The Structure of Online Activism

Kevin Lewis, Kurt Gray, Jens Meierhenrich

Sociological Science, February 18, 2014
DOI 10.15195/v1.a1
Despite the tremendous amount of attention that has been paid to the internet as a tool for civic engagement, we still have little idea how “active” is the average online activist or how social networks matter in facilitating electronic protest. In this paper, we use complete records on the donation and recruitment activity of 1.2 million members of the Save Darfur “Cause” on Facebook to provide a detailed first look at a massive online social movement. While both donation and recruitment behavior are socially patterned, the vast majority of Cause members recruited no one else into the Cause and contributed no money to it-suggesting that in the case of the Save Darfur campaign, Facebook conjured an illusion of activism rather than facilitating the real thing.

Kevin Lewis: Department of Sociology, University of California, San Diego. E-mail:

Kurt Gray: Department of Psychology, University of North Carolina, Chapel Hill. E-mail:

Jens Meierhenrich: Department of International Relations, London School of Economics and Political Science. E-mail:

  • Citation: Lewis, Kevin, Kurt Gray, and Jens Meierhenrich. 2014. “The Structure of Online Activism.” Sociological Science 1: 1-9.
  • Received: September 16, 2013
  • Accepted: October 16, 2013
  • Editors: Jesper Sørensen, Sarah Soule
  • DOI: 10.15195/v1.a1

, , , , , ,

10 Reactions to The Structure of Online Activism

  1. Zeynep Tufekci February 18, 2014 at 10:46 am #

    This article provides a fascinating window into how Save Darfur failed to convert clicks to donations, but I believe an important qualification is necessary. This article tests an “online only” outcome of a case that functions more as a celebrity-driven cause rather than a movement like the Civil Rights Movement, or Occupy Wall Street or the Arab uprisings that began in 2011.

    Social media obviously will have different mediating impacts on movement engagement depending on conditions and goals of collective action. Variables such as whether the collective action undertaken under the banner of participants’ own interests (like Civil Rights Movement, Arab Uprisings, Turkey’s Gezi protests, or Occupy Wall Street) or a cause/solidarity campaign conducted in rich nations in the name of other, distant (often brown) people who need saving (like “Save Darfur” or “Stop Kony”); levers of impact (whether the issue is local, national or global); openness of the public sphere and the role of censorship; the level of repression; and degree of online/offline integration of action (which tend to be fairly little in online cause campaigns such as the one here) all have significant consequences for role of social media in civic engagement.

    With regard to testing on successful cases, it’s true that most movements fail. That’s always been true. Selecting on the dependent variable (success) allows us to examine necessary but not sufficient causes; selecting on unsuccessful movements, on the other hand, do not allow for rejecting necessary or potential causes. (In other words, social media may well be a gateway to civic action in some successful cases; however, that claim is not negated by the finding that it is not a gateway to civic action in non-successful cases. It’s practically impossible for movement success to be monocausal).

    Finally, I think this study highlights a deficiency that many studies in this field share. Too many look at only online or only offline components of behavior as if those are two separate universes. Do we know how many high school students who never considered donating to Save Darfur, since that age group rarely does, nevertheless became engaged participants in the civic sphere after being introduced to politics through this movement? I’m not making a claim but highlighting measures we need.

    I’m convinced by these findings that “Save Darfur” cannot convert clicks to donations, though I’m not clear on how more donations would have allowed Darfur to be saved, and it’s possible many participants felt the same way. Donations or other online-only (worse, one platform-only) behaviors as outcome measures are frustratingly limited windows into civic behavior. I emphasize this as a comment on the field rather than only to this paper.

    Finally, even as I remain unconvinced that the (otherwise very interesting) findings provide as broad a result as the authors indicate in their conclusion, I consider this paper to be a valuable contribution to the sizable literature of empirical studies of online activism. Congratulations to the authors and Sociological Science! May open access and debate thrive!

    • Kevin Lewis February 28, 2014 at 11:32 am #

      Thank you for your reaction, Zeynep. We are grateful that our work has generated so much interest. You raised excellent concerns about 1) offline behavior, 2) causality, and 3) case selection. Let us briefly address each.

