Abstract
Although recruitment is often studied by social movement scholars, the literature lacks an integrated account of what recruitment actually entails and there is disagreement about who the recruiters are (weak- or strong-tie contacts). This study develops a theoretical account of different recruitment “functions” that can be fulfilled by social ties and that enhance protest participation. Participants need to be informed about the protest event, many of them need to be persuaded before they join a demonstration, and most of them also want to have suitable company to attend the protest with. We expect that these three recruitment functions—information, persuasion, and company—are often fulfilled by the same social tie (homogeneous pattern). However, when this is not the case, we expect that strong ties become more important the closer one comes to actual participation (centripetal pattern). Drawing on individual-level information about more than 14,000 protesters, we find that these patterns are robust across 84 demonstrations in eight countries. Nevertheless, there are systematic differences across countries and across types of social movements. The centripetal recruitment pattern is significantly more apparent when people protest for socio-economic issues compared to when they participate in demonstrations on other, non-socio-economic issues.
Recruitment is one of the most researched processes in the broad field of social movement scholarship. The main conclusion of this large body of work is clear: there is an overwhelming agreement that being recruited is paramount for movement participation. Scholars also concur that recruitment is mainly a “micromobilization” process: individuals are mobilized via interpersonal social networks by people they know. Nevertheless, despite the size of the micromobilization literature, some important theoretical, empirical, and methodological matters remain.
First, the literature fails to conceptualize movement recruitment in a theoretically satisfying manner. Scholars agree that recruitment refers to the process leading to the decision of an individual to join (or not) a protest event and/or a social movement. But what does recruitment exactly entail? What happens when people are recruited? And what functions are fulfilled by recruiters? In other words, the literature does not generate a systematic theoretical account of micromobilization. Second, there is empirical disagreement about who the recruiters primarily are. Studies that look into recruiters came to contradictory results. Some studies suggest that people are mostly recruited via weak-tie contacts, such as co-members of an organization, while other studies conclude that participants are recruited predominantly by people they have strong ties with, like family members or friends. Third, the micromobilization literature suffers from a methodological bias because the evidence comes predominantly from U.S. case studies. Apart from a handful of exceptions, there is a dearth of comparative studies and, as a consequence, we are unsure whether the found patterns are idiosyncratic or applicable to protest participation on different issues and in different countries. In sum, while scholars agree that micromobilization is crucial for movement and protest participation, there is room for improvement regarding the depth and breadth of what we know. We lack theory about potentially different recruitment functions, there is no empirical clarity about who the exact recruiters are, and, due to the limited scope of the available evidence, there are doubts about the generalizability of the existing micromobilization findings.
It is this study’s ambition to start filling that void. We draw upon evidence collected among more than 14,000 individual protest participants during 84 demonstrations in eight European countries. The events in our sample are all legal street demonstrations, yet they cover a wide variety of issues (e.g. from anti-austerity to gay pride events) in a variety of democratic systems. This gives us the opportunity to check the robustness of our findings and to look for differences across issues and countries. We develop a systematic theoretical framework of what recruitment entails and distinguish three recruitment functions: information, persuasion, and company. These three functions match questions in our survey in which participants were asked from whom they received the information that a protest event was taking place, by whom they were asked to participate (persuasion), and in whose company they eventually attended the demonstration.
Our study draws on evidence about participants only; we do not compare participants with non-participants. As a consequence, we cannot say whether information, persuasion, and company provided by specific ties actually cause participation. In fact, it may be the case (though unlikely) that non-participants were informed, persuaded, and offered company by the same type of people as the participants were. Research comparing participants and non-participants in specific movement activities is rare because the evidence is hard to gather (Klandermans and Oegema, 1987; McAdam and Paulsen, 1993; Van Laer, 2017). Since we cannot assess causality by looking at participants only, we will talk about recruitment “functions” and not about participation “determinants” and we avoid causal language when reporting the results. Still, we believe that we can learn a lot about the mechanisms of recruitment by looking at participants only. We may not be able to conclude that some ties are more effective than others in bringing people to the streets, but we are able to observe varying frequencies of ties fulfilling different functions among the people who protested.
We find that recruitment is indeed a multifaceted process with three functions—information, persuasion, and company—quite often performed by different ties. Many different combinations of ties fulfill the three recruitment functions but two patterns prevail: a homogeneous pattern in which the same tie provides information, persuasion, and company and a heterogeneous pattern whereby different ties fulfill the three functions. Within the heterogeneous pattern, it seems to be the case that strong ties play an increasingly important role as one moves from information to persuasion to company. This points to a “centripetal” pattern: the closer one comes to actual participation, the more strong ties, close to the individual matter. Strong ties prevail especially when it comes to the company function. This general pattern holds across countries and demonstrations. Yet, we do see notable and understandable differences across protest issues. The centripetal recruitment pattern is more prevalent in protest on non-socio-economic than on socio-economic issues.
Recruitment functions and tie strength
It has become a truism in social movement research that interpersonal social networks are key to explain people’s engagement in social movements and participation in protest (for rare opposite accounts, see Jasper and Poulsen, 1995; Lim, 2010). Starting with the work of Snow et al. (1980) in the early 1980s, a large number of studies, both old and recent, show over and again that participation is a matter of recruitment by other individuals through a person’s personal network (see among many others, for example, Gould, 2003; Kitts, 2000; Knoke, 1990; Lim, 2008, 2010; McAdam, 1988; McAdam and Paulsen, 1993; Schussman and Soule, 2005; Verba et al., 1995). Snow et al. (1986) coined the concept of “micromobilization” to refer to the role played by micro networks which foster movement and protest participation by increasing the chance that one would be exposed to recruitment attempts. Verba et al. (1995) aptly summarized the state of the literature more than 20 years ago: “Any attempt to understand the roots of participation must take into account the impact of requests from others” (p. 138). Notwithstanding the impressive amount of evidence regarding the importance of micromobilization, it remains unclear what interpersonal recruitment precisely entails and by whom people are recruited.
Our main claim in this article is that interpersonal recruitment in networks is a multidimensional phenomenon (see also Passy and Monsch, 2014). Making people participate in a protest event supposes that several thresholds are overcome. We are not the first to conceptualize mobilization as a process that entails the clearing of several hurdles. The best-known example probably is the drop-out mobilization model proposed by Klandermans and Oegema (1987) or the earlier work by Granovetter (1978; see also Verhulst and Walgrave, 2009). What these barrier approaches have in common is that participants are considered to some extent to be rational actors who weigh the costs and benefits of participation and who only participate when benefits outweigh costs. We argue that interpersonal recruitment helps people to overcome three potential barriers that prevent them from participation: informational, attitudinal, and social hurdles.
First, recruitment implies factual information transferal: a recruiter informs the potential recruit that event X is taking place at place Y on date Z. It is pretty straightforward that people need to know that a protest event is about to take place before they can start thinking about participating. Second, recruitment has a persuasive function as well: the potential participant must be convinced to take part in the protest. He or she needs to be motivated to join. Most studies tend to focus on this part of recruitment: being asked by others to join. Asking someone to participate implies an attempt to convince the other to partake. A third barrier is social. Protest, like many other forms of activism, is something people mostly do in the company of others. Recruitment implies in most cases that a potential participant finds suitable company to attend the event with. If you do not find such a company, you may not show up after all, even if you were informed and persuaded.
