My paper examines the effects of negative group primes on policy preferences.My primary hypothesis states that negative group primes will have an effect not only on the policy preferences of respondents who receive the primes about a group in which they are not a member (out-group), but they will also be affected by negative primes about their own group (in-group). For members of the more powerful groups, the negative out group primes are very likely to have an effect on policy preferences in ways that widen gaps in public opinion. However, negative in-group primes are unlikely to have an effect on their political attitudes. For members of the less powerful groups, the inverse will be the case as they resist these negative primes by supporting policies linked to their groups with greater furor.To test these hypotheses, I have conducted experiments on three sets of groupmale and female, Caucasian and African American, mainline and fundamentalist Protestant. These groups were selected for two reasons. First, there is compelling evidence in literature of political behavior of gaps in policy preferences. Second, they represent groups with impermeable and permeable membership boundaries.The results suggest that negative out-group primes affect the attitudes of members of the more powerful. However, the effects differ across groups, sometimes in unexpected ways. For whites, the negative African-American primes increased their support for social welfare spending and racial policies. For mainline Protestants, the negative fundamentalist primes decreased their support for social welfare spending and spending for the environment but more supportive of school vouchers and equal roles for women. With all but one exception, these results run counter to my hypotheses. While the negative out-group primes affect whites and mainline Protestants in similar ways, it is unlikely that the cause is the same. In the case of gender, negative female primes decrease support for programs such as social welfare programs per my hypotheses. It is clear from the results that group members interact and conflict in the political realm in very different ways. The results for members of less powerful groups are less clear.