Challenging the conclusions drawn from Zimbardo’s Stanford Prison Experiment

The invitation to design the 'most important experiment that's never been done' is an interesting one, but one that entails a great many dangers. One of these is that by inviting researchers to focus on experiments per se, it encourages them to forget that the fundamental purpose of experiments is to test theoretically derived ideas. Really, then, the question should be 'what is the most important theoretical idea that has never been empirically challenged or supported?'

An answer to this question needs to be informed by an analysis of both (a) ideas that dominate the field but which are misleading, and (b) ideas which offer an alternative, superior understanding. In my own field of social psychology, there are a great many candidates for (a), but relatively few for (b).

In this regard, I would assert that the most important class of ideas that need to be challenged in our understanding of social life are those which lead us to believe that social problems encountered in the world at present (e.g. tyranny, prejudice, abuse, discrimination) are the 'natural' manifestation of inherent processes (e.g. evolutionary, socio-biological, or social psychological). The most important studies are therefore those which take influential studies that appear to support such ideas, and turn them on their heads.

In this regard, one study that I would like to conduct would involve using the paradigm of the Stanford Prison Experiment (SPE, Haney et al., 1973) as the basis for a study showing that, under certain conditions, people (e.g. prisoners) can resist oppression as well as commit and fall victim to it — thereby challenging the idea that tyranny is an inevitable consequence of assigning people to powerful and powerless roles.

In fact we were given the opportunity to design and conduct just such a study several years ago (the BBC Prison Study — BPS; Reicher & Haslam, 2006). The findings to this both (a) challenged received models of tyranny, (b) supported an alternative analysis (derived from social identity theory; Tajfel & Turner, 1979), and (c) bore striking resemblance to patterns of resistance displayed in real-life prisons (and elsewhere in society).

The problem here, though, is that because our findings diverged markedly from those of Zimbardo, he (and others) assumed that the study must have had an inherent design flaw. Of course, as an experiment, the study had certain limitations (e.g. lack of a control condition, small sample size) that it would be good to try to address in follow-up research. Indeed, it would be great to conduct an experiment which resolved this debate conclusively.

Such a study might involve two conditions with multiple prisons in each: in one, the prisons would recreate and replicate patterns of tyranny observed in the SPE; in the other a relevant manipulation would heighten shared social and political identification among the prisoners while weakening that of the guards in order to show that, over time, this was a basis for resistance of the form displayed in the BPS (and in prisons like Robben Island and the Maze; Buntman, 2001; McEvoy, 2001).

Here again, though (as we have found out), there is a danger in thinking that the resolution of such matters is only ever an empirical issue — a question of ethics, resources, and careful design. These things are important, but ideology, politics, group memberships and vanity also have a role to play. You can lead an experimentalist to data, but you can't always make them think. The most important experiments are those which make such disengagement harder, and which encourage fresh minds to change the world not just reproduce it.

About the author

Dr Alex Haslam is Professor of Social and Organisational Psychology at the University of Exeter, UK.