Although these three functions suggest a kind of chronological order in the recruitment process—first information, then persuasion and finally company—they can happen at the same time or even in reverse order. Information can chronologically precede persuasion and it is likely that finding company mostly comes at the end of the process, but this is not necessarily so. Also, the three functions can be fulfilled outside a potential recruit’s interpersonal network. For instance, a potential participant can be informed about the event by a poster, he or she can be persuaded to take part by reading a call to arms on an organization’s website, and can attend the event with co-members of an organization he or she did not know personally before the event. Although we acknowledge that the recruitment functions can be performed by other actors than social ties, this study focuses on interpersonal recruitment only and on how social ties may perform the three functions of information, persuasion, and company.
Extant research focuses predominantly on the persuasive function of micromobilization—the being asked part. It teaches us much less about the information and social functions. The neglect of the multifunctional nature of interpersonal recruitment may be why previous research did not come to clear conclusions about who does the recruiting (see below; note that some approach the recruitment puzzle in a more complex way, see, for example, the work of Beyerlein and Hipp (2006) and Klandermans and Oegema (1987) who speak of recruitment as a multi-stage process). In fact, the information, persuasion, and company recruitment functions can be fulfilled by different sorts of recruiters. The people who inform others about an upcoming event, those who persuade people to take part, and those who invite others to participate in their company can be different people. This is the second main argument this study makes.
Before we continue with arguing why the three functions may be fulfilled by different others, we want to emphasize that in many cases the three functions are probably fulfilled by the same person or social contact. Indeed, the distinction between the three functions, while analytically valid, is to some extent also artificial. For instance, a friend, during a single chat, can tell someone about the event, convince him or her to take part, which then leads to the agreement to attend the event together. As far as the three functions are fulfilled at the same time and in a synchronous way, the informer, the persuader, and the prospective company will be the same other person (or group of persons). In sum, we expect that there is an important basic pattern of homogeneous recruitment. In addition, however, if people are informed, persuaded, and accompanied by different people—heterogeneous recruitment—we expect there to be a distinct secondary pattern with an increasing importance of strong ties as one goes from information over persuasion to company. This centripetal pattern implies that the closer one gets to actual participation, the more important the role played by close ties. We now further elaborate how and why this would be the case.
A useful way to classify types of recruiters is Granovetter’s (1973) classic distinction between strong- and weak-tie contacts. Strong ties are loaded with trust and intimacy, they refer to people one has a close and tight relationship with, people with whom one spends a lot of time; typical examples are family members or partners. Weak-tie contacts are more distant, they are people one knows, with whom one may even entertain friendly interactions but these contacts are less familiar, less time-intensive and less mutual trust is implied; typical examples are colleagues. Drawing on this well-known distinction between strong and weak ties, the argument we develop in the remainder of this section is that strong ties become more important as one moves from the informative, over the persuasive, to the social function of recruitment. In other words, strong ties are especially important because they provide good company to attend a protest event with, and also because they are persuasive recruiters, but they are less important as providers of information about an upcoming event.
Granovetter’s (1973) seminal study about the strength of weak ties claimed that weak ties are “stronger” than strong ties in the sense that they are excellent conduits for letting information flow between far-off groups. Via strong ties people typically do not get new information as strong-tie contacts only dispose of the information that they already had. Applied to protest recruitment, the advantage of weak ties is that they allow factual information about a protest event to diffuse far and fast. But the advantage of strong ties, in contrast, is that they may support costly and high-threshold social action (Centola and Macy, 2007). While weak ties are good at disseminating information, they are less appropriate at actually persuading people to do certain things. Strong ties are more suited for persuasion for several reasons (see Somma, 2009). People are more emotionally involved with their strong-tie contacts and they trust them more. This increases their persuasive leverage. Also, people with whom one holds strong ties are on average more similar than those with whom one holds weak ties, and similarity (and common identity) breeds persuasion. Finally, social conformity pressure is stronger when it comes to strong-tie contacts: the relationship at stake is more important and this increases the chance that a potential recruit will just give in and agree to participate because he or she wants to keep an important relation intact or because they care about their reputation (see, for example, Brady et al., 1999; Gould, 1993).
We expect strong social ties not only to be more important persuaders than weak ties, but we also expect them to be more important providers of attractive company to attend the event with. Walgrave and Wouters (2014) recently argued that when people ask others to attend a protest event, they actually ask these others to attend the event with them. If we accept that people generally like to be in the company of strong ties—an assumption that is not too far-fetched when it comes to good friends or spouses—it makes a lot of sense to expect that they are relatively more likely to attend protest events with strong-tie contacts and less with weak-tie contacts.
Our claim that different ties match different recruitment functions seems straightforward. Not surprisingly, other scholars formulated similar theoretical intuitions. For example, McAdam (1986), in his original Freedom Summer study, wrote: “Although weak ties may be more effective as diffusion channels, strong ties embody greater potential for influencing behavior” (p. 80). And later, together with McAdam and Paulsen (1993), he repeated it in other words: “At the microlevel, ties are less important as conduits of information than as sources of social influence. And the stronger the tie, the stronger the influence exerted on the potential recruit” (p. 655). Or, Somma (2009) stated in a more recent study: “Although less effective than weak ties as channels of nonredundant information, strong ties could have their own specific ‘strength’ by increasing interpersonal influence to a greater extent than weak ties” (p. 291).
All these authors suggest that weak and strong ties play a different role with regard to different aspects of the recruitment process. But none of the previous empirical studies conceptually distinguished multiple recruitment functions, none of them developed a systematic argument about how these functions are fulfilled by different ties, and no studies empirically examined whether this is actually the case in reality. Our study directly builds on these previous studies but adds theoretical integration and empirical proof.
Looking into the empirical literature, the lack of a systematic account integrating and differentiating different recruitment functions becomes clear. All studies we know of focus on one function only. That is why they probably came to contradictory conclusions regarding which type of tie matters most for recruitment. For example, in their pioneering micromobilization study, Snow et al. (1980: 791) summarize the early literature by quoting 10 earlier case studies, mostly about recruitment in religious movements. A minority of these studies found that recruitment was done by strong-tie contacts (family and relatives), a majority contradicts this and concluded that participants were recruited by weak-tie contacts (acquaintances, neighbors, colleagues, etc.), and, in one study, participants were recruited by people they did not even know personally. The reason for these mixed findings is, we believe, that recruitment was not clearly conceptualized by Snow and colleagues and that different recruitment functions were actually confounded in the studies they listed.
