Affect control theory (Heise 1979, 2007, 2010; MacKinnon 1994; Smith-Lovin and Heise 1988) contends individuals' sentiments for social identities result from three sources: (1) fundamental cultural norms, (2) individual experiences, and (3) error. Importantly, affect control theory adopts a consensus assumptions that individuals are homogenous with regard to sentiments due to the powerful and equivalent influence of cultural norms across individuals. Yet, a considerable body of evidence about the role of interpersonal, intrapersonal, social network, and system-organizing processes leads to the expectation of sentiments that are localized within areas of sociodemographic space. As such, existing empirical evidence suggests that a modification of the consensus assumption may be necessary. Research in the affect control tradition relies on sentiment data collected from sociodemographically homogenous individuals. Therefore, queries regarding the consensus of sentiments using existing sentiment data are very limited due to sample homogeneity on numerous sociodemographic dimensions such as race, age, education. The current investigation examines the validity of the consensus assumption embedded in affect control theory.

The current investigation employs two data sources to explore the validity of the consensus assumption: (1) sentiment, sociodemographic, and social network data from sociodemographically diverse respondents in Durham, NC and (2) sentiment data from a homogeneous sample of Indiana University students (Frances and Heise 2006), currently used as database of cultural meanings for affect control theory analyses (Heise 1997). The dissertation explores three fundamental research questions using three different, but related measures of consensus: (1) does the sociodemographic distance between individuals influence the level of interpersonal agreement, (2) is cultural inculcation (affective socialization) equivalent among sociodemographically diverse respondents, and (3) are the affective ratings of sociodemographically diverse respondents equally reliable?

The results suggest (1) consensus for the affective meanings of social identities is dimension dependent, with the highest degree of consensus noted for the evaluation of social identities; (2) the types of measures used to estimate consensus in meaning substantially influence the conclusions regarding consensus, with, the results for the individual to cultural measures indicate a significant degree of consensus, but the results for the person-to-person measures indicate that consensus decreases with sociodemographic dissimilarity; (3) sociodemographic distance is negatively associated with interpersonal agreement, especially for achieved statuses; and (4) neither sociodemographic nor social network factors are associated with significantly more normative or reliable sentiments for social identities, though there is a consistent, but non-significant, positive relationship between education and sentiments that are normative and reliable. Thus, the results provide some support for a consensus model in sentiments for social identities such that affective sentiments are widely shared within a sociodemographically diverse sample of respondents. However, the results also suggest that sociodemographic distance is negatively related to interpersonal agreement.