Assessing the Structure of Policy Preferences: A Hard Test of the Low Dimensionality Hypothesis

This paper analyzes belief systems in a novel way, modeling relational patterns of policy disagreements using nonmetric multidimensional scaling. Because of its flexible assumptions, the approach enables us to conduct a notably harder test of the "low dimensionality" hypothesis than is found in previous work. The results support the proposition that a basic space (consisting of a small number of interwoven issue domains) anchors the policy dimension of public opinion. Among our findings, we show that voters---especially those meeting a minimum threshold of political sophistication---neither lack meaningful attitudes nor hold distinct preferences across a wide range of issues. Rather, their policy attitudes are organized alongside relevant core values and affective evaluations in a common, low-dimensional cognitive space. A unidimensional approximation of these belief structures often exhausts the explanatory power of vote choice models.