In most of the more recent studies, recruitment was not defined very precisely either nor was it conceptualized as a multifunctional process. Most scholars conclude that strong-tie rather than weak-tie relationships foster protest participation. But when looking more closely at these strong-tie studies, it becomes clear that they are all addressing the persuasive or company function of interpersonal contacts. Passy (2001), for instance, explains that, rather than weak ties, strong ties bestowed with trust and intimacy are needed to overcome the cost and uncertainty that comes with protest participation (see also McAdam and Paulsen, 1993). She operationalizes recruitment as the incitement of others to join the organization, which clearly refers to persuasion. The same applies to other studies finding that recruitment happens through strong ties. McAdam’s (1986) Freedom Summer study concluded that strong ties are most helpful for overcoming the barrier to high risk/cost activism. McAdam’s impressive study did not explicitly differentiate between different recruitment functions but he implicitly addressed the persuasion and company function of ties since all subjects of his study had been informed about the event. The study by McAdam and Paulsen (1993) showed that, what they call, social support by strong-tie others (i.e. parents) was more important for participation than support from weak-tie people (i.e. religious figures). This comes close to what we call the company function of recruitment. In sum, extant case studies looking at the persuasive and social aspects of recruitment found strong ties matter more than weak ties (Brady et al., 1999; for similar claims, see Gerlach and Hine, 1970; Jasper and Poulsen, 1995; Klandermans and Oegema, 1987; Somma, 2009).
In contrast, the few studies that—again mostly implicitly—emphasized the information function of recruitment found that weaker, often organizational, ties are more instrumental for information dissemination and, eventually, participation. For instance, Fisher (2010) compared large-scale protests in France and the United States and reports that most people heard about the protest via weak or formal ties to organizations (see also Fisher and Boekkooi, 2010). A study by Bedoyan et al. (2004) of anti-European Union (EU) summit protesters in Brussels shows the same: participants mostly mention weak ties or even the total absence of ties (e.g. mass media) when they are asked about how they were informed. Lim (2008), like many others using the US Citizen Participation Study data, also downplays the effect of strong ties for participation: ties with organizational co-members—we would call those weak ties here—are what fosters participation most, he says.
We explained that the three functions can be fulfilled by the same social ties (homogeneous recruitment) or by different social ties (heterogeneous recruitment). Within heterogeneous recruitment, we forecast a centripetal pattern whereby persuasion is provided by a stronger tie than information, and company is provided by a stronger tie than persuasion. So far, our reasoning is exclusively situated at the individual level. Since the social environment in which recruitment happens varies across issues and movements, there could also be differences in recruitment patterns across issues/movements. For example, gay pride marches are rooted in very different social circles than manual workers’ protest, and this could affect the recruitment patterns underlying these protests. The literature is rife with studies analyzing recruitment for specific events or movements and sometimes these accounts emphasize the specificity of the recruitment process of each of these movements (e.g. Cable, 1992; Passy and Giugni, 2001; Swank, 2018).
For our purpose here, we believe the most useful distinction is the one between protest for socio-economic issues, on the one hand, and non-socio-economic protest, on the other. In fact, the social movement literature has long distinguished between so-called “old” and “new” social movements with the old social movements basically being the labor movement and the trade unions (see, for example, Kriesi et al., 1995; Melucci, 1989). One of the key differences between both movement types is their organizational backbone and the type of beneficiary they fight for. For instance, protest against austerity measures—in the period under investigation austerity was very much present—is typically organized by strong trade unions and, most of the time, aims at defending the socio-economic (or material) interests of clearly defined groups. Climate change protests, to give a contrasting example, are often staged by loosely organized networks of smaller organizations. In addition, these protests do not defend specific groups’ material interests. Since non-socio-economic protest is less organized and draws on a potentially broader support base we expect the centripetal recruitment pattern to be more prevalent in these protests than in socio-economic protests. For a climate march, for example, one could imagine very diverse social ties such as parents, family, friends, acquaintances, colleagues, and co-members to be relevant categories that could play a role in the recruitment process. For a trade union demonstration protesting the closing of a factory, in contrast, the only relevant social ties in the recruitment process probably are the colleagues working in the same factory.
On top of these meso-level differences (issues/movements), there could be macro-level differences in recruitment patterns between countries as well. It could be the case, for instance, that degrees of politicization of family life or of friendship bonds vary across countries, civil society could be more politically active in one country compared to another, and the workplace may spark more recruitment in some countries than in others. Nevertheless, we do not have specific a priori expectations with regard to differences between countries regarding recruitment patterns—the interplay of different ties and the different recruitment functions. In the results section, though, we empirically test whether real country differences exist and, where we find such differences, we explore them and come up with a preliminary interpretation.
Data and methods
Evidence comes from protest surveys of, initially, 17,000 individuals fielded during 84 demonstrations in eight countries (Belgium, the Czech Republic, Italy, the Netherlands, United Kingdom, Spain, Sweden, and Switzerland). All demonstrations took place between 2009 and 2014. The sample of demonstrations is a convenience sample in the sense that, first, countries were selected based on the presence of a research team and that, second, within the eight countries, the biggest and most visible demonstrations during the research period were covered. Data were collected by an international team of scholars in each of the countries under study. The data collection followed the established field work method for protest surveying (Walgrave and Verhulst, 2011). Following that method, two groups of interviewers, each directed by a fieldwork supervisor, select demonstrators in a systematic way in order to give every demonstrator the same chance to be surveyed. Respondents are asked to accept a questionnaire, to complete it at home and to send it back by land mail to the researchers afterward. Some respondents also answer a few oral questions before they are asked to accept the postal questionnaire. Comparing the brief oral questionnaires with the sent back postal questionnaires allows to assess the representativeness of the postal questionnaires. This method produces sample data that are mostly representative for the people participating in a covered protest event but there sometimes is a response bias with regard to age and education (Walgrave et al., 2016). However, respondents’ age and educational level are not correlated with the different recruitment functions of weak and strong ties. Therefore, as weighting may actually cause more problems than it solves, we decided not to weigh the data.
Table 5 in Appendix 1 provides a full overview of all covered demonstrations in the eight countries with their date and issue. The overview also shows the varying response rates per demonstration. The average response rate is 26 percent (in total, 57,109 questionnaires were distributed), but that summary statistic conceals considerable differences (going from 45% at a Dutch demonstration to protest against austerity in the cultural sector to 6% for a Czech protest against corruption). All countries are European democracies, so they form a coherent group of nations. We cannot claim that our data set of 84 demonstrations forms a perfectly representative sample of all demonstrations occurring in these eight countries during the research period. Yet, in most countries, all large demonstrations held during that period were covered. The least we can say is that we have a diverse sample of demonstrations on a good many different issues. Our study has a broad comparative scope that outperforms previous research on micromobilization.
The demonstrations were, based on their underlying issue, split up in socio-economic and non-socio-economic events. We classified as socio-economic the following demonstrations: demonstrations against budget cuts, austerity, and economic crisis; demonstrations in favor of workers’ rights and jobs; and May 1st celebrations. Non-socio-economic are the protests against climate change, nuclear energy, abuse of animals, corruption, racism, hate crime, xenophobia, violence against women, inequality between men and women, abortion, as well as protests in favor of changes in the voting system, regional independence, respect for LGBTs (lesbian, gay, bisexual, and transgender), coming out of gays, more public awareness of LGBTs, and peace.