      Offline Behavior

      As we acknowledge in the article, our data are indeed limited to online behavior and do not preclude the possibility that Cause participants were active offline (in fact, it is perhaps worth mentioning that our larger project—on the Save Darfur movement more generally—aims to link analyses of online AND offline behavior, precisely for the reasons you mention). On the other hand, it stands to reason that online donations, or the lack thereof, can have very real offline consequences, blurring the distinction between the two types of behavior; and given how little time, effort, and money were required to recruit just a single other person into the Cause or to donate just a single dollar, it is difficult to imagine that Cause participants who were unwilling to undertake either of these basic activities were nonetheless contributing to Darfur in other substantial yet unobserved ways. In other words, just because our window was “frustratingly limited” does not necessarily mean that there was more to see. This is the POSSIBILITY that we would like to raise; but this is also of course an empirical question, and we are certainly interested in substantiating our hunch in future work.


      We agree that the impact of social media on political engagement may vary depending on whose interests are being served, the openness of the public sphere, the degree of online/offline integration, and various other characteristics of the movement, the medium, and the socio-historical context. Human movements are multi-faceted and incredibly complex, and success or failure are unlikely to be monocausal. Unfortunately, any form of research requires simplifying empirical complexity, and we attempted to distill the richness of online activism into one parsimonious framework, all the while cognizant of the fact that a comprehensive account of social activism in our case requires the integration of a number of different research designs—not to mention the integration of network and qualitative data. We also look forward to future research exploring more generally the moderating variables of the general phenomenon we have uncovered—the illusion of effectiveness in online activism. That said, we are interested to hear more about the claim that “selecting on the dependent variable” allows one to examine causality of ANY sort, necessary or sufficient; and we regret that we provided insufficient detail in the article to understand the full context and motivation of our own case selection strategy.

      Case Selection

      We selected the case of the Save Darfur Cause on Facebook because one of us was pursuing research on the movement’s offline behavior. Yet seeing that online behavior (and not just, or even primarily, on Facebook) was believed to be central to this particular movement’s diffusion, we set about to devise a research design that would allow us to capture a slice of cyberactivism in this case, including its nature, structure, and dynamics. Put differently, we did not select the case of the Save Darfur Cause on Facebook because it was “unsuccessful.” In fact, we were totally unaware of the level of recruitment and donation behavior among its participants before the data arrived. The framing of the article, therefore, parallels our actual aims—aims that were based around addressing an open empirical question: What role for Facebook in the making of a social movement?

      We were as surprised at the findings as many readers probably have been or will be, and felt that these results warranted attention. We would therefore like to optimistically suggest that our results (again, based on a massive social movement, a ubiquitous social medium, and an urgent social issue) might carry just a bit more weight than many prior studies—studies that focus on cases where social media WERE important for civic engagement and then use these data to support the conclusion that social media ARE important for civic engagement. There are three more reasons why our data deserve to be taken seriously. First, our data reflect actual behavior, rather than commonly used self-report measures of activism. Second, our data include precise and nuanced network information, which is lacking in most studies. Third, these behavioral and network data were collected for an entire, enormous population of interest, rather than the typically used small convenience samples. Although Facebook was but one of multiple online tools that served anti-genocide activism in the case of the Save Darfur movement, its modest role raises important questions about the specter of online activism.

  2. Alex Hanna February 20, 2014 at 7:03 am #

    In this piece, Lewis et al. attempt to understand the “nature and scale of the typical activist’s invovlement with social media” in the Save Darfur Cause on Facebook. They look into the distribution of Facebook activism, its overtime trends, and the effects of networks on mobilization.

    Some of the very interesting elements of this article — longitudinal network data of recruitment for a social movement is a bit of a dream. Having this kind of granuality of data in different contexts would be incredible. Also interesting (although similar findings have been shown in the literature) are the findings that most people were recruited by others, and that the amount donated depended highly on whether one was recruited or joined independent. I particularly find fascinating the odds that a group member was 610% more likely to donate if their recruiter also donated.

    There are some questions and negatives of the piece. Stylistically, the article’s setup is needlessly self-important — “empirical studies of online activism are surprisingly scarce”, which is patently false with a cursory look at literatures in communications, political science, and the growing field within sociology; “rivaling the US civil rights movement” in size, which is a bit of an apples to oranges comparison since there are varying definitions of participation in each.

    Beyond that, the finding that there is a power law distribution in contribution isn’t particularly new, and the citation to Oliver’s work on “hyperactivists” and differential contributions is indicative here. This sort of phenomenon persisted before the internet and may even be exacerbated by it.

