1 ! Group Bias , Identity , and Social Preferences

This paper reveals that a subset of subjects adopting extreme behavior underlies intergroup bias. In a simple task, we replicate previous results that, on average, subjects are less inequity averse allocating income to out-group members. Using a with-in subject design and new econometric techniques, however, we find a group division does not matter for most subjects; social preferences are the same for out-group and in-group members. But twenty percent of subjects adopt particularly pernicious behavior—sacrificing own income to destroy out-group incomes. Thus we uncover a new sort of heterogeneity. Some people are “groupy,” responding readily to a group division, while others do not. This new picture of bias follows from the first analysis of individuals across more or less salient group treatments: (a) a non-group control, (b) a minimal group treatment dividing subjects by arbitrary criteria, and (c) a political group treatment dividing subjects by opinions and party affiliations. We compare Democrats to Democraticleaning Independents, subsets large enough for statistical power, and with identical politics but for political party affiliation. Democrats show bias in both minimal and political group treatments; Independents only exhibit bias in the political treatment. Thus, behavior mirrors both individual propensities to join groups and individual identities. * Contacts: [email protected], [email protected], [email protected], [email protected]. Authors are listed in alphabetical order following the convention in economics, except for the lab director who is listed last following the convention in psychology & neuroscience. We are grateful to George Akerlof, Jeff Butler, John Miller, Pedro Rey-Biel and seminar and conference participants at Berkeley, Canadian Institute for Advanced Research, Conference on the Economics of Interactions & Culture (EIEF), Duke, Ecole Polytechnique, Erasmus Choice Symposium, Pompeu Fabra, Institute for Economic Analysis & Universitat Autònama de Barcelona, Maryland, Paris School of Economics, THEEM, and Washington University for their comments. We thank Catherine Moon, Robert Richards, Sierra Smucker for research assistance. We thank the Social Science Research Institute at Duke for sponsoring our faculty fellows program in 2010-2011, “From Brain to Society (and Back),” and we are grateful for funding from the Transdisciplinary Prevention Research Center (TPRC) at Duke, as supported by the National Institute on Drug Abuse grant DA023026.!