We only consider participants in street demonstrations. Demonstrating probably is the most prototypical protest activity of the last decades (Norris et al., 2006). Moreover, the advantage of just concentrating on demonstrations is that we keep the action form, and the related cost and risk of the activity, constant. Also, the fact that all countries are Western European countries is an advantage as the risk of police repression is similar across the countries. Indeed, previous work suggests that recruitment for more risky/costly forms of action may be different from low risk/cost activism (e.g. McAdam, 1986). By selecting similar events in similar countries we can fully focus on examining which social ties perform which recruitment function while keeping other things constant.
The surveys fielded during 84 demonstrations contain identical measures of how people got to know about the event (information), by whom they were asked to participate (persuasion), and with whom they attended the protest (company). Of our 17,000 respondents, some did not answer all three recruitment questions. That is why the actual number of respondents used below is only 14,692. Descriptive statistics of the three key variables are presented in Table 1.
Variable descriptives (N = 14,692).
Information Tie is based on the following question: “How did you find out about the demonstration? (Check as many as apply).” We presented respondents with 10 different potential information channels including interpersonal ties (fellow members of an organization, friends, and/or acquaintances; people at school or work; and family and/or partner) and some non-personal channels (organization, newspapers, radio or TV, online social networks, alternative online media, and ads/flyers and/or posters). Responses were categorized into five dummy variables, going from weak to strong: no tie, co-members of an organization, colleagues or co-students, friends or acquaintances, and partner or family. The categorization of specific ties into strong-weak follows earlier work, for example, by Somma (2009). The crudeness of this simple classification leaves a lot to be debated. First, friends and acquaintances are combined into one variable here although this is far from ideal. Aiming to keep the survey within limits we, unfortunately, took friends and acquaintances together in the Information Tie question. We are aware that this is a suboptimal situation as friends form definitely stronger ties than acquaintances. The only advantage we see is that this noise in our data probably works against finding the hypothesized centripetal recruitment pattern. Second, one could debate whether, in a context of protest recruitment, co-members actually represent the weakest tie. It all depends on the degree of engagement of a person in the organization, of course. Yet, we do not have any information about this, so we decided to code co-members in a conservative way as weak ties. Third, the category of colleagues and co-students is very crude as well. In fact, the social institutions (work and school) from which these two ties originate are different. Our measure does not allow to assess the different recruitment weight crucial social institutions may have. Fourth, we decided to consider “no tie” as the weakest tie category. While this may make intuitively sense, it is clear that a no tie situation is not simply the final step on a linear scale from strong to weak ties; it is something different. Finally, ties are multiplex. Another person can at the same time be a friend and a colleague of a respondent. How that respondent then answered our questions—which ties he or she ticks as applying to his or her recruitment—is unknown to us. This creates additional noise in the evidence as some ties are probably counted twice and other ties may be underestimated as they are “overruled” by other ties. Since we are mainly interested in differences across functions and since there is no reason to expect that our three core tie questions are answered differently in this respect, we believe the noise is not systematic and, again, rather works against than in favor of our homogeneous and centripetal recruitment pattern expectations.
A similar measure is used with regard to Persuasion Tie. We draw on the question: “Which of the following people specifically asked you to take part in the demonstration? (Check as many as apply).” Respondents got seven answer options (no-one, partner or family, relatives, friends, acquaintances, colleagues or fellow students, and co-members of an organization). We can again classify the answers into dummy variables going from the weakest to the strongest tie: no-one, co-members of an organization, colleagues or co-students, acquaintances, friends, and partner or family. Note that the exact question, querying by whom participants were asked to participate, does not strictly match the persuasion aspect of recruitment. Yet, we believe that it is safe to assume that asking someone to participate actually is an attempt to persuade that person to take part.
For Company Tie the question was “Were you at this demonstration with . . . (Check as many as apply).” and we gave them the option to choose between eight options: alone, with partner, with children, with relatives, with friends, with acquaintances, with colleagues or fellow students, and with co-members of an organization. This led again to six identical dummy variables from weak to strong ties: alone, co-members of an organization, colleagues or co-students, acquaintances, friends, and partner or family.
The three tie questions we use to tap the three key variables refer to specific types of people (note that our information question is more ambivalent in this regard). Our measures do not tell us about the exact medium via which these types of people have been interacting with the potential participant. The interaction may have occurred via face-to-face encounters, via the telephone, via text message, via email, or via social media. There is a lot of debate among social movement scholars about the role of social media (see, for example, Van de Donk et al., 2004). Social media allow people to keep in touch with, especially, their weak-tie contacts. We do not aim to address the social media debate here though. We are interested in the types of people that perform the recruitment functions, not via which medium they do so.
Results
Our first expectation was that there would be a good deal of homogeneity in the recruitment process, meaning that the same tie performs all three recruitment functions. We start with presenting some bivariate evidence of this homogeneous recruitment pattern in Figures 1 and 2. Figure 1 shows, among those who received information about the event from a certain tie, the percentage of respondents who were persuaded by the six possible ties. Figure 2 is made up similarly and compares the company of people within the group of people that were persuaded by a certain social tie. A reading example can clarify things: consider the first few bars in Figure 1 dealing with respondents who mentioned that they were informed about the imminent protest event by their partner or family. Of those 2200 people, 65 percent say that they also were persuaded by their partner or family, 24 percent said they were persuaded by friends, 9 percent referred to acquaintances, and so on. Please note again that respondents were allowed to check multiple persuasion ties. The persuasion percentages within one information category, therefore, add-up to more than a hundred percent. Please note, as well, that because of the same reason the total N is higher than the number of respondents (14,692). The 95 percent confidence intervals are shown for all bars in the two figures.
Social ties and their role in information and persuasion. Percentage of participants mentioning a Persuasion Tie within each Information Tie group.
Social ties and their role in information and company. Percentage of participants mentioning a Company Tie within each Persuasion Tie group.
The figures quite forcefully suggest that people who are informed about a demonstration by a certain social tie also tend to be persuaded by the same social tie (Figure 1), and that those who have been persuaded by a particular social tie tend to show up in the company of that same social tie (Figure 2). Of all the corresponding bars, in 9 of the 11 cases, the bar referring to the same social tie is the tallest. The first exception is the company of people who were persuaded by acquaintances (Figure 2): they show up more often in the company of friends (62%) than in the company of acquaintances (53%). The other exception is the company of people who were not persuaded by anyone (Figure 2): more than half of them (55%) went to the protest with their partner or family members instead of going alone (23%). In all other cases, the “matching” bar is the tallest—often by a large margin. In other words, there is a strong, baseline pattern of recruitment homogeneity: social ties tend to reproduce themselves across the three recruitment functions. Still, a lot of variance remains: for many people, the three recruitment functions are fulfilled by different ties. We see in Figures 1 and 2 that, across the board, the combination of the other social ties is higher than the matching social tie (e.g. the first six bars in Figure 1 show that 65% mentioned partner or family, but 76% of the respondents (also) mentioned other or no ties).
