¹ Michael Forsythe, Yu-Huay Sun, and Andrea Wong, ‘Ma Wins Second Term as Taiwan Voters Back His Push for Closer China Ties’, Bloomberg, 15 January 2012, italics added (accessed 15 January 2012).

² ‘Ma Must Forge PRC Peace Accord’, The China Post, 16 January 2012 (accessed 22 January 2012).

³ ‘Ma Re-elected Taiwan Leader, Mainland Reiterates 1992 Consensus’, Xinhua, 14 January 2012 (accessed 22 January 2012).

⁴ ‘Preamble’, Constitution of the People's Republic of China (accessed 4 January 2012).

⁵ Gilley, Bruce, ‘Not So Dire Straits’, Foreign Affairs, 89, 1 (January/February 2010): 58Google Scholar.

⁶ Glaser, Charles, ‘Will China's Rise Lead to War? Why Realism Does Not Mean Pessimism’, Foreign Affairs, 90, 2 (March/April 2011).Google Scholar

⁷ On Taiwanese politics, see Rigger, Shelley, Why Taiwan Matters: Small Island, Global Powerhouse (Rowman & Littlefield, 2011)Google Scholar. On Cross–Strait relations, see Bush, Richard, Untying the Knot: Making Peace in the Taiwan Strait (Brookings, 2005)Google Scholar.

⁸ For instance, response categories included ‘status quo forever’ and ‘status quo now, decision in the future’. For an overview of the extant Taiwanese survey literature on cross–Strait relations, see Sobel, Richard, Haynes, William-Arthur, and Zheng, Yu, ‘The Polls − Trends: Taiwan Public Opinion Trends, 1992–2008: Exploring Attitudes On Cross–Strait Issues’, Public Opinion Quarterly, 74, 4 (2010): 782–813.CrossRefGoogle Scholar

⁹ The general population figures for ethnicity came from the Taiwan Social Change Survey (TSCS), as the MOI does not measure ethnicity.

¹⁰ For more on sample matching, see Douglas Rivers, Sample Matching: Representative Sampling from Internet Panels, White Paper, Palo Alto, CA, 2011. Available online @ http://www.rochester.edu/College/faculty/mperess/srm2012/Polimetrix_Methodology.pdf. Eric Yu at National Chengchi University has been developing a similar methodology to conduct Internet surveys focused on elections and voting behavior.

¹¹ F(1,526) = 4.006, p = 0.046, η_p ² = 0.01. Data available from TEDS at www.tedsnet.org.

¹² We did not use the 0° to 100° scale in the first place because of ambiguities surrounding the meaning of different degrees in Fahrenheit and Celsius, such as whether 50° would be understood as hot or cold.

¹³ Sample size was N = 2,506. See Gries, Peter, ‘Disillusionment and Dismay: How Chinese Netizens Think and Feel about the Two Koreas’, Journal of East Asian Studies, 12 (2012), 31–56CrossRefGoogle Scholar.

¹⁴ Items inspired in part by the ‘colonial debt’ factor of the CMS. See David, E.J.R. and Okazaki, S., ‘The Colonial Mentality Scale (CMS) for Filipino Americans: Scale construction and psychological implications’, Journal of Counseling Psychology, 53 (2006): 245CrossRefGoogle Scholar.

¹⁵ A two-step hierarchical regression was run predicting warmth towards Japan. The standard demographics were entered in the first step, and age (β = −0.11, p = 0.09) and income (β = 0.08, p = 0.09) were both marginally significant predictors. But when liking Japanese popular culture was entered in at the second stage, age and income fell to statistical insignificance, p = 0.40 and 0.18 respectively. Liking Japanese popular culture by itself contributed an R² change of 0.18 to explaining warmth towards Japan.

¹⁶ K-pop (M = 3.6) was much less liked than US and Japanese popular culture (M = 4.6), t(497) = −11.84, p <0.001.

¹⁷ In a three-step hierarchical linear regression predicting warmth towards South Korea, entering the standard demographics in the first step explained 8% of the variance in feelings towards South Korea, with women (β = 0.20) feeling substantially warmer than men, and younger people and the less educated feeling marginally warmer. Partisanship added a small amount (β = 0.09) of predictive power in the second step, with green feeling warmer than blue. But when our prejudice item (β = −0.35) and liking Korean popular culture (β = 0.36) were added into the regression, the effects of gender, age, education, and partisanship dropped out, and the overall explanatory power increased to a remarkable R ² = .38.