    Two other outstanding questions — in figure 2, what caused random spikes in donations? Wondering what kind of exogenous elements may have caused an uptick. Secondly, what may be the causal mechanism in the finding that people were much more likely to donate if their recruiter donated? Is it possibly the case that there’s some homophilic effect — there is a class of people who have the capital to donate and since their network is full of donors they also donate? Or is this an influence effect? Can the recruited see that their recruiter donated?

    • Jens Meierhenrich March 5, 2014 at 12:38 am #

      Thank you for your comment, Alex, and your kind words about the longitudinal dataset that we pieced together. We are glad that you share our enthusiasm for its uniqueness and the granularity of the data available in it. You also raised substantive questions about 1) the existing literature, 2) the role of hyperactivists in relation to member contributions, and 3) causality. Let us briefly address each.

      Existing literature

      You wonder whether our claim that “empirical studies of online activism are surprisingly scarce” might be an overstatement. When we started this research in 2009, our claim was certainly truer than it is today. Nonetheless, we do not think that it is “patently false” to suggest that good and rigorous scholarship is in rather short supply, even now. Empirical analyses on the scale made possible by our longitudinal data are still rare. But even less ambitious research designs are perhaps not quite as numerous as you purport—especially in the sociological literature. Having said that, we would like to benefit from your expertise. Since this is an interactive forum, it would be very helpful for us—and perhaps also for other readers of Sociological Science—if you publicly shared the ten or so leading empirical studies of online activism. Since ours is, as we mentioned in response to Zeynep, an ongoing research project, we would be keen to situate ourselves better in the existing literature than you think we have.


      We agree that the finding of differential contribution certainly is not novel; though here, also, we would benefit from your direction to specific literature documenting power law distributions (or distributions of any sort) of donations—as again, our sense is that data such as ours are relatively scarce. The fact that the phenomenon of hyperactivists “persisted before the internet” is interesting, but, of course, of little import for our analysis, which is solely concerned—at least in this article—with online behavior. If indeed the distribution of contributions is similar online as offline (assuming apples-to-apples comparisons, i.e. representative data on all contributions and accurate data on all contributions), despite the myriad differences between the two types of activism that have been the focus of so much attention—is this not itself a worthwhile finding to report? And in the online context, hyperactivists have arguably not yet received the kind of scholarly attention that they deserve, which is precisely why we alerted our readers to the coordinating roles they played in the case of the Darfur Cause on Facebook. However, we do agree with you that it would be worthwhile to ascertain whether the apparent centrality of hyperactivists in movement recruitment “may even be exacerbated” by the Internet, as you suspect.


      As to the spikes in figure 2, we are confident that some of these spikes are more or less “random.” On one hand, why a single user suddenly decided to contribute $1,500 on the 230th day of the cause (contributing to one such spike) is probably not worth a great deal of our speculation. On the other hand, broader surges of activity—in which any such spike may be embedded—are almost certainly meaningful. One obvious possible explanation would point to events and circumstances entirely “external” to the cause—e.g. events taking place in Darfur or in the broader socio-political landscape. Another would point to events exogenous to individual users but still endogenous to the online movement—for instance, the organizers of the cause might have initiated an e-mail campaign to try to encourage participants to donate (we are aware that such campaigns exists, but do not yet have data on their timing). A third would point to internal “cascades” of behavior, i.e. dynamics totally endogenous to the network. Regrettably, we do not consider any such explanations in this article—but we are excited to pursue these possibilities in our ongoing research.

      As to the causal mechanism linking the donation behavior of donors and that of their recruiters, your guess is as good as ours, and possible explanations are not limited to the mechanisms you suggest (see, for instance, the great introductory section in Shalizi and Thomas 2011, which discusses all of the many possible reasons you might jump off a bridge if your friend jumps off, too). In terms of disentangling these possible explanations, the longitudinal precision of our data is certainly a strength and makes some form of modeling very tempting; though given that we do not have data 1) on actors’ covariates or 2) the broader population of Facebook users who COULD have been recruited to the cause but were not, we fear that such an effort would be doomed from the start. Selection undoubtedly played some role (as it does in virtually every study of network homogeneity), and influence would indeed be a likelier suspect if donation behavior was amplified in the network (we are not aware that it was).

