Myers, D. G., & Kaplan, M. F. (1976) Group Polarization in Simulated Juries. Personality and Social Psychology Bulletin
by Elnaz Alikarami *
Suppose a group of friends give a slightly higher chance to Donald Trump over Bernie Sanders for 2020 election. Having a “Trump vs Sanders” debate in the group will probably make them more certain about Trump. This is an example of group polarization. It is called “risky shift” and was a big topic in social psychology in 70s before echo chambers became fashionable through social media.
Myers and Kaplan (1976) asked if this phenomenon generalizes to jury decisions when judging guilt and innocence.
Three cool things about this paper: (1) the simulated jury discussions are very similar to real-life group-decision making; (2) predictions do not rely on pay-off; (3) the experiment design (see Fig 1) is simple yet carefully controlled and focused on one target hypothesis.
In a single session, 60 subjects all gathered in a big room. Each person read a booklet of 8 stories of traffic felonies (4 low-guilt and 4 high-guilt) and rated how much punishment each case deserved. Then, subjects were randomly organized in 10 groups of 6 people to discuss half of the cases (no judgment was recorded). Then groups were reshuffled and a second round of discussion followed. At the end, subjects went through all of the cases once again individually and judged each.
The key result is depicted below. Judgments were more “polarized” after discussion. Group discussion (1) led people to increase punishment for the high-guilt cases and (2) decrease punishment for the low-guilt cases.
A very interesting feature of this result is that they are the opposite of what would be expected from regression to the mean. Talking to others changes our opinion in very interesting and non-trivial ways.
(*) E Alikarami is a qualified dentist from Iran currently doing an MSc in Canadian Center for Behavioral Neuroscience, Lethbridge, Alberta.