Perceptions of Gender in Taiwan’s Elections: Results from an Experimental SurveyTaiwan is primed to be the first country in the region in which a female candidate is not connected by blood or marriage to a previous male leader
Taiwan is frequently praised for its gender parity compared to other countries in the region. Using the same criteria as the United Nations’ Gender Inequality Index (GII), Taiwan would rank second only to the Netherlands. The percentage of legislative seats held by women in Taiwan also surpasses virtually all of their neighbors. This can be attributed in part to educational opportunities dating back to the Japanese colonial era, but more so to gender quotas established during Taiwan’s democratization. These quotas not only extend to legislative seats and subnational elected offices, but to candidates within the Democratic Progressive Party (DPP) and Chinese Nationalist Party (KMT). Furthermore, while female presidents and prime ministers are not rare in Asia, Taiwan appears primed to be the first in the region in which the female candidate is not connected by blood or marriage to a previous male leader.
Such trends do not necessarily overcome cultural factors that promote a patriarchal society with gender-based division of labor. For example, quotas in public sector employment and women in senior management roles hide that less than half of adult women are employed full-time. Confucian elements are also often believed to undermine the formation of a liberal democracy; however, Taiwan nevertheless has evolved into a stable democracy with greater gender equity than many in its region. In contrast, the Politburo Standing Committee of the Chinese Communist Party has yet to have a female representative at all.
The current electoral context, where until last month both of the major parties were running female presidential candidates, provides a unique opportunity to measure public perceptions of gender in politics. As others have stated, this election could show that Taiwan is diverging from traditional patterns of gender inequality seen in Chinese culture. We designed a web survey through PollcracyLab to tap into whether the public places different requirements on male versus female candidates. Within in a larger political opinion survey, we asked people to evaluate three statements on a five-point scale from strongly disagree (1) to strongly agree (5). However, we randomly assigned whether a person received a question about a generic male or female candidate. Thus we could identify whether the public showed differing standards for candidates based solely on their gender.
The questions were:
1. A man without a college degree is unfit to be president.
A woman without a college degree is unfit to be president.
2. An unmarried man is unsuitable for political office.
An unmarried woman is unsuitable for political office.
3. A man with a criminal record is unsuitable for political office.
A woman with a criminal record is unsuitable for political office.
Previous research suggests that the gender of candidates matters, whether in terms of political office or other forms of employment (see examples here, here, and here). For example, a study of Norwegian voters found that evaluations of candidate abilities varied considerably when identical speeches were presented as from a male versus female candidate, with similar distinctions on trait attributions found in experimental designs elsewhere (examples here and here). Based on existing research we expected that the public would place higher demands on female candidates, thus agreeing with these questions at a higher rate than when presented as a male candidate.
However, what we found ran counter to expectations (see Figure 1). For each question, respondents who received the female version (shown in red) were more likely to disagree than when presented a male version (shown in blue). Regarding needing a college degree to be suitable for the presidency, the average score for those receiving the male version was 2.76, compared to 2.39 for the female version. For the marriage question, the male version averaged 1.94 compared to 1.64 for the female version. Finally, the male version of the criminal record question averaged 3.67, compared to 3.23 for the female version. While the range in each was rather small, the distribution of responses based on the gender version was statistically significant in each. In other words, the public was more likely to disagree with the statement consistently when the candidate was presented as a woman.
Figure 1: Summary Responses to Three Experimental Questions
1 (Strongly Disagree) to 5 (Strong Agree)
Further analysis shows that the findings are largely motivated by the gender of the respondent. In other words, female respondents were far more likely to disagree with the above questions when presented as a female candidate than when presented as a male one. Put another way, the variance between versions when presented to male respondents was consistently smaller than when presented to females. The figure below (Figure 2) presents this distinction, with male respondents in blue and red and female respondents in green and purple. Among males, only the criminal record question elicits a statistically significant difference between the hypothetical male and female candidate. Meanwhile, among female respondents all three questions elicit a statistically significant difference based on whether the candidate is presented as male or female.
Figure 2: Summary Responses to Three Experimental Questions
1 (Strongly Disagree) to 5 (Strong Agree)
Additional statistical analyses that controlled for partisan identification, age, political identity (e.g. Taiwanese, Chinese, both) still found a statistically significant decline in support of the three questions when presented as a female candidate rather than a male. Furthermore, borrowing from the AsianBarometer survey, which includes a battery of questions aimed at measuring traditional values, we created a traditional values index based on nine questions. As expected, traditional values correlated with agreeing with each of the three questions, but again a significant distinction was apparent based on whether the candidate was presented as male versus female.
What does this tell us? First, the findings suggest that female candidates, at least in the abstract, are not directly discounted. Even male respondents did not expect additional requirements out of female candidates compared to their male counterpart. However, it is unclear why a stronger distinction was evident among female respondents. This may be tapping into feelings of historical discrimination, for example. Nor does our survey attempt to assess policy preferences that may correlate with the gender of the candidate. For example, evidence from Taiwan suggests a higher rate of government spending for social welfare under female mayors. Similarly, the public may rely on stereotypes of the expectations of females, although it remains unclear when such stereotypes are employed or even actively suppressed at the individual level (examples here and here). Regardless, the findings are generally consistent with perceptions of gender parity in Taiwan in spite of cultural factors that may discourage such perceptions.
Timothy S. Rich is an assistant professor in political science at Western Kentucky University. His main research focuses on the impact of electoral reforms in Taiwan, South Korea, and Japan compared to similar legislative systems. His broader research interests include electoral politics, domestic and international politics of East Asia, and qualitative and quantitative methods.
Hannah Neeper is a junior at Western Kentucky University majoring in English and Chinese with a minor in Political Science. She is currently researching gender perceptions in Taiwan through an internal research grant (FUSE).