  3. Brayden King February 27, 2014 at 2:25 pm #

    As Alex said, the authors’ statement that “empirical studies of online activism are surprisingly scarce” is just not true. Unfortunately, the authors missed an opportunity by overlooking this growing literature. Most research about online activism tends to have an overly-sunny outlook, offering theories about the advantages that social media provide to activists and glorifying technology, as illustrated by a fascination with hyperlinking and retweets. In contrast, this analysis shows that online activism is weakly associated with deeper forms of movement participation. This finding is a nice addition to the debate about the consequences of online activism and provides evidence that online movements may deter more traditional forms of movement participation. More engagement with the literature on online activism would have showed just what a significant finding this is!

    I also want to draw attention to another study that supports their findings and offers a psychological mechanism to explain why participating in an online activist group might deter serious engagement with the cause. Kristofferson et al.’s (2014) paper in the Journal of Consumer Research ran an experiment in which they showed that people who join a public Facebook activist group (public = their friends can see they joined) were less likely to agree to help stuff envelopes to support the campaign than were individuals who privately joined the group. Only when people were encouraged to think about how their values aligned with the cause did the difference between the public and the private joiners disappear. The authors suggest this is because publicly joining online groups satisfies impression management needs (and encourages “slacktivism”) but doesn’t necessarily lead people to demonstrate behavioral consistency with their values. The two papers together provide compelling evidence that slacktivism might be more common than other scholars of online activism have believed and provide an explanation for why this might be happening.

    • Kurt Gray March 5, 2014 at 7:28 am #

      Brayden: Thank you for your thoughtful reaction. Regards to the novelty of our studies—some past research has indeed investigated online activism, but as Kevin and Jens suggest above, these studies offer only limited insights into the structure of these movements. We agree wholeheartedly that past work has often painted a rosy picture of online movements, implying that social media holds the keys to salvation. As you note, our data offers a corrective perspective to the glorification of technology (to paraphrase your compelling sentiment).

      Thanks very much for pointing out the Kristofferson et al. (2014). As you note, this paper provides an excellent investigation into the psychological mechanisms for why people are such “slackers” in social media. Why pay the costs of activism, when you can reap the reputational rewards for free? We look forward to more studies—and the healthy debate—generated by questions of whether online social activism is better, worse or no different than offline activism.

  4. Dave Karpf March 5, 2014 at 8:11 pm #

    This is an interesting empirical study, and I’ll echo Alex’s note that longitudinal data on movement activation is just dreamy. One major problem stood out for me, though: the authors are equating the Save Darfur “Cause” with the Save Darfur Movement. is a for-profit company. Lead generation is its business model. It is, in this sense, a competitor with (also a for-profit “b corp”).

    The business of is to identify potential supporters of the Free Darfur Coalition (or the Sierra Club, or Americans for Responsible Solutions). Causes *also* allows people to launch their own campaigns (thus attracting more potential supporters to the site), but the company’s business model is selling names to existing organizations so that they can fundraise. The site architecture and online presence is optimized for one-time actions*.

    During the time of your study, Causes had a slightly different format (it was launched as Facebook Causes). But it was *still* a for-profit company with a business model that emphasized breadth over depth. And, lo and behold, your data shows that they were achieving breadth over depth.

    Normatively, I see nothing wrong with the Causes business model. Free ridership remains a big problem for social movements/advocacy organizations. Direct mail fundraising (also a loss-leader with heavy communication flows and tiny response rates) was a major revenue stream for many organizations and is now in steep decline. This is causing a disruption in the advocacy sector (I discuss this all in my book, The MoveOn Effect, incidentally). These companies are filling a market demand and providing a service to nonprofits that I personally admire. That’s fine.

    But I don’t think scholars should treat these companies as black boxes. They are actively shaping the user experience. What looks like “slacktivism” from the user-only perspective looks like a business plan from an alternative perspective.

    A few elite interviews with the people who launched the Cause, with staffmembers at Causes, and with staffmembers at related social movement organizations would dramatically enhance this project. Might make for an interesting follow-up piece, if you have the taste for it.

  5. Bridgit Evans March 13, 2014 at 2:08 pm #

    I worked as a culture change strategist within the Save Darfur movement from 2003 to 2006, and I have to agree with Alex: “One major problem stood out for me, though: the authors are equating the Save Darfur “Cause” with the Save Darfur Movement.”

