This paper deconstructs social preferences in group contexts and resolves the contradiction between laboratory findings of inequity aversion and the destructive reality of group conflict. The paper replicates previous results that subjects are—on average—inequity averse towards out-group participants and more so towards in-group.  However, the mean is not representative. Using a with-in subject design and new econometric techniques, we find that more than half (54%) of subjects systematically diverge.  Twenty percent destroy total income when facing an out group participant—sacrificing own income to lower out-group incomes. At the other extreme, thirty-four percent of subjects exhibit no bias at all; social preferences are the same for out-group and in-group.  Thus some people are “groupy,” responding readily to group divisions, and a third of subjects are “non-groupy,” with unchanged allocations even when group divisions are salient.  Response times and individual demographics support these findings.