Now our second expectation comes into play: if different ties provide information, persuasion, and company, strong ties should matter most for the company function and least for the information function, with the persuasion function in between. In order to examine whether this centripetal recruitment pattern is frequent, we coded for each respondent which tie performed each of the three functions. If a function was performed by more than one tie (e.g. someone was asked to participate by a friend and by a family member), then the strongest tie prevailed (the respondent was coded to have been asked by family). We use five categories of ties, here presented from weak to strong: no ties (0), co-members (1), colleagues/co-students (2), friends/acquaintances (3), and partner/family (4). As mentioned, we are able to distinguish between friends and acquaintances for persuasion and company. But for reasons of clarity, we decided to create the same five-step scale for all three functions. If we recalculate the variable using the fuller scales for persuasion and company, the results presented below remain almost exactly the same (only very minor changes; see Appendix 2). Being interested in whether the role of strong ties varies across the three functions, we can now compare the strength of the ties across recruitment functions. Table 2 contains the results presenting three comparisons.
Comparison of tie strength across recruitment functions.
Note: The figures in the table refer to the number (share) of cases in which the first function (1) was performed by a weaker/same/stronger tie than the second function (2).
According to the centripetal recruitment pattern, we expect for all three comparisons in Table 2 that the “stronger tie” percentage is larger than the “weaker tie” percentage. We can see, for instance, that for 22.7% of the respondents, the persuasion tie was stronger than the information tie, but for 20.6% the opposite, unexpected pattern presents itself. For 56.7% these two functions were performed by the same tie. The table further indicates that more than half of the respondents (51.1%) were accompanied by a stronger tie than the tie they were persuaded by. Only 7.2% of the respondents went to the protest with a tie that was weaker than the tie they were persuaded by. All in all, for two of the three comparisons, the expected centripetal pattern seems to hold. But the difference in tie strength between the information and persuasion tie is as good as nonexistent. This suggests that while the company function is mostly fulfilled by people with whom the recruit has a stronger tie than with those who informed and who persuaded him or her, it is hardly the case that persuasion ties are on average stronger than information ties.
Going one step further, we calculate a single recruitment pattern score for each individual: an additive metric consisting of the three comparisons presented in Table 2. If the persuasion function was performed by a stronger tie than the information function, this respondent got “+1.” If the company function was performed by a stronger tie than the persuasion function, the respondent got “+1” again. Another “+1” was added when the company tie was stronger than the information tie. Similarly, “−1” was given when the social tie was weaker in the subsequent mobilization function. This way, each respondent received a centripetal recruitment pattern score between −3 and +3, indicating the extent to which the respondent followed the expected pattern of increasingly stronger ties as we go from information over persuasion to company. Table 3 shows the frequency distribution.
Frequency distribution of centripetal recruitment pattern score.
Note: The higher the centripetal recruitment pattern score (−3 to +3), the more a respondent is informed by a weaker tie, persuaded by a stronger tie and attends the event in the company of an even stronger tie.
A bit more than a third of the respondents (36.2%) has a score of 0. This score refers to homogeneous recruitment whereby the same tie performs the three recruitment functions (but note that respondents with a score of 0 are not necessarily informed, persuaded, and accompanied by the same type of ties, they also could have had +1, −1, and 0). The largest group in our sample (39.5%) has a score of +2—for this group, two of the three steps were performed by stronger ties and one of the three steps was done by the same tie. This indicates that many people to a large extent follow the expected centripetal pattern of increasingly strong ties. Nevertheless, only 6.9% of the respondents were fully mobilized as we expected them to be, with an increase in tie strength in every recruitment step (+3 score). As we saw in Table 2, the reason is that the information and persuasion functions are quite often fulfilled by a tie of the same strength. Our results are strongly driven by the fact that people systematically attend protest events in the company of stronger ties. Remarkably, the data show that negative scores (−1 and −2) are very infrequent (only 8%). So the opposite centrifugal pattern of weak ties becoming more important as we move through the recruitment process is rare. If recruitment is heterogeneous, it predominantly follows the centripetal pattern and only rarely the centrifugal pattern.
Figure 3 presents a way to visually inspect the centrifugal pattern. The thickness of the lines shows how many people follow which recruitment pattern. Each pattern has a distinct color (e.g. the thick orange line represents the large group of people who were informed and persuaded by friends/acquaintances but attended with partner/family). Note that the homogeneously recruited participants were deleted from the graph to be able to inspect better whether those with varying ties display the centripetal or rather the centrifugal pattern. The upward lines in the graph—the thick orange and blue lines, and also the red and green lines—clearly exemplify the centripetal pattern. There are much less downward lines (centrifugal pattern) and they are on average much thinner.
Respondents’ recruitment patterns across all demonstrations (N = 10,845).
To further examine the predominance of homogeneous and centripetal recruitment, we look at the full wealth of all combinations of ties and recruitment functions. In total, there are 5 × 5 × 5 = 125 different combinations possible (of which 124 actually showed up in our sample!). Of the 20 most frequent combinations of ties (together covering more than 71% of the respondents), only four are not of the homogeneous or centripetal type (and they contain 8.7% of all respondents). In contrast, 11 of the most frequent combinations follow the centripetal pattern (36.4%) and 5 follow the homogeneous pattern (26.2%). The most frequent single combination (N = 1411; 9.6%) is being informed and persuaded by no-one, but attending the event in the company of partner/family. Of the 20 most infrequent combinations (covering only 72 demonstrators, 0.5% of our sample), none matched the homogeneous or centripetal pattern. The most infrequent combination (N = 1; 0.01%) is a demonstrator being informed by his family or partner, being persuaded by colleagues or co-students but (sadly) attending the event alone.
The homogeneous and centripetal patterns are widespread, we showed, when we aggregate all data from all demonstrations in all countries. In the theoretical section, we argued that the centripetal pattern should be more frequent in non-socio-economic compared to socio-economic demonstrations since more different types of ties are relevant to the issue underlying these protests. Indeed, for a march opposing global warming, people could be recruited by all types of ties as all these types of ties may care about the issue. For a march against factory lay-offs, the types of ties that matter should be more limited (colleagues, of course, and maybe also family). Table 4 presents a multilevel mixed-effects linear regression analysis with random intercepts (individuals (N = 13,675) nested in demonstrations (N = 84)), predicting an individual’s recruitment pattern score (with higher scores pointing to a more centripetal pattern). The model contains one demonstration level variable named socio-economic and country dummies. Furthermore, it includes a number of individual-level variables that control for characteristics of demonstrators and a demonstration-level variable that controls for the size of the protest event (number of participants/1000). Because not every respondent answered these survey questions, the N is lower than in our previous analyses.
Multilevel mixed-effects regression predicting the centripetal recruitment pattern score.
