Unpredictable Distinction in Cultural Inequality: Social Class vs Social Networks

• In Event: 20420 - Section on Sociology of Culture Roundtables
  In Refereed Roundtable (90 minute): Table 3: Cultural Capital and Taste

Abstract

Implementing machine learning analysis, this study compares the influence of social class and social networks in shaping cultural inequality, offering a data-driven perspective on the structural and relational dynamics of cultural capital. While cultural tastes are largely shaped by social class, they cannot be fully explained by class alone. This study seeks to empirically examine how cultural preferences that fall outside traditional class-based explanations are shaped and differentiated through social networks, shedding light on the relational dynamics of cultural consumption. Using the National Leisure Activity Survey in South Korea (N = 10,040, collected in 2023), we applied UMAP and k-means clustering to classify individuals into two groups: omnivores, who engage in diverse leisure activities, and univores, who participate in a more limited range of activities. A CatBoost model identified region, social embeddedness, and age as key predictors of leisure participation. Based on survey responses regarding 87 cultural consumption behaviors, this study applies machine learning techniques to classify individuals and predict patterns of cultural engagement using a range of socioeconomic variables. The CatBoost model achieved an 75.86% accuracy rate in predicting cultural consumption patterns based on these factors. Additionally, further analysis was conducted to examine why the algorithm failed to predict the cultural engagement of the remaining 24.14% of individuals, offering insights into the limitations of socioeconomic determinants in explaining cultural preferences. Misclassification analysis revealed that omnivores misclassified as univores tend to be older, have lower incomes and poorer health, and engage in leisure primarily for health-related and family-oriented reasons. Conversely, univores misclassified as omnivores generally have higher incomes and better health, engaging in leisure activities that prioritize self-satisfaction, stress relief, and social interaction. Moreover, misclassified omnivores exhibited a tendency toward solitary leisure, whereas misclassified univores participated more in social leisure. Finally, network analysis showed that misclassified omnivores had more interconnected leisure structures, while misclassified univores formed highly concentrated networks, primarily engaging in sedentary leisure while interacting within close social circles. While social stratification remains an important factor, this study highlights the role of social embeddedness in cultural consumption.

Authors

• Yeojin Yang, Korea University

• Eun Kyong Shin, Korea University