Social isolation by design: Bias in measuring core networks in Taiwan?

The estimation and measurement of the size of egocentric networks have sparked vigorous discussion and debate. Drawing on datasets from the Taiwan Social Change Survey, this study explores methodological issues pertaining to the change of core networks in Taiwan from 1997 to 2017 via a modified Poisson mixture approach, assesses the efficiency of name generators as a survey instrument via Fisher Information Maximizer, and investigates the role of social desirability in reporting core networks. Net of other effects, the study finds that individuals expressing a strong sense of social desirability report significantly fewer close contacts and face a higher risk of social isolation. Name generators in this study are associated with trivial design errors and can yield estimates comparable to those produced by exact enumeration. These findings are situated in the drastic changes in face-to-face survey interviews as well as the cultural context of Taiwan and, more broadly, East Asia. They call for further research inquiries into methodological issues regarding measuring and estimating egocentric networks in a transnational and modern setting. • This study shows changes in Taiwan’s core discussion networks from 1997 to 2017. • We use modified Poisson mixture regression and Fisher Information Maximizer. • High social desirability linked with fewer contacts and higher isolation risk. • Name generators show minimal design errors, comparable to exact enumeration. • It highlights cultural context and calls for transnational methodological research.