Protest event analysis (PEA) and the related concept of repertoire of contention are widely used in the study of social movements. Are they appropriate for the study of feminist protest? I argue that conventional forms of protest event analysis may have significant limitations when applied to feminist protest. Unobtrusive or individualized forms of resistance and protest associated with feminism are difficult to measure through typical protest event data. Moreover, the concept of repertoires of contention retains within it a number of unwarranted gendered assumptions. Some flow from being too reliant upon protest event data. I suggest that repertoires may be gendered, that this is unacknowledged by those who use the concept, and that this has implications for its normative dimensions.

The Trouble with PEA

PEA has proved attractive as it facilitates the consideration of a wide variety of movements and actions, enabling both historical-comparative analyses and quantitative analysis (Olzack Reference Olzack1989). Some have suggested that it is now the dominant methodology in the field of social movement studies (Earl et al. Reference Earl, Andrew, McCarthy and Soule2004). Since Herbert Danzger (Reference Danzger1975), there have been numerous attempts to measure and correct for PEA's biases. While some have maintained that it remains the best methodology for the study of social movements and that the problems of bias and validity are minor, others have concluded the opposite. Newspaper data are not as robust as other forms of social science data, do not measure social movement activity (they are simply analyses of biased newspaper reporting of protest), and should be treated as such (Ortiz et al. Reference Ortiz, Myers, Wall and Diaz2005).

Researcher bias arises from the researcher's practices of sampling or coding. Not all daily editions of a newspaper may be selected due to cost and time, resulting in the selection of only those protests involving larger numbers of people or those that involve heightened levels of physical conflict. As a result, PEA may undercount feminist protest because such protest is often not in the settings covered by newspaper accounts (Rucht Reference Rucht, Banaszak, Beckwith and Rucht2003, 259–60; Sawer and Grey Reference Sawer, Grey, Grey and Sawer2008, 3). When one examines the operationalization of what counts as a protest event, it appears that in many cases what is described is typically masculine forms of behavior. Often there is a concern with violence and its intensity and the size of the gatherings involved (McCarthy, McPhail, and Smith Reference McCarthy, McPhail and Smith1996). Obviously, this may exclude smaller nonviolent gatherings. As a result, given that there is good reason to assume that feminist mobilizations are more likely to use nonviolent tactics or unobtrusive repertoires (Bagguley Reference Bagguley2002; Katzenstein Reference Katzenstein1990; Taylor and Van Dyke Reference Taylor, Van Dyke, Snow, Soule and Kriesi2004), they are unlikely to be reported in the press and measured using the usual techniques of PEA.

In these debates, there is no recognition that questions of researcher bias might influence the analysis of feminism as a social movement. It is notable that there have been very few sustained attempts to study feminist protest using protest event data (e.g., Costain Reference Costain1992; Grey Reference Grey2009; Rucht Reference Rucht, Banaszak, Beckwith and Rucht2003; Soule et al. Reference Soule, McAdam, McCarthy and Su1999).

Selection bias occurs due to the processes of news reporting. Some events are seen as more newsworthy than others, news agencies might not cover certain geographical locations, or some events might fit in with short-term media attention cycles (McCarthy, McPhail, and Smith Reference McCarthy, McPhail and Smith1996). The vast majority of protest event studies fail to face up to the possibility that their data suffer from selection biases that are impossible to control for (Ortiz et al. Reference Ortiz, Myers, Wall and Diaz2005). The principal concern here would be the systematic underreporting of feminist protest. Many of the discussions of selection bias remain resolutely ungendered.

Some events are seen as contentious while others are not, and what is seen as contentious may change over time, reflecting description bias. Pamela Oliver and Gregory Maney (Reference Oliver and Maney2000) suggest that there is a triadic relationship among institutionalized politics, the media, and protest. Furthermore, as protest has been routinized, with movements becoming more formalized and using less disruptive tactics, their activities are of less interest to the media. What exactly counts as protest, according to Oliver and Maney, has become vague and fuzzy. Protest is increasingly symbolic and less of a serious challenge to the dominant social order (Oliver and Maney Reference Oliver and Maney2000, 468). This is especially the case with feminist protest (Bagguley Reference Bagguley2002; Katzenstein Reference Katzenstein1990; Rucht Reference Rucht, Banaszak, Beckwith and Rucht2003; Sawer and Grey Reference Sawer, Grey, Grey and Sawer2008; Taylor and Van Dyke Reference Taylor, Van Dyke, Snow, Soule and Kriesi2004).

Until recently, the vast majority of journalists were men, yet those using PEA have not generally recognized that this might influence the types of protest and movements that get reported. Thus, to the catalog of already existing problems, we can add that of gender bias within the newspaper industry. It is almost a truism of contemporary social science that the mass media are in some way or form biased in their reporting. Journalists not only select what to report but also decide on how to report it (Lester Reference Lester1980).

