not available period. Contrast this with the fact that of the 70 new groups active in , only 18 (or 26%) of these were observed in the period. Finally, the data show that over time, more and more initiating groups become inactive (following some activism) during this period of observation. 17 The Institutionalization of Response The final claim made by the MS argument is that the state will institutionalize a set of responses to protest. One way to measure the extent to which response has become institutionalized is to examine how police respond to protest. We might expect that, over time, police respond differently to events, especially those that they perceive as threatening (Earl 2003). Past research (McAdam 1982; Davenport 2000; Earl, Soule, and McCarthy 2003) has shown that certain events are more threatening to elites: Large events and events in which protesters use confrontational tactics have been found to be more threatening, thus to elicit a higher degree of police response. Table 4 presents information about whether or not police are more likely to attend such threatening events over time and, if so, whether or not they are more likely to take action at such events. In particular, we examine three different types of threatening events: moderate to large events (100-9,999 people), mass events (10,000 or more), and events that used confrontational tactics. 18 Table 4. Police Presence and Action at Threatening Protest Events, (percent) Police Presence at Event % If Present, Police Took Action at Event % Events Between 100-9,999 People Events with 10,000 or More People Confrontational Events

15 A Movement Society Evaluated 359 Table 4 shows that with respect to police presence at threatening protest events, there appears to be an overall downward trend. That is, in all of these categories of threatening events, police are less likely to attend the events in 1985 than they were in For example, in % of confrontational events drew police presence; in 1985, only 44% of such events drew police. However, this is coupled with an overall upward trend in the probability that once at an event, police will take action; for all of these categories of protest events, police are more likely to take action in 1985, given that they attended a protest event, than they were in For example, at confrontational events in 1960 in which police showed up, 83% drew some sort of police response (beyond just presence at the event). In 1986, over 89% of such events drew some sort of police response (beyond just presence at the event). This suggests an interesting shift in how police treat protest events that are likely to be threatening. First, in the 1980s, they attended fewer of these threatening events than they did in the 1960s. But, of events with police present, in the 1980s a higher percentage of these drew some sort of police action. These findings suggest the importance of paying attention to the level of threat posed by protesters at events, rather than merely police presence at events (Earl et al. 2003). It could be that police in the latter part of the period are better able to assess the level of threat and are thus able to better choose those events that they will attend. In other words, these findings suggest better filtering mechanisms on the part of the police; they now attend only those events that are likely to be especially threatening, thus explaining why a greater percentage of those attended are met with some sort of response (beyond just attendance). But what kinds of police action became more prevalent over this period? Did police respond in a kinder and gentler manner to these threatening events? According to the MS argument, over time we should see that police respond in a more controlled manner to protest in general, we ought to see a decline in the use of force-based policing of protest. This should be especially true for those events that are more threatening (moderate to large or mass events or those using confrontational tactics). To examine this claim, it makes sense to look at the use of force, violence, and arrests at these protest events. We define police use of violence as whether or not physical force was used by police (e.g., hitting, shoving, etc.). We define police use of force as whether or not police used equipment to control protesters (e.g., tear gas, guns, etc.). We also coded whether or not police arrested protesters. Table 5 shows how these police Table 5. The Use of Force, Violence, and Arrests at Threatening Protest Events (percent) % of Events in Which Police Used Force % of Events in Which Police Used Violence % of Events in Which Police Made Arrests Events Between 100-9,999 People Events with 10,000 or More People Confrontational Events

16 360 Mobilization strategies change over time. The figures presented in table 5 largely confirm the MS hypothesis. That is, in all types of threatening events (moderate to large events, mass events, and events using confrontational tactics), there is a downward trend in the percentage of events in which police used violence and force. However, the tendency to use arrests at such threatening events has remained largely the same over this period. Thus, it appears that there has been a decline in force-based policing associated with the protection of rights of citizens, per the MS argument. SUMMARY AND CONCLUSIONS Over the past several years, social movement scholars have advanced the argument that protest in most advanced, industrial democracies has become part of the standard repertoire of political participation. Building on the early insights of Barnes and Kaase (1979), these scholars have argued that protest is no longer only used by individuals situated outside of the institutional, political system and, instead, is used by a variety of different groups, articulating a variety of different claims. Moreover, they have argued that protest is more organized and institutionalized than it once was and that, in turn, protest tactics have become somewhat less radical than in the past. Finally, these scholars posit the state and its agents have developed their own repertoire for how to manage protest, a repertoire that has become more tolerant of protesters and which is designed to protect the rights of protesters. This article has evaluated the central claims of these arguments, which together comprise the core of the MS literature, using newly available data on public, collective protest in the U.S. during the period spanning The findings presented above are somewhat mixed and deserve some summary and discussion. First, contrary to the MS arguments, we find that there are fewer protest events in the latter years of this period than there were at the beginning, but that these later events tend to draw more people (congruent with the MS argument). In other words, our findings suggest that there may be greater coordination of protest toward the end of our period, which would explain the greater number of participants that these more recent events drew. However, one of the other findings that we present in this paper suggests that this may not be the only explanation for these findings: that is, when we examine the average number of social movement organizations present at protest events, there were fewer organizations at more recent events than there were at earlier events. Nonetheless, it could be the case that there has been a consolidation in the social movement organization sector, such that there are now fewer, but larger, social movement organizations in the U.S. In fact, our finding that a greater proportion of events had some organizational involvement (despite the fact that events have fewer organizations participating) lends some credence to this interpretation. That is, it could be the case that there are now larger social movement organizations that are more actively organizing protest events and which do a better job of mobilizing more people. If this is the case, then our seemingly contradictory findings regarding protest size (increasing), protest frequency (decreasing), numbers of organizations involved per protest event (decreasing), and percentage of events with some organizational involvement (increasing) make sense and, in fact, support the core claims of the MS argument. Second, with respect to the institutionalization of protest tactics, we find support for the MS prediction that protest has become more moderate over this period. That is, the percentage of events that use insider tactical forms has increased over time, while there has been a decrease in the use of outsider tactical forms. Similarly, there is less property damage and violence in the latter part of this period. It is interesting to note, however, that Rucht (1998) finds that in West Germany, there has been an increase in recent years in the percentage of events using violence and in which there is property damage. Thus, over time more events have at least one organization present, events draw more people, and they tend to be more moderate (i.e., they use less radical tactics and damage less property). These findings suggest support for Everett s (1992) argument that over time, organizers of protest events have deliberately used less radical tactics in an attempt to draw more