ABSTRACT

Presence of social network among the respondents in a survey may have an influence on the patterns of response. Latent class analysis identifies hidden subgroups in the respondents in a survey and simplifies the structure of heterogeneity amongst them. In the present study, a new model of social network-informed multiple group latent class analysis is developed by introducing a parameter measuring the influence of a social network on a respondent. For suitable application of the proposed model, data from 75 Indian villages with detailed demographic characteristics and a diverse social and economic network are used. The present study examined the methodological aspect of the proposed model of multiple group latent class analysis adjusted for the impact of social network.

摘要

调查中受访者之间存在的社交网络可能会对响应模式产生影响。潜在类别分析在调查中识别受访者中隐藏的子组，并简化他们之间的异质性结构。在本研究中，通过引入衡量社交网络对受访者影响的参数，开发了一种基于社交网络的多组潜在类别分析的新模型。为了适当应用所提出的模型，使用了来自 75 个印度村庄的数据，这些村庄具有详细的人口特征和多样化的社会和经济网络。本研究检查了针对社交网络的影响进行调整的多组潜在类别分析所提出模型的方法学方面。