Taiwan

The two percentage-point gaps for Taiwan are shown next to the unweighted mean across settings. Both are defined as female minus male. A negative choice gap means that among students who have already cleared the same academic bar, women are less likely than men to rank a STEM program first.

The table summarizes features of the admission system that bear on the construction and interpretation of the analysis. Untranscribed values are marked as pending.

Taiwan allocates seats at most of its general universities through a centralized clearinghouse operated by the Joint Board of College Recruitment Commission (JBCRC): 33 of the 35 public general universities and 32 of the 36 private ones participate, 65 of 71 in all. The admission unit is a university-major combination, termed a program, and roughly 1,659 programs are on offer in a typical year. The paper's data cover this clearinghouse for every application cycle since the 2002 multi-channel reform. Technological and vocational universities and colleges, 77 institutions in total, form a separate market and are outside the data.

Students take two national examinations: the General Scholastic Ability Test (GSAT) in January or February and the Advanced Subjects Test (AST) in July. Only after knowing their scores do students submit their rank-ordered lists to the clearinghouse, ranking up to 100 programs. University programs rank students by evaluating their exam performance, mainly on the AST, with subject weights set by the program. Students are then assigned to programs via a program-proposing deferred acceptance algorithm, so the standard strategy-proofness result for applicants does not apply directly in this setting.

Nor is the hundred-cell list limit slack in practice: the admissions committee's published statistics show applicants using a substantial share of the hundred available slots, so the paper does not rest anything on the cap being non-binding here. Across the ten settings, list caps counted in college-major cells run from two in Brazil and 24 in China to a hundred in Taiwan, with no numerical limit in Greece, and the caps do not order the choice gaps: Taiwan's choice component is the second smallest in the sample, on a hundred-cell list, while Brazil's two-cell list, the tightest cap observed, sits in the middle of the range. Taiwan's gap is in any case neither an outlier nor a driver of the paper's results.

Final placement results are announced in August. Compliance is strictly enforced: students can enroll only in the program to which they are assigned, or opt out of the system altogether, with very few exceptions applied to top-scoring students.

The cycle opens with the General Scholastic Ability Test in January or February. The Advanced Subjects Test follows in July. Students submit their rank-ordered lists to the clearinghouse after their scores are known, so the list can respond to where the student actually stands in the score distribution. Final placement results are announced in August, after which students enroll in their assigned program or opt out of the system.

The project's sources record the calendar at this resolution — the two exam windows, the fact that list submission follows score release, and the August announcement — but not the precise dates of the list-submission window between the release of AST scores and the announcement of results. That finer detail would need to come from the country team.

Participation in the clearinghouse is defined over Taiwan's general universities: 33 of 35 public and 32 of 36 private general universities admit through it, 65 of the 71 in the country. The two public and four private general universities outside the platform are not named in the project's sources, nor is their share of enrollment recorded there. Across the participating institutions, roughly 1,659 university-major programs are offered in a typical year. Tuition is charged, and financial aid is available.

The technological and vocational sector sits entirely outside this system. Taiwan's technological and vocational universities and colleges, 77 institutions in total, form a separate admissions market, and the paper's data do not cover it. The 65-of-71 figure therefore describes coverage of the general-university sector, not of Taiwanese higher education as a whole.

The paper's Data section notes that in eight of the ten settings the centralized platform is the route into the most selective universities and their STEM programs, and that where participation is partial the institutions outside the platform are typically less selective; the two exceptions it flags on this point are Brazil and Uganda, not Taiwan.

The performance measure is the score students are ranked on for admission: Advanced Subjects Test subject scores, weighted by program. Programs evaluate applicants mainly on AST performance, and students know their scores before they submit their lists. The underlying records contain students' scores on all subjects in both the GSAT and the AST, alongside their reported rank-ordered lists and final admission outcomes.

For the paper's analysis, performance groups are formed within setting and cohort over the full distribution of score holders. The measure is constructed identically for women and men; no admission adjustment, weighting, or bonus enters it. Because the score scale generates ties at the cutoff, the top decile covers 9.7 percent of Taiwanese score holders rather than exactly 10 percent.

In the 2010 analysis cohort there are 72,067 score holders. The top decile contains 6,998 students, 2,766 women and 4,232 men, so women are 39.5 percent of the high-achieving group against 49.8 percent of all applicants. The pipeline gap runs against women in this setting: women are under-represented at the top of the score distribution.

The data cover the 2002 through 2010 application cycles — every cycle since the 2002 multi-channel reform — and the analysis uses the 2010 cohort. The system processes about 72.1 thousand applicants per year, with about 27.7 thousand admissions to STEM programs per year. Taiwan is one of only two settings in the paper, with Uganda, where women do not outnumber men among applicants: women are 49.8 percent of the applicant pool.

The analysis sample is defined over score holders in the 2010 cycle: 72,067 students form the full distribution against which performance groups are constructed, and the top decile, the group the paper's main-text analysis is based on, contains 6,998 of them. Ties at the cutoff mean the top decile covers 9.7 percent of score holders rather than exactly 10 percent.

Additional record-level exclusions, repeat-applicant rules, and field mappings are documented in the replication code.

Admission rules in this setting do not differ by sex: the paper's Table A.III comparison of the ten systems records no sex-conditional admission rule for Taiwan. Eligibility runs through academic performance alone — Taiwan is not one of the two settings, Greece and Uganda, where the paper qualifies that claim — and the performance measure used in the analysis is constructed identically for women and men, with no admission adjustment, weighting, or bonus entering it.

The data come from the administrative registers of the Joint Board of College Recruitment Commission (JBCRC), the agency that operates the centralized clearinghouse, and cover the 2002 through 2010 application cycles, that is, every cycle since the 2002 multi-channel reform. The analysis uses the 2010 cohort. The records contain students' exam scores on all subjects in both the GSAT and the AST, their reported rank-ordered lists, and their final admission outcomes.

The admission records are augmented with insurance records from Taiwan's Ministry of Labor, which supply rich information on students' demographics and family background, including age, gender, residential location, and parental income and education level. Attributes of the high schools from which students graduated are supplied by the Ministry of Education.

Access conditions and the agency contact route have not yet been supplied for publication.