Existing research has shown that anticipating discussing socially divisive topics is associated with feeling threatened, both in terms of self reported emotion and automatic reactions (the dissensus effect). The current research aims to test two further hypotheses about this effect. First, given the proposed causal mechanism (that divisive topics are socially challenging, and hence discussing them conflicts with social goals), the effect should be stronger for participants who are more motivated to have positive interactions. Second, given that threat is an avoidant response, it is predicted that dissensus will lead people to think less about the topic in question when social considerations are salient. These hypotheses were tested across two studies. Both manipulated participants' social goals (manipulated using subliminal priming in Study 1, and by a scrambled sentence task in Study 2) and then asked them to imagine discussing consensual or divisive topics (manipulated within-subjects using actual issues in Study 1, and between-subjects using fictional polling data in Study 2). Study 1 measured participants' affective responses to discussing the issues, whereas Study 2 measured the amount of thought about the issue. Both studies also measured participants' attachment styles (or internal working model of relationships). The first hypothesis received mixed support; across both studies, participants with more insecure attachment styles showed a stronger dissensus effect, but activating social goals did not have a comparable effect. The second hypothesis received very little support; while Study 2 found evidence that thinking about discussing a divisive (as opposed to consensual) issue leads participants to adopt less extreme attitudes and feel more ambivalence, there was no evidence for an effect on the amount of thought (or bias in this thought).