¹⁸ Our survey also included an item asking respondents how warmly they felt towards US President Barak Obama. On average, Taiwanese felt as warmly towards Obama (65°) as they did towards the Dali Lama (64°). Interestingly, while there were no partisan differences in feelings towards Obama, there was a statistically significant difference in feelings towards the Dali Lama, even after controlling for belief in Buddhism, with Blue identifiers (68°) cooler than Green (73°). Obama: F(1,355) = 0.05, p = 0.83; Dali Lama: F(1,355) = 4.69, p = 0.03, η_p ² = 0.013. Both ANCOVAs (analysis of covariance) controlled for standard demographics. The second ANCOVA also controlled for belief in Buddhism. The Dali Lama is likely seen as both a religious and political figure in Taiwan, perhaps accounting for this partisan difference.

¹⁹ This may be because Filipinos come to Taiwan more for work than for marriage.

²⁰ ‘Foes’ is not intended to suggest that Taiwanese see these countries as enemies, only to contrast them against those countries in the other factor more clearly viewed as ‘friends’.

²¹ South Korea is not ‘backwards’, but is not a ‘friend’ either.

²² Countries for which there were no statistically significant partisan differences are not displayed.

²³ F(1,341) = 4.78, p = 0.029, η_p ² = 0.014 in an ANCOVA covarying the standard demographics.

²⁴ F(1,341) = 121.54, p < 0.001, η_p ² = 0.035.

²⁵ Entering pride in Taiwanese democracy as a covariate into the ANCOVA reduced the effect of PID to statistical insignificance, F(1,340) = 0.21, p = 0.654.

²⁶ F(1,357) = 6.47, p = 0.011, η_p ² = 0.018.

²⁷ The indirect path was statistically significant. See the appendix for details.

²⁸ The effect size of the difference was relatively small, F(1,341) = 7.34, p = 0.007, η_p ² = 0.021.

²⁹ F(1,357) = 8.744, p = 0.003, η_p ² = 0.024, controlling for standard demographics.

³⁰ F(1,341) = 6.34, p = 0 .012, η_p ² = 0.018; F(1,341) = 15.51, p < 0.001, η_p ² = 0.043; F(1,341) = 15.95, p < 0.001, η_p ² = 0.045, for South Korea, the United States, and Japan respectively, all controlling for standard demographics.

³¹ F(1,341) = 33.39, p = 0.029, η_p ² = 0.089, controlling for standard demographics.

³² Interaction ΔR² = 0.01. F(1,272) = 3.14, p = 0.078.

³³ Chris Wang, ‘DPP presidential candidate Tsai Ing-wen's election platform’, Taipei Times, 9 January 2012 (accessed 13 January 2012).

³⁴ Wilks Lambda = 0.87, F(2,354) = 27.63, p < 0.001, η_p ² = 0.14 in a MANCOVA controlling for age, gender, region, education, and income.

³⁵ F(1,338) = 2.02, p = 0.16 in an ANCOVA controlling for age, gender, region, education, and income.

³⁶ F(1,357) = 76.04, p < 0.001, η_p ² = 0.18 in an ANCOVA controlling for age, gender, education, income, and region.

³⁷ Each of the three R ² change figures was significant at p < 0.001.

³⁸ ANCOVA statistic: F(1,442) = 16.69, p < 0.001, η_p ² = 0.036, controlling for standard demographics.

³⁹ Both indirect paths were statistically significant. See appendix for indirect effect statistics.

⁴⁰ α = 0.81, 0.77, 0.64, and 0.76 respectively.

⁴¹ T(497) = 2.25, p = 0.025.

⁴² F(1,356) = 0.22, p = 0.64; F(1,357) = 5.01, p = 0.028, η_p ² = 0.014.

⁴³ F(1,357) = 15.13, p < 0.001, η_p ² = 0.04, and F(1,357) = 7.90, p = 0.005, η_p ² = 0.02 respectively. Both ANCOVAs controlling for standard demographics.