A few weeks ago, the FET Open project I was coordinating, IBSEN, came to a successful conclusion. We had begun our work in September 2015, when teams from the Universities of Aalto, Amsterdam, Cambridge, Oxford, València, Zaragoza, and my own institution, Universidad Carlos III de Madrid, joined forces to study human behaviour in large groups. We wanted to determine whether the results of behavioural and economic experiments involving just a few people also apply to larger groups.
To carry out our research, we developed tools for carrying out large scale experiments, such as experimental software making it possible to coordinate at least 1000 people simultaneously and a controlled volunteer pool that now has around 25 000 subjects. We worked on a range of interesting problems in economics and the social sciences, such as investigating how bubbles form in economic markets by asking volunteers to predict the price of a particular item. The project team in Amsterdam had already discovered that, for groups of six people, the behaviour of the participants could be understood in terms of simple rules. We wanted to see if this conclusion was still valid for larger groups, as smaller groups could perhaps be influenced by one person acting erratically. Therefore, we set out to do the experiment with 100 people, which required working simultaneously from two labs, València and Amsterdam – and found that the results of the six-person groups were reproduced nicely when 100 people were involved.
Another experiment we ran, with 1000 online participants, was the so-called public goods game, where participants contribute money to a common pot that is later shared equally among all participants, irrespective of their contribution. As many studies have shown, when this game is repeated contributions steadily decline, as participants who contribute to the pot realized that they are being taken advantage of by those who do not, and stop contributing themselves. Size matters here too, as after every round players learn what others have done, but with 1000 people in the group you cannot possibly track everybody.
In our version of the experiment, we provided summary information. In one experiment, we just told the players what the average contribution was, whereas in another we told then how many players had contributed between 0 and 20% of their initial endowment, how many between 20% and 40%, and so on. We found that that when people knew the average contribution, their own contributions tended to cluster around that average; on the contrary, when people knew a rough distribution of contributions, they tended to divide in two groups, one contributing all their endowment and one contributing nothing.
Working with large groups clearly makes a difference, and more research still needs to be done to understand how information drives behaviour in large group interactions more generally, such as in large public and private organizations, and online platforms and social media. In fact, society is the perfect example of a large, complex system, and only by understanding how the people who make up society interact can we apply the tools of complexity science to address many of the challenges we are currently facing. The IBSEN project is already working to find insights into real-world questions, such as how to drive a car on a road full of driver-less vehicles, how to make investment decisions as a trade-off between the impact our money makes in the real world and the returns we receive, how a universal basic income could influence our performance, and how we could participate in actions to mitigate climate change and global warming.
I am personally very excited about all these possibilities IBSEN has opened, and I am convinced that they will make an impact on evidence-based policy making. I hope to meet some of you, those who need answers on similar issues, so we can put the IBSEN expertise to work in the near future.
Video on experimental economics and LINEEX
Video showing the conclusions of the climate change experiment
Introduction to IBSEN