Improving Cultural Cartography with Correlational Class Analysis: Which Shared Cultural Schemas Structure American Musical Tastes?

• In Event: 365 - Section on Sociology of Culture Roundtable Session
  In Refereed Roundtable: Table 09. Music and Taste

Abstract

The task of revealing intelligible cultural schemas within complex collections of cultural data is among the most central ones posed by sociologists of culture. In an influential recent paper, Goldberg (2011) proposes an innovative technique for identifying such schemas in survey data, which he terms Relational Class Analysis. This approach rests on a novel measure called "relationality," which Goldberg contends can quantify the extent to which two respondents organize their attitudes according to a shared cultural schema. However, relationality is a flawed answer to a measurement problem that has long had a superior solution. In this paper, I show that Goldberg's theoretical reasoning implies that relationality must measure the degree of linear dependency between vectors—the quantity captured by Pearson's correlation. I then demonstrate with Goldberg's own example that correlation produces a far more accurate result than relationality. A further 10000 simulations confirm that this switch from "Relational" to "Correlational" Class Analysis (CCA) reliably increases accuracy, so much so that, when the methods disagree, the odds that CCA's results are more accurate exceed 16:1. Finally, when I revisit Goldberg's analysis of the 1993 GSS music taste data with CCA, my results contradict Goldberg's controversial assertion that a third of respondents arrange their tastes according to the traditional distinction between “highbrow” and “lowbrow” genres. Instead, CCA documents two classes of respondents whose cultural schemas echo the exclusionary omnivorousness described by Bryson, in which a handful of musical tastes closely associated with low-education listeners are distinguished from tastes for all other genres.

Author

• Andrei G. Boutyline, University of California-Berkeley