    In general, I find the analysis in this article to be predicated on two problematic assumptions: 1) that Facebook Causes (later “Causes”) was the gold standard (preferred) platform for online activism in relationship to the Save Darfur movement; and 2) that the period from 2008-2009 was the most representative period of online (and offline) activism for this movement to be assessed in your research.

    I think most people in the online mobilizing space would agree that Facebook Causes was ultimately not a very effective online mobilizing platform in general — it’s why so many organizations began to rapidly migrate to other Facebook platforms (Pages), social networks (Twitter,, direct messaging platforms (Constant Contact) and funds collection services (Network for Good) in 2009/10 to support their online mobilizing. In my universe of discussions amongst campaigners venturing into the online space in the mid/late ’00s, it was commonly known that it was near impossible to generate donations via Causes (people just never trusted the idea of donating on Facebook). Instead, best practices recommended that organizations find creative ways to migrate social network ‘friends and fans’ into email lists in order to maximize their online donation and social action potential. Indeed with one client in 2009, we dedicated significant resources to transitioning their Causes audience into their Pages community, which they found to be far more effective at engaging people in digital activism, and into their email list, where they had most success with donations. Does your study address the fact that — in general and not specific to Save Darfur — Causes was wildly viewed as a troubled platform that didn’t yield the user engagement (actions, donations, recruitment) it originally promised?

    Another helpful thing to consider: the window of time covered in the study — 2008 to 2009 — is after the peak in growth of the public’s activism around the Darfur crisis — and its rates of engagement.

    2006 was truly the peak year for the American public’s activism on this issue, in my opinion. In 2006, when Save Darfur completed its 6-month “Million Voices for Darfur” postcard campaign, they had achieved a movement milestone: concrete social action, both online and offline, from over 1,000,000 Americans (who signed postcards in support of the appeal to President Bush to better protect the people of Darfur). They also realized two iconic indicators (at the time) of effective online and offline base-building: a rally on the National Mall with over 100,000 people from across the country that garnered global media attention, and an email list of 1M+ subscribers that was regularly mobilized by movement leaders to demonstrate public will. These numbers — and their visible expressions in the rally crowds, boxes of postcards delivered to the White House, etc. — gave a clear sense of majoritarian demand for a Congressional/Presidential response to the Darfur crisis.

    Within this frame, the movement’s mobilization goals were met. Was Darfur ‘saved’? Absolutely not — but that’s a rather complex matter for another study, I think you’ll agree. But this movement did create the political climate that made the signing of the Darfur Peace and Accountability Act in late 2006 more possible.

    After this, a great deal of the advocacy work was in the hands of human rights advocates, lawmakers and UN leaders, and I’d wager to guess that this accounts for some of the reduction in online activism and donations from everyday people (vs. long-committed activists).

    If an evaluation of the effectiveness of Causes to spur online action and donations is of interest, it might be worthwhile to consider movements who experienced their peak periods of online activism during the window you’ve isolated. Perhaps the human trafficking movement in 2008-2010. ECPAT had a similar online support base on Causes in 2009, a period of significant activism around the human trafficking issue, they might be an interesting comparable:

    Happy to discuss further, if helpful.

  6. Rense Corten May 8, 2014 at 11:03 am #

    Very nice work, and good to see the discussion flourish here! I just wanted to point out that some of the results – namely that those who joined independently are more likely to donate than those who were recruited – seem consistent with the findings of Parigi et al (2012, link below) on Couchsurfing, which is also in some ways an online social movement.
    In that paper, using also detailed temporal data on members’ activities and social networks, we found “a significant impact of new friendship ties on participation, compared to a negligible impact of pre-existing friends, defined here as ties to other members formed outside of the organization’s context.”

    Our theoretical interpretation is that CouchSurfing succeeded in forming a strong identity among its members, thereby reducing the importance of embedding participation within pre-existing networks to foster participation. Might something similar be going on here.

    The paper is here:

  7. Diane McMahon November 13, 2016 at 6:03 am #

    Donating funds parallel both a sense of urgency and a clearly established fact that the funds will make a difference. In this case “How will donated funds save Darfur?” If the organization cannot verify how this will happen it will be very difficult to raise funds for the cause. People give to people – not to websites – or appeals for money made through the internet. There is data to this effect collected by the international organization –