What direction should PEA take with respect to the study of feminist protest? First, we should take a broad approach akin to that of “claims analysis” (Koopmans and Statham Reference Koopmans and Statham1999). This would enable us to examine the wider cultural and political context of feminist mobilization as it is reported in the press. Within this approach however, it is important to distinguish claims and actions arising from different sources—movements as well as more institutionalized forms of political organization. Second, a key concern of the analysis should be the changing patterns and biases of the reporting of feminism broadly conceived. We should not fall into the trap of assuming that all that is reported is all that is happening. It is not the analysis of social movement actions but the analysis of the way that a public political debate is reported in the sources that we use. Third, PEA tends to treat all events as of equal significance and the quantity of events as what matters. The interpretation and meaning of those events for participants, bystanders, and opponents is lost in many cases. In these instances, we should look at particularly influential protest events in more detail as qualitative case studies.

Repertoires and Social Movements

The concept of repertoires of collective action was introduced by the late Charles Tilly in his studies of social movements in nineteenth-century Europe. People learn protest strategies and tactics through the process of political struggle (Tilly Reference Tilly and Mark1995, 26–27). This original formulation focused on the shift in the repertoire of contention in Britain between the eighteenth and nineteenth centuries.

Tilly described the eighteenth-century repertoire as parochial, particular, and bifurcated. He described it as parochial because the protests and issues concerned were limited to a specific locality, region, or community. They did not extend geographically to the level of the nation-state or trans-nationally. This repertoire was particular inasmuch as the tactics used differed significantly among locations, actors, and circumstances. Finally, the old repertoire was described as bifurcated as people took direct action locally to address local grievances, but national-level grievances were addressed indirectly through a local person in a position of authority. In contrast to this repertoire, Tilly described the new nineteenth-century repertoire as national, modular, and autonomous. It was national inasmuch as it involved interests across many localities and regions. It was seen as modular as the same tactics, strategies, and forms of action were used by many diverse social constituencies from across the country. Finally, it was autonomous in that the protesters sought to directly influence the national centers of power, rather than work through some locally significant intermediary (Tilly Reference Tilly and Mark1995, 33–34).

Toward the end of his life, Tilly modified this account somewhat. The repertoire now became a central component of his conceptualization of the “modern social movement.” This consisted of publicly organized collective campaigns aimed at those in power. It used a historically distinctive repertoire of creating organizations, public meetings and demonstrations, petitions, attempts to influence the press, and other “propaganda.” The movement also represented itself as composed of worthy, unified, numerous, and committed individual citizens (Tilly Reference Tilly2004, 3–4).

The close connection between the conceptualization of repertoires and the methodology of protest event data raises the issue of whether or not what is being measured and how it is measured significantly affects the construction of the concept. Only those forms of action reported in the press become conceptualized as part of the repertoire. Furthermore, these measurements are typically quantitative, albeit often illustrated with some suitably dramatic examples of description from the contemporary press. If women's protests are relatively unreported, then this might have implications for how repertories are conceptualized. Since it seems that Tilly constructed the concept of the repertoire “inductively” from his data, missing data would affect that process. It might have been rather more appropriate to refer to repertoires of media representation of movements.

Tilly seems not to have considered that repertoires might be gendered. Indeed, the dimension of modularity that characterizes the modern repertoire effectively excludes from consideration how the diversity of the form of action might be related to the type of actor concerned. As we have seen, the very methodology of protest event data analysis based upon newspaper reports may actually under-represent feminist protest. As feminist forms of protest may not appear in the newspapers due to selection bias or not be identified in PEA due to researcher bias, the concept of the repertoire of contention based on such studies may be too narrow (Sawer and Grey Reference Sawer, Grey, Grey and Sawer2008; Taylor and Van Dyke Reference Taylor, Van Dyke, Snow, Soule and Kriesi2004, 268).

Repertoires can also vary during the “life-time” of a movement. Many have suggested that the repertoire of feminism has moved from public protest to less visible forms of institutionalized political action and influence since the 1970s (e.g., Andrew Reference Andrew2009; Bagguley Reference Bagguley2002; Katzenstein Reference Katzenstein1990; Sawer and Grey Reference Sawer, Grey, Grey and Sawer2008). These are less visible at least to methodologies such as PEA and traditional conceptions of the repertoires of contention. Such changes arise from the interaction between the movement and its wider political context, as well as changes within the movement (Rucht Reference Rucht, Banaszak, Beckwith and Rucht2003).

While the notion of a repertoire might remain useful as a general metaphorical description of the form of collective action, we still need to ask what kind of repertoires are used and under what circumstances. It might prove useful as a sensitizing conceptual device for orienting some quite general research questions. However, in doing so we need to be fully aware of its gendered character.