The results underscore the effect of demonstration issue. Demonstrators participating in non-socio-economic demonstrations follow the centripetal recruitment pattern significantly more than those who participate in socio-economic events. On average, non-socio-economic demonstrators have a centripetal recruitment score, that is, 0.12 higher than socio-economic demonstrators. This effect may appear slim on the full −3 to +3 scale. However, the effect is quite considerable if we take into account that the lowest mean score for a demonstration is −0.12 (Occupy The Netherlands, 5 November 2012), while the highest mean score is 1.34 (May Day Sweden, 1 May 2010). The effect of non-socio-economic versus socio-economic is not driven by any particular subtype of demonstrations. Even if we look at the average scores of demonstration types, we see that most non-socio-economic types—environment (0.99), antiracism (1.02), democracy (1.06), and women (1.03)—have higher recruitment pattern scores than the socio-economic types—austerity (0.84) and May Day (0.96) (these results hold when we use demonstration issues as dummies in the regression analysis). There is one exception, though, with the gay demonstrations that we labeled as non-socio-economic having the lowest recruitment pattern score (0.80), while our typology would have expected it to have high scores (since it deals with non-socio-economic issues).
We also checked whether socio-economic versus non-socio-economic is a driver of homogeneous recruitment. After constructing a homogeneity pattern dummy variable—1 if a person was informed, persuaded, and accompanied by the same tie and 0 in all other cases—we ran the same (but logistic) regression analysis as the one presented in Table 4 (results not shown here). The effect of socio-economic protest issue is not statistically significant (B = 0.132; p = 0.102). Hence, the issue type only affects centripetal recruitment, not homogeneous recruitment. Across all sorts of demonstrations, there appears to be a basic tendency of recruitment homogeneity, whether a secondary centripetal pattern is added depends on the issue of the demonstration (non-socio-economic).
Looking again at Table 4, we can see that demonstrations in Spain (reference category in the models) follow the centripetal pattern more than demonstrations in other countries. Spanish demonstrators are significantly different from all other nationals (except from Belgians, but this difference as well is close to being significant) when it comes to how they are recruited. Differences among the other countries exist, but they are not significant; it is Spain that is the true outlier here (with a recruitment pattern score of 1.14, while the average of the other seven countries is 0.90). This probably is a true country effect, as we control for some individual features and demonstration features. Furthermore, all Spanish demonstrations in our data set (five socio-economic and six non-socio-economic) score above the mean (being 0.94) which implies that they all show the pattern whereby the informing about an upcoming protest event tends to be done by weak ties and persuasion by intermediary ties, while the company function is fulfilled by stronger ties.
We did not develop expectations about country differences in the theory section as we did not have any clues what to expect. But since we find such a strong and robust outlying pattern in Spain, we dig a little deeper and give some directions for how we might understand this finding. On closer inspection, the diverging pattern for Spain is generated by the fact that Spanish demonstrators attend demonstrations much more with their partner and family, both being strong ties. No less than 77 percent of the Spanish demonstrators report attending the event accompanied by (at least one of) these two strong ties, while the average in the other countries is “only” 51 percent (the Spanish demonstrators differ significantly from all other nationals in this respect). The two most frequent recruitment combinations in Spain are being informed and persuaded by no-one while participating with family members (17.5%) and being informed, persuaded, and accompanied by family (9.8%). Another particularly frequent pattern in Spain is being informed/persuaded by friends but eventually attending with family (6.7%). So, it seems to be the case that the Spanish, much more than demonstrators in other countries, prefer to attend protest events in the company of their family. This plausibly points to a distinct underlying sociability in Spain, compared to the other countries. Note that this pattern is not just “South-European” as the Italians, the other Southern country in our data set, score rather low on the recruitment pattern scale (0.93) and Italians were not attending the events particularly frequently with their family (55%). Going to a demonstration with one’s family appears to be a specific Spanish thing to do. We can only speculate that Spanish families are maybe more politicized than families in the other countries under study or that families in general, beyond the political realm, play a larger role in the daily lives of the Spanish. Another possibility is that the explanation lies on a higher level and that the way, for instance, the Spanish police deals with protest is particularly family-friendly—although that seems not very likely. In any case, we think this comparative finding opens up interesting possibilities for theorizing and testing country differences with regard to recruitment patterns. Recruitment clearly is affected by broader cultural and social differences but we do not really know how.
Conclusion
Scholars have shown convincingly, and extensively, that recruitment is crucial to explain why some individuals participate in social movements and protest, while others do not. We also know that recruitment to a large extent takes place in interpersonal social networks. People tend to be recruited by people they know. Yet, there still is disagreement about who these interpersonal recruiters are exactly. Previous studies presented mixed results, with some (most) claiming that strong tie contacts—such as family members and friends—are the key recruiters, while others found that rather weak ties—like co-members or colleagues—are more important for recruitment into movement activism. We argued that both can be true at the same time: recruitment is a multidimensional process existing of an information, persuasion, and company dimension. While these three functions can be performed by the same ties—the basic homogeneous recruitment pattern—these different recruitment functions can also be performed by different sorts of ties. We argued that information may be acquired via both weak and strong ties, but persuasion and especially the company function of recruitment should be more a matter of strong ties. We theorized that individuals are more easily convinced by people they trust and know very well, and that they also prefer to have these kind of others as their company when they attend a protest event. This increasing role of strong ties through the recruitment process is what we called the “centripetal” recruitment pattern. Furthermore, we expected that the centripetal pattern would be more prevalent in demonstrations dealing with non-socio-economic issues.
These theoretical expectations were tested based on a large sample of 14,000 protest participants spanning 84 protest demonstrations in eight countries. The data showed that there are clear recruitment patterns and that recruitment functions are correlated with tie strength. First, the basic homogeneity pattern is evident in our data: many people are informed, persuaded, and accompanied by the same tie. Yet, at the same time, we see that the centripetal recruitment pattern is quite forcefully at work as well. If people are informed, persuaded, and/or accompanied by different types of ties, strong ties generally matter more for the company than for the information and the persuasion aspect. However, the expected, linear centripetal pattern was not entirely corroborated. While we observe a very clear difference between the company function, on the one hand, and the information and persuasion functions, on the other hand, we do not find the expected differences between the information and persuasion functions—people are not convinced by stronger ties to take part in a demonstration than they are informed about the upcoming event. Basically, overall, we see similar ties performing the information and persuasion functions. Maybe these two recruitment steps mostly happen at the same time; when someone informs another person about an upcoming event, this implies automatically that an attempt is undertaken to convince the other to take part in the event (why mention the event in the first place if one does not want the other to take part?). Our data suggest that recruitment may be a two-step process rather than a three-step process, and while information provision and persuasion may be analytically different things, in the real world they are often entangled.
The homogeneity and centripetal patterns were robust and held across the majority of the 84 events examined. Yet, at the same time, while homogeneous recruitment did not vary much across demonstrations, centripetal recruitment did. Demonstrations with non-socio-economic goals displayed the centrifugal pattern more than socio-economic demonstrations. Furthermore, of all eight countries in our study, Spain displayed a clearly diverging pattern with much more centrifugal recruitment (due to the fact that the Spanish prefer attending protest events with their family/partner).
We believe our study contributes in three ways to an already large literature on micromobilization. First, it solves an empirical inconsistency that has been central in the micromobilization debate and that pitched strong-tie against weak-tie studies. Our data suggest that both types of ties play a role in recruitment at the same time. Second, we offer an integrative theoretical framework that distinguishes several recruitment functions and that explains why and how some of these functions can be performed by what sort of social ties. Our framework advances an already pretty elaborate but not integrated thinking about the exact mechanisms of recruitment. Third, as the largest empirical study of micromobilization so far, we go beyond existing studies that mostly draw on meticulous case studies and show that the theoretically predicted and empirically found patterns apply to a large variety of protest events in eight countries while there are some interesting and explicable differences between protest issues and countries. In any case, we can be more confident with regard to the generalizability of our findings than those of previous work.
