Attitude Instability Due to Ambivalence: A Reconsideration of Recent Evidence

One of the most enduring debates in political science concerns attitude instability -- do individuals hold stable, well-defined attitudes on political matters, or do they hold unstable attitudes, which vary randomly or in response to influences that change from moment to moment? Recent work in this area suggests that political attitudes are unstable when individuals are ambivalent, or hold conflicting considerations about some political matter. Much of the empirical evidence of attitude instability due to ambivalence comes from what are often termed ``inferential tests,'' which demonstrate that measures of ambivalence are related to less predictable political attitudes. These tests suggest that attitude instability due to ambivalence is common, existing not only for narrow political issues, but also for attitudes towards political candidates, parties, and political institutions -- in other words, these tests suggest that the public is conflicted and confused about even the most basic political matters. However, I demonstrate in this paper that these inferential tests of attitude instability due to ambivalence are inconclusive, failing to rule out the possibility that all individuals, even those who appear to be ambivalent, actually hold stable attitudes on political matters. I first show that the measures of ambivalence used in past work may actually indicate neutrality, or indifference between two options on some political matter. I then show that under these inferential tests, neutrality and ambivalence are empirically indistinguishable. Thus, much of the empirical support for attitude instability due to ambivalence is open for reinterpretation.