Two-mode Belief Networks: The Dual Nature of Chinese People and their Beliefs

• In Event: 41003 - Institutions, Values, and Family Changes in Asia

Abstract

Network analysis of mass belief systems has emerged as a powerful approach for studying public opinions. This methodology transforms correlation matrices derived from subjective survey questions—items measuring attitudes and beliefs—into weighted networks where beliefs denote nodes and correlation values signify tie strength. Such transformations of survey data into belief networks, however, capture only one projection of an inherently bipartite structure—where survey respondents and beliefs constitute two distinct modes connected through responses. Current research faces two critical limitations: the methodological constraint of analyzing only the belief network projection, and the predominant focus on Western democratic contexts. This study addresses both gaps through a comprehensive bipartite network analysis of the 2017 Chinese General Social Survey.

The dual-projection analysis reveals distinct structural patterns. In the belief network, Confucian values—particularly filial piety and traditional gender roles—demonstrate high betweenness centrality and strong inter-node connections, while political beliefs occupy peripheral positions with low centrality measures and weak ties to other beliefs. In the respondent network, we apply community detection algorithms to identify distinct opinion groups. Subsequent multinomial logistic regression reveals systematic associations between group membership and sociodemographic factors such as urban residency, gender, education levels, and age cohorts. Interestingly, central filial piety beliefs show minimal variation across detected communities, while traditional gender role beliefs exhibit significant differentiation.
This study advances belief network methodology by demonstrating how a bipartite, dual-projection analysis captures complementary structural dimensions of beliefs and people, while empirically documenting how local traditional values remain influential under the authoritarian context.

Authors

• Jingkai Huang, Pennsylvania State University

• Diane H. Felmlee, Pennsylvania State University