Still, a number of weaknesses remain. To start with the most important limitation, only drawing on actual participants, we cannot say which ties are most effective in performing which recruitment function. The only thing we can say is that, for people who were successfully recruited, some sorts of ties fulfilling some recruitment function are reported more frequently. But many recruitment attempts fail as people receive the information but are not persuaded, as people are persuaded but do not find suitable company, and as people might have been convinced to take part if they had had the information about an upcoming event, or any other combination of these factors leading to a no-show. Our design does not produce any information about that drop-out process. So, we cannot really say that certain social ties fulfilling some recruitment function actually lead to participation. Yet, we think our study, because of the measures that were used and the framework that has been developed, can lead the way for other studies that do have access to non-participants.
Second, we sketched a rather simplified picture of information, persuasion, and company as three comparable parts of the recruitment process. The company function, however, is essentially different from information and persuasion. Information and persuasion are performed before someone attends a protest event, while the company aspect takes place at the event itself. The dissimilarity of company also came forward in the results: we found starker contrasts for company versus information and persuasion than for information versus persuasion. Furthermore, we treated the three recruitment functions as one-dimensional phenomena, while they are actually dual processes. For instance, attending a demonstration with someone is not only the result of being asked (persuaded) by a contact, people can themselves ask others as well. Protesters can be both recruited and recruiter. This duality probably applies more to the company function than to the information and persuasion functions. Future research could investigate whether the same centripetal pattern is found if one looks at it from the perspective of the recruiter.
A final limitation is that we only take one type of protest activity into account: participation in legal (and large) street demonstrations. Methodologically, this was an advantage as it kept a good deal of possibly confounding factors constant. Yet, in terms of generalizing our findings, the fact that we only study one protest type is a disadvantage. In general, we would expect that for more high risk/cost activities, the role of strong ties becomes more important across the board (see, for example, Passy and Monsch, 2014). More trust, familiarity, and reciprocity are needed to engage in things that are costly and risky. Yet, even if the role of strong ties is more important to recruit high risk/cost activists than what we found here for street demonstrations in Western Europe, we still think that the differential role played by strong/weak ties across recruitment functions would remain. Even for disruptive forms of protest, information probably flows relatively more through weak ties while convincing potential activists to engage in high risk/cost protest, and, especially, the decision to engage in action with specific others, is a matter of stronger ties.
In conclusion, interpersonal recruitment for protest (and also for other types of activities) is a complex process. We probably need to go beyond the idea that some social ties are simply more important than other ties for recruiting individuals. It critically depends on the type of barrier standing in between non-participation and participation: different ties are functional in overcoming different sorts of barriers.
Footnotes
Appendix 1
Overview of the 84 sampled demonstrations.
| S. No. | Name | Issue | Country | Date | Turnout | Response rate (%) |
|---|---|---|---|---|---|---|
| 1 | Climate Change | Environment | Belgium | 5 December 2009 | 15,000 | 38 |
| 2 | March for Work | Austerity | Belgium | 29 January 2010 | 30,000 | 24 |
| 3 | 1st of May March | May Day | Belgium | 1 May 2010 | 2000 | 20 |
| 4 | No to Austerity | Austerity | Belgium | 29 September 2010 | 70,000 | 24 |
| 5 | No Government, Great Country | Democracy | Belgium | 23 January 2011 | 45,000 | 43 |
| 6 | Non-profit Demonstration | Austerity | Belgium | 29 March 2011 | 15,000 | 29 |
| 7 | Not in Our Name | Democracy | Belgium | 7 May 2011 | 700 | 27 |
| 8 | We have alternatives | Austerity | Belgium | 2 December 2011 | 70,000 | 20 |
| 9 | Fukushima never again | Environment | Belgium | 11 March 2012 | 1000 | 31 |
| 10 | National Manifestation | Austerity | Belgium | 6 November 2014 | 120,000 | – |
| 11 | Prague Pride | Gay Pride | Czech Republic | 18 August 2012 | 9000 | 19 |
| 12 | Stop the Government | Austerity | Czech Republic | 17 November 2012 | 6000 | 28 |
| 13 | The End of Godfathers | Democracy | Czech Republic | 7 April 2013 | 500 | 18 |
| 14 | Czech Communist Party May Day | May Day | Czech Republic | 1 May 2013 | 1000 | 19 |
| 15 | In defense of academic and civil freedoms | Democracy | Czech Republic | 23 May 2013 | 300 | 27 |
| 16 | Veggie Parade 2013 | Environment | Czech Republic | 15 June 2013 | 300 | 19 |
| 17 | Prague Pride 2013 | Gay Pride | Czech Republic | 17 August 2013 | 10,000 | 13 |
| 18 | DIY carnival 2013 | Democracy | Czech Republic | 28 September 2013 | 1000 | 6 |
| 19 | Demonstration against extreme-right | Antiracism | Czech Republic | 17 November 2013 | 600 | 18 |
| 20 | “Velvet” feast 2013 | Democracy | Czech Republic | 17 November 2013 | 100 | 20 |
| 21 | Euromayday | May Day | Italy | 1 May 2011 | 5000 | 11 |
| 22 | May Day | May Day | Italy | 1 May 2011 | 500 | 23 |
| 23 | General Strike | Austerity | Italy | 6 May 2011 | 15,000 | 20 |
| 24 | Demonstration Perugia-Assisi | Antiracism | Italy | 25 September 2011 | 150,000 | 23 |
| 25 | Gay Pride | Gay Pride | Italy | 9 June 2012 | 15,000 | 20 |
| 26 | No Monti Day | Austerity | Italy | 27 October 2012 | 25,000 | 18 |
| 27 | Joining forces for another Europe | Austerity | Italy | 10 November 2012 | 3000 | 12 |
| 28 | Seeds of Justice | Antiracism | Italy | 16 March 2013 | 30,000 | 19 |
| 29 | No Mous | Environment | Italy | 30 March 2013 | 10,000 | 12 |
| 30 | Retirement demonstration | Austerity | Netherlands | 21 November 2009 | 7000 | 36 |
| 31 | Climate demo | Environment | Netherlands | 12 December 2009 | 3500 | 36 |
| 32 | Student demo 1 | Austerity | Netherlands | 21 May 2010 | 2000 | 29 |
| 33 | Culture demo Amsterdam | Austerity | Netherlands | 20 November 2010 | 15,000 | 45 |
