Myth of Independent Voters and the Prediction of Closet Partisans’ Party Identification in Taiwan





Published date: 

December, 2016


Frank C. S. Liu
Ai-jhen Tsai


Researchers of partisan voters have been assuming that there is a solid difference between “independent” voters and partisan voters (including leaners). This is hardly the case in the Taiwan context, a democracy with a two-party presidential system, where over 40 percent of voters are partisans, but claim to be independent in most telephone surveys. Pollsters, researchers, and journalists have been calculating the distribution of party supporters by either omitting these “independent” voters due to the unavailability of the data, or simply applying counterintuitive formulae to guess the distribution of the respondents with missing data. This study  avoid the definition of these not-so-well-defined “independent” voters, but takes aim at these “invisible” or “closet” voters and attempts to  the partisan orientation behind their ambivalent answers to telephone surveys.

With this in mind, we took a series of steps, including qualitative and quantitative ones. First, we used a representative sample, conducted in January 2014 (N=1,072) in Taiwan via an RDD telephone survey. This survey included the conventional party identification question plus a series of theory-based alternative questions that we evaluated as triggering respondents’ mobilized reasoning regarding the two major political parties, the Kuomintang (KMT) and Democratic Progressive Party (DPP). We then created an index for partisan respondents of the two political camps, and applied the score patterns to the closet respondents. In another follow up survey (March 2014) that targeted the closet respondents, we found that the correctness of prediction using the index was about 70%. We then targeted and interviewed the most ambivalent closet voters and explored how their partisan mobilized reasoning was (and failed to be) triggered by the alternative survey questions. We concluded with a few survey questions that future electoral studies can use for probing closet voters. The rich implications of the findings for improving the accuracy of predicting partisan votes, the debates about the characteristics of independent voters, and the development of partisan motivated reasoning theories are discussed.