Who holds Japan's national data science and AI program certification? Institutional sector, size and region in the MEXT MDASH scheme, 2021 to 2026

Since 2021, Japan's Ministry of Education, Culture, Sports, Science and Technology (MEXT) has certified university programs in mathematics, data science and AI at a literacy level and an applied basic level (the MDASH scheme), toward a national goal of literacy-level skills for every university and college student. Certification is voluntary and by application. This paper links the current list of certified institutions to the MEXT school code register and the 2024 national university directory, and asks how certification varies with sector (national, local public, private), size (undergraduate entry capacity) and region. The registered analyses follow an analysis plan recorded before the first result; the plan is not public and its timing cannot be verified independently (Appendix A). Among 767 four-year universities with a known capacity, 64.8% of private universities hold the literacy-level certification. In the registered primary contrast, 45.3% (106 of 234) of private universities with an entry capacity of 300 or fewer hold it, against 91.0% (121 of 133) of those with a capacity above 1,000: a difference of -45.7 percentage points (95% Newcombe interval -53.0 to -36.8). For the applied basic level the shares are 9.8% and 49.6% (difference -39.8, interval -48.8 to -30.3). Among local public universities the share also rises from the smallest band (46.0% of 63) to the next (77.4% of 31); national universities in the two bands with at least 20 institutions hold it at 88.0% and 100%. Within the small private band, universities outside the three large metropolitan areas hold it more often than metropolitan ones (51.6% against 37.7%), and certified small private universities first appear in the annual announcements in later rounds than large ones. Because certification is by application, the results describe who has been certified, not who teaches data science.