Variability or Moderation? The Effects of Ambivalence on Political Opinions

Prior research theorizes that ambivalence makes opinions about an object unreliable in the sense of being haphazard, unpredictable, or variable. As such, ambivalence is a prominent explanation for seeming nonattitudes in opinion surveys. This study proposes an alternative account of the effects of ambivalence on attitudes. It posits that people who are ambivalent about an issue split the difference between their conflicting considerations by taking a position near the middle of the bipolar opinion scale, which reflects a moderate attitude. I show how the widely-used method of modeling the supposed variability of ambivalent opinions conflates variability and moderation. This problem is addressed by modeling variability and moderation of attitudes separately, without this confound. Using this strategy in analyses involving four datasets and three policy domains, the results show that ambivalence is associated with moderate, not variable, attitudes. Ambivalence does not increase the variability of opinions but, rather, moves them quite predictably toward the middle of the response scale. The results recast our understanding of the effects of ambivalence on political opinions and raise questions about the ability of ambivalence to explain nonattitudes in surveys.