Making sense of online mobilisation and radicalisation

Making sense of online mobilisation and radicalisation


Resonant Voices Initiative gathers individuals and organisations pushing back against anti-democratic forces that weaponise social media in an attempt to manipulate and distort reality, to sow mistrust, and to spread fear. These malicious actors – from terrorists and violent extremists to foreign influence operators and autocratic governments – recruit and mobilise supporters, intimidate and harass opponents, and suppress dissent through amplification of content and exploitation of system and audience vulnerabilities using tools built by social media platforms.

The system vulnerabilities as well as the malicious actor exploitation strategies have received a lot of attention and solutions ranging from regulation to platform policy changes have already been proposed in response.

So far, these efforts have focused primarily on tackling terrorist content – mostly of ISIS provenience – Russian political influence, and, in more recent years, right-wing terrorism. While these are the low-hanging fruits, they represent the extreme end of the spectrum. To address the problem of online radicalisation effectively, the solutions must not only be designed to address the supply side – the exploitation by malicious actors and their “content” or “conduct” by banning or removing it. They must also consider the demand side, that is, the audience of this messaging.

Communication-based responses to social media manipulation attempts (commonly referred to as counter-, alternative, or positive narratives) do consider the audience’s perspective. This practice has increased in recent years. Nevertheless, insufficient attention has been given to assessing the effectiveness of these narrative pushback strategies, particularly in light of audience vulnerabilities and biases.

This is in part because success in this endeavour lacks simple metrics. So far, those working on strategic communication campaigns have utilized social media marketing approaches from commercial and political campaign contexts as their guide. However, unlike in those sectors, audience receptiveness to counter-narratives has no clear behavioural outcome, like voting or making a purchase. For this reason, the effectiveness of these efforts can be especially challenging to assess.

Our Research

Measuring Online Mobilisation

In simple terms, the Online Mobilisation Score (OMS) is based on the following components of online mobilisation:

  1. Action-readiness: (self-declared) likelihood the person will engage in an offline action, specifically volunteering, donating, promoting, participating in protest, etc., in relation to the issue they are concerned about
  2. Emotion activation: (self-declared)
  3. Action rationale: perception of a need for social change, specifically a change to the economic, political, and/or socio-cultural status quo on behalf of the marginalised groups affected by the issue of concern.

Based on the model, we were able to estimate the OMS per participant, which then allowed us to show average variance of each individual or group from the mean OMS of the entire sample. The overall average scores for OMS, need for change, and action-readiness were defined as 0, and the variations are calculated relative to the overall average, where an individual’s positive OMS value indicates he or she has higher than average online mobilisation (as understood from the three dimensions described above), and an individual’s negative OMS value indicates he or she has lower than average online mobilisation.

Each person in the dataset is comparable on the basis of the OMS as a complex factor, as well as on the basis of the component factors of OMS, action-readiness and belief in change, and the binary variables of anger and frustration reported in reaction to online content.

Figure 1 shows the differences between respondents in Austria, Croatia, and Germany in their average OMS values along with a breakdown of average scores in the action rationale and action-readiness metrics, as well as numbers of those who felt angry or frustrated about the content they find online regarding their issues of concern.

Figure 1. Average OMS and its components for all narratives

As the table shows, respondents in Croatia had a higher average OMS, including higher action rationale and action-readiness scores, as compared to those of respondents in the other two countries.

There are three important caveats related to interpretation of this score to note at this point:

  • The sample in our data is currently not representative of the country’s population, as that was not the objective of the research. This limitation could be achieved with building a probability sample, for those interested in findings that can be extrapolated to the country’s population.
  • While it serves as a useful quantitative reference point, the OMS can only be meaningfully interpreted in relation to Testing this was beyond the scope of the current research.
  • The research could potentially be replicated with a narrower issue-specific focus – collecting only micro-narratives related to a particular issue (e.g. corruption, environment, gender equality). This approach would be suitable for researching mobilisation on a particular issue on either a probability or purposive sample of respondents.

Analysing Narratives

In the next step of the analysis, we coded the micro-narratives describing issues of concern that were volunteered by respondents, linking them to broader narratives and debates.

Because our research was designed to explore issues volunteered by respondents rather than specific issues defined by the research, the range of collected narratives is broad and narratives cannot always be grouped into discrete categories, as they would be in a ranking or single choice issue salience questions.

