Affective Intelligence Theory (AIT) asserts that anxiety reduces the effect of party identification on candidate preferences ( Marcus, Neuman, and MacKuen 2000 ), but recent studies have raised doubts about this causal claim. Rather than functioning as a moderator of party identification, perhaps anxiety has a direct effect on preferences, or perhaps the relationship is reversed and preferences drive emotions ( Ladd and Lenz 2008 ). Alternatively, Marcus et al.’s measure of anxiety may simply be capturing partisan ambivalence, so the posited relationship is spurious ( Lavine, Johnston, and Steenbergen 2012 ). This paper addresses each of these questions by examining the effect of experimentally induced emotions on the types of considerations that came to mind when a national sample of adult Americans was asked what they liked and disliked about Barack Obama. By directly manipulating anxiety, this experiment avoids the causal ambiguity plaguing this debate and ascertains the true nature of the relationship between anxiety and ambivalence. Consistent with AIT, anxiety led respondents to recall more contemporary considerations, whereas enthusiasm brought to mind more long-standing considerations. Because the political context at the time of the study (fall 2013 ) was a very tumultuous time for the Obama administration, the increased accessibility of contemporary considerations led Democratic participants to experience more ambivalence in the anxiety condition. This effect was concentrated among those Democrats who were exposed to the most newspaper coverage.