| 34 | Culture demo Utrecht | Austerity | Netherlands | 20 November 2010 | 2500 | 42 |
| 35 | Student demo 2 | Austerity | Netherlands | 21 January 2011 | 15,000 | 29 |
| 36 | Together strong for public work | Austerity | Netherlands | 17 February 2011 | 8000 | 32 |
| 37 | Stop racism and exclusion | Antiracism | Netherlands | 19 March 2011 | 350 | 27 |
| 38 | Anti-nuclear demo | Environment | Netherlands | 16 April 2011 | 2500 | 44 |
| 39 | Military demo | Austerity | Netherlands | 26 May 2011 | 4000 | 22 |
| 40 | Stop budget cuts (care and welfare) | Austerity | Netherlands | 19 September 2011 | 4500 | 26 |
| 41 | Occupy Netherlands | Austerity | Netherlands | 5 November 2011 | 120 | 16 |
| 42 | Pink Saturday Parade Survey | Gay Pride | Netherlands | 7 July 2012 | 1000 | 22 |
| 43 | Demonstration against language decree | Democracy | Spain | 21 January 2010 | 40,000 | 30 |
| 44 | Against the Europe of Capital, Crisis, and War | Austerity | Spain | 28 January 2010 | 1500 | 24 |
| 45 | Demonstration Against Abortion | Women | Spain | 7 March 2010 | 65,000 | 30 |
| 46 | 1st May, Labor Day | May Day | Spain | 1 May 2010 | 8000 | 22 |
| 47 | Self-determination is democracy | Democracy | Spain | 12 June 2010 | 10,000 | 40 |
| 48 | Demonstration against the new labor law | Austerity | Spain | 30 June 2010 | 10,000 | 20 |
| 49 | We are a nation, we decide | Democracy | Spain | 10 July 2010 | 1,000,000 | 27 |
| 50 | Against Labor Law | Austerity | Spain | 29 September 2010 | 55,000 | 32 |
| 51 | Celebration May Day | May Day | Spain | 1 May 2011 | 15,000 | 22 |
| 52 | Defend Our Rights | Austerity | Spain | 1 May 2011 | 2000 | 44 |
| 53 | Real Democracy Now! | Democracy | Spain | 15 May 2011 | 25,000 | 39 |
| 54 | Climate March | Environment | Sweden | 12 December 2009 | 40,000 | 28 |
| 55 | May 1 March, Left Party | May Day | Sweden | 1 May 2010 | 4200 | 32 |
| 56 | May 1 March, Social Democratic Party | May Day | Sweden | 1 May 2010 | 3000 | 32 |
| 57 | Against racist politics | Antiracism | Sweden | 4 October 2010 | 2800 | 25 |
| 58 | Anti-nuclear demonstration | Environment | Sweden | 26 April 2011 | 1000 | 35 |
| 59 | May Day (Left Party) | May Day | Sweden | 1 May 2011 | 2000 | 31 |
| 60 | May Day (SAP/LO) | May Day | Sweden | 1 May 2011 | 900 | 30 |
| 61 | May Day (Left Party) | May Day | Sweden | 1 May 2012 | 3300 | 34 |
| 62 | May Day (Social Democratic Party/LO) | May Day | Sweden | 1 May 2012 | 2450 | 28 |
| 63 | Rainbow Parade (LGBTQ festival) | Gay Pride | Sweden | 3 June 2012 | 3300 | 29 |
| 64 | Stockholm Pride | Gay Pride | Sweden | 2 August 2014 | 60,000 | – |
| 65 | World March of Women | Women | Switzerland | 13 March 2010 | 6000 | 32 |
| 66 | May 1st Demonstration | May Day | Switzerland | 1 May 2010 | 8000 | 13 |
| 67 | May 1ste demonstration 2011 | May Day | Switzerland | 1 May 2011 | 1000 | 20 |
| 68 | Women demonstration Geneva | Women | Switzerland | 14 June 2011 | 1500 | 32 |
| 69 | Anti-nuclear Manifestation | Environment | Switzerland | 22 June 2011 | 20,000 | 45 |
| 70 | Gay Pride Geneva | Gay Pride | Switzerland | 2 July 2011 | 3000 | 21 |
| 71 | Anti-nuclear | Environment | Switzerland | 11 March 2012 | 3000 | 45 |
| 72 | Pride demonstration | Gay Pride | Switzerland | 16 June 2012 | 3000 | 27 |
| 73 | National Climate March | Environment | United Kingdom | 5 December 2009 | 50,000 | 36 |
| 74 | May Day Labor March | May Day | United Kingdom | 1 May 2010 | 5000 | 15 |
| 75 | Take Back Parliament | Democracy | United Kingdom | 15 May 2010 | 2000 | 31 |
| 76 | No to Hate Crime Vigil | Antiracism | United Kingdom | 23 October 2010 | 2000 | 13 |
| 77 | Unite Against Fascism National Demo | Antiracism | United Kingdom | 6 November 2010 | 3000 | 17 |
| 78 | Fund Our Future: Stop Education Cuts | Austerity | United Kingdom | 10 November 2010 | 30,000 | 13 |
| 79 | National Climate March 2010 | Environment | United Kingdom | 4 December 2010 | 1500 | 31 |
| 80 | Second Student National Demo | Austerity | United Kingdom | 9 December 2010 | 40,000 | – |
| 81 | Million Women Rise | Women | United Kingdom | 5 March 2011 | 3000 | 15 |
| 82 | TUC’s March for the Alternative | Austerity | United Kingdom | 26 March 2011 | 250,000 | 12 |
| 83 | Occupy London | Austerity | United Kingdom | 12 November 2011 | 600 | 12 |
| 84 | London Pride Parade | Gay Pride | United Kingdom | 7 July 2012 | 20,000 | 17 |
Appendix 2
In the presented analyses, we use five categories of ties. We combine friends and acquaintances in one category because we are not able to differ between these ties for the information function. We therefore run some extra analyses in which we do differ between friends and acquaintances for the persuasion and company function. We calculate the recruitment pattern score in the same way as we did before (giving respondents a +1 when a subsequent mobilization function was performed by a stronger tie and a –1 when it was performed by a weaker tie), but we now use six different categories of ties: no ties (0), co-members (1), colleagues/co-students (2), acquaintances (3), friends (4), partner/family (5). Respondents who said they were informed by friends/acquaintances were now all coded as having been informed by friends (4). Table 6 shows the frequency distribution for the variable we used in the article (five tie categories) and the same variable but now with six tie categories.
Table 6 indicates that more respondents do not at all follow the pattern we expect when using six tie-categories variable (100 vs 52 who have a score of –3). Yet, this is still less than 1 percent of the protesters. In general, the distribution of the variable does not change a lot. As a consequence, the results of the regression analyses presented in Table 4 also only very slightly differ if we use the six-categories variable as the dependent variable (only very minor changes in sizes of effects).
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Evidence was gathered in the framework of the CCC project. This project was funded by the European Science Foundation under grant 08-ECRP-001 and funded by the National Science Foundations of the Netherlands (Bert Klandermans), Belgium (Stefaan Walgrave), Italy (Donatella della Porta), Sweden (Abby Peterson), Spain (José Manuel Sabucede Cameselle and Eva Anduiza), the United Kingdom (Christopher Rootes and Clare Saunders), and Switzerland (Marco Giugni). Information on the project is available at 
(project proposal, manual for data collection, etc.). See a special issue of the journal Mobilization (vol. 17, no. 3) for an introduction to project framework, design, and data.