Political Belief Networks. Socio-cognitive Heterogeneity in American Public Opinion

Most research in public opinion and political sophistication relies on the assumption that Americans organize their political belief system according to the liberal and conservative categories. Yet not all of them do. We hypothesize that citizens’ sociodemographic profiles make them disposed to espouse different understandings of the political debate, and document systematic heterogeneity in Americans’ organization of their political attitudes over the last two decades. We interpret this diversity as the coexistence of multiple belief systems. Relational class analysis (RCA), a network-based method for detecting heterogene- ity in collective patterns of opinion, is used to identify distinctive opinion structures – or belief networks – that are shared within different groups of respondents. The analysis of ANES data between 1984 and 2004 leads to the identification of three stable groups of American citizens: of the ANES data over the 1984-2004 period leads to the identification of three stable groups of respondents: Ideologues, whose political attitudes strongly align with either liberal or conservative categories; Alternatives, who are in- stead morally conservative but economically liberal, or vice versa; and Agnostics, who exhibit weak associations between political beliefs. Respondents’ sociodemographic profiles, particularly their income and religiosity, stand at the core of the different ways in which they understand politics. When their social identities and related political interests are incompatible with the prescriptive liberal-conservative polarity (i.e., high earners with weak religious commitments), indi- viduals gravitate toward alternative ways of conceptualizing the political debate. These results raise important methodological questions concerning the limitations of traditional analytical techniques that assume population homogeneity in the organization of political beliefs