Experiments to Investigate Online Collective Action and Leadership

Project Date: 
Jan 2009 - Dec 2010
Emergent Leadership Screenshot

These experiments took place as part of the Rediscovering the Civic project run by the Institute of Political and Economic Governance at the University of Manchester and the University of Southampton. This ESRC Ventures project involved a range of experiments investigating how we can increase civic behaviour and the most effective ways to encourage active citizenship. The project was funded mainly by the ESRC with co-funding from Communities and Local Government and the North West Improvement and Efficiency Network.

The research team leading these experiments was Helen Margetts and Peter John, Tobias Escher, Stephane Reissfelder, and Scott Hale.

Purpose: The purpose of these experiments were to explore the dynamics of collective action on-line and to test empirically how different information environments affect collective action decisions at various stages in a mobilization. Specifically, we wanted to examine the effect of different forms of real-time ‘social influence’ on people’s participatory decisions.

First Experiment: The Effect of Social Information on Online Collective Action, 2008-9

This quasi-field experiment involved a randomised controlled trial (RCT) testing whether providing information on the number of signatures of an online petition affects participation. It was expected that where very low numbers of people have signed a petition, the information will have a negative impact on an individual’s likelihood of signing, as they may consider it a hopeless cause. Where very high numbers of people have signed, the information may have a positive impact, generating excitement, social pressure and a feeling that they can be part of change.

Subjects participated remotely in the experiment via a custom-built interface. The experiment involved 668 people from OxLab’s subject pool, who were presented with a screen which asked them to examine a number of issues and then asked to (a) express their willingness to sign a petition supporting the issue and (b) donate a small amount of their participation fee to supporting the issue. Subjects were randomly allocated across a control group and three treatment groups with information of low, medium and high numbers of other signers.

The experiment provided evidence that where potential participants receive the message that one million other people have signed a petition, they are significantly more likely to sign themselves. Below this figure, the information about other participants does not positively affect their likelihood of signing and can even have a negative effect. These findings support the claims of those sociologists who argue that a ‘critical mass’ in political participation makes large groups viable (Marwell and Oliver, 1988) and add to a line of work which re-evaluates Olson’s arguments about the non-viability of large groups to form.


Second Experiment: Leaders, Followers and Tipping Points in Online Collective Action, 2010

In this laboratory-based public goods experiment, individual subjects were asked to contribute to the collective good at a cost, receiving a higher return if the number of participants is higher than a determined point. We wanted to test the hypothesis that different personality types would be differentially susceptible to different information environments, with some consistently acting as ‘leaders’ or ‘starters’ and some more likely to be ‘followers’. ‘Starters’ will be more inclined than ‘followers’ to sign up to mobilizations where they have no idea about other numbers of participants, while ‘followers’ will be more influenced by feedback information and being made ‘visibile’ to other participants. There will be no generalised effect of personality on people’s propensity to sign: rather, we expect an interaction between personality and the information environment. The research addresses previous work on leadership and tipping points in collective action (Marwell and Oliver, 1993; Schelling, 2005).

To test this hypothesis, we implemented two treatments. First, in the ‘social information’ treatment, subjects were provided with real-time information on how many other subjects in their randomly allocated group had contributed. Second, in a ‘visibility’ treatment, subjects’ contributions were made visible to other participants in the room. The control group received no social information and their contributions were not made visible. All subjects took part in both treatments and the control at different points during the experiment. Post-experiment personality tests enabled us to examine the personality characteristics of those subjects who habitually ‘lead’ rather than ‘follow’.

We recruited 185 subjects to the OxLab laboratory. After the experiment, subjects completed a questionnaire, which included a short personality survey to determine subjects’ locus of control using the Rotter scale. At each round (n=28), subjects were shown a step-level public good scenario phrased as a request to fund a local initiative. Subjects were endowed with 10 tokens and informed about the provision point (60 tokens) and the number of participants in their group (N = 10). If the provision point was met, a fixed bonus was redistributed amongst all participants. Subjects were paid for one round only, which is selected at random, a design which means that the most rational behaviour is to treat each as if it were the only round (Bardsley 2000). Groups were randomly allocated at each round, so that players never interacted with the same group.



Post new comment

The content of this field is kept private and will not be shown publicly.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd>
  • Lines and paragraphs break automatically.

More information about formatting options

By submitting this form, you accept the Mollom privacy policy.