Although public opinion in Taiwan is increasingly in favor of gay rights, there is no corresponding trend at the elite level and no concrete policy changes have been achieved. Based on the logic of electoral competition and political socialization, this paper takes the 2014 elections for municipal councilors in Taiwan as an example and examines the factors influencing support for LGBT equality among local politicians. Theoretically, although the centrifugal effect of the SNTV system motivates candidates of the same party to differ from each other on a variety of issues, political socialization, on the contrary, encourages a more unified issue stance taken among party members. The results of multilevel models suggest that there is little linkage between the characteristics of constituents, the electoral rule, and candidates’ stances on gay rights. However, both partisanship and age play an indispensable role. Candidates who are members of the DPP or one of the liberal parties are more likely to support LGBT rights than their KMT counterparts.

One of the problems with empirical studies of Taiwanese/Chinese identity in Taiwan is the use of over-simplified measurements based on responses to a question involving three choices: is your nationality Taiwanese, Chinese, or both? This study attempts to produce a new model with a more fine-grained conceptualization of national identity in Taiwan. The model is derived from Rawi Abdelal et al.’s idea of social identity, and applies Partial Least Squares Structural Equation Modeling (PLS-SEM) to survey data, to develop a social psychological framework using three independent latent variables: “national norms,” “national closeness,” and “national purposes,” and a single dependent latent variable: “state identity.” The results of this re-analysis show all three of the independent variables have significant positive correlations with the dependent variable “state identity.” Of the independent variables, national norms has the highest total effect.
For respondents self-identifying as Taiwanese (T respondents) and respondents self-identifying as Chinese (C respondents), there were significant differences in two dimensions: national purposes and national norms. The strength of T respondents’ national purposes is higher than C respondents while the strength of C respondents’ national norms is higher than T respondents. In addition, a comparison of total effect value and outer weight found that T respondents and respondents self-identifying as both Taiwanese and Chinese (B respondents) also differed. Both T and B respondents stress on “state-building,” a component of the latent variable national purposes. For the dependent variable state identity, however, B and T respondents differ. T respondents take a pro-Taiwan and anti-unification stance. B respondents, however, take a pro-“Republic of China” and prodemocratic unification stance. Variables such as age, education, and social contacts all have moderating effects for both T and B respondents but not great enough to change the path direction.