The open-ended nature of the question and a longer narrative response required offers a unique insight into how the problem is framed by the participant. This was followed by a battery of questions allowing the respondents to provide further details in relation to their understanding of their own micro-narrative and the issue it represents, in addition to calculating their online mobilisation score.

The dashboard below shows average OMS estimates per cross-section of key demographic groups and the positioning of each respondent on the spectrum within countries.

As can be seen in the data above, three specific groups seem to be the most motivated to engage in a political action related to issues of their concern, mediated by online content triggering their anger and/or frustration.

Young women in Croatia feel their rights are at risk: The group with highest online mobilisation score were women in Croatia aged 18-30 in our sample. The issue they are most concerned about is the anti-gender movement, including anti-abortion/pro-life protests in their countries and abroad, violence against women, prejudice against certain groups of women, such as those wearing a headscarf, and a perceived threat from right-wing radicalisation in general.

Middle-aged diaspora men in Austria are frustrated about inequality: Another group with higher online mobilisation score relative to the rest of the sample are men belonging to the Balkan diaspora in Austria aged 31-50, though their concern spectrum is wider, from income inequality, the treatment of immigrant workers in Austria as second-class citizens, experienced or observed racism and xenophobia, to climate change and a recent corruption scandal.

Men over 50 in Croatia stress over corruption: Although not as pronounced as in the first two groups, men over 50 in Croatia have a high online mobilisation relating to their concern over corruption and abuse of public resources, although mistreatment of migrants also appears to be one of the concerns for that age group.

The resulting report explores the interplay between the surveyed audience’s online mobilisation levels and their micro-narratives on their issues of concern.

Figure 2. Online mobilisation score across demographic categories

Threat to Democracy

Our analysis model suggests that those respondents with higher online mobilisation scores are individuals ready to take action to effectuate change on their cause of choice, driven by a desire to topple the status-quo, stimulated by online content about their cause that leaves them feeling angry and frustrated.

While the action they decide to take can be as benign as rescuing a stray animal, or organising garbage cleaning in the neighbourhood, it may make individuals vulnerable under certain circumstances because:

  • Such individuals are susceptible to the stimuli of online content, which may be a vulnerability exploited by malicious actors; or this susceptibility might signal existing exploitation;
  • They are experiencing frustration and anger in relation to the problem they care about and are compelled to do something about it, presumably because they feel the institutional or community response is inadequate;
  • They are likely to engage offline which exposes them to multiple risks, such as their action not being successful at bringing about a meaningful change, or, worse, being a target of a backlash.

Even though the OMS is not a measure of an anti-system or anti-elite sentiment and a predictor of populist voting, a lot of the frustration and anger likely comes from electoral politics not yielding a desired change. Nor is it a direct measure of extremist radicalisation or a predictor of violence, even though they have many common antecedents and risk factors.

We theorise two pathways along which online mobilisation can lead to populist voting, extremist radicalisation, and violence, which may be mutually reinforcing in some cases.

Frustrated activism – negative feedback loop

  • When activism is not translated into meaningful policy change, activists will seek more extreme ways to bring about that change.
  • When activism is not translated into meaningful policy change, activists will retreat from the public arena, abdicating space to extremists.
  • Regardless of their impact on policy or practice change, an activist engaging publicly can become a target of opponents and experience negative personal consequences which will make them retreat from public arena.

Narrative hijacking by extremist propaganda

  • Mobilisation – the belief in the need for change and the eagerness to be a part of it – can be co-opted by extremist propaganda through narrative hijacking. Extremist propaganda can re-frame the issue, a legitimate concern or anxiety about something. It can be done by diagnosing the problem in a biased fashion, manipulating perception of its prevalence, or support for their solution. It can be done by smearing and de-legitimising alternative explanations, solutions and their supporters, and in extreme cases by legitimising, calling for, or orchestrating violence. For example, the narrative of “great replacement” exploits concern over the consequences of migration by re-framing the issue as a matter of cultural or literal survival. The use of this tactic can become a winning electoral strategy by the radical right, which may in turn trigger vigilante responses by right-wing extremists.
  • The absence of a viable alternative to pressing challenging can reinforce the extremist propaganda narrative. This is a particularly potent possibility in cases when the absence of alternative is not only touted by those who would oppose it, but when there is a true lack of a viable solution backed by concrete policy for addressing complex issues related to migration, environmental issues, economic crises, etc.