This too is potentially problematic. I lived in CNMI for a while. The local population that was the dominant group was Chamorros. There were 13 white people on the whole island I lived on and 3 African Americans out of about 3,000 people. There was a sizeable chunk of Chinese workers who worked at the casino. Yes, I was part of the minority community as a white person and, yes, I was subject to bias. (It gave me a much better appreciation for African Americans in the USA to be watched like a hawk everytime I entered a store, having people occasionally follow me around to watch me.) Some countries have radically different experiences with race. Think of Japan, or places like Korea. I have had white, female American acquaintences tell me they were mistaken for Russian prostitutes in both countries.

If I was trying to get a better feel for this as a potential issue, I would do a combination of research methods and then collect them all, and try to arrive at a wider picture. This would include examining any survey research done on the existing body of literature, getting infobox data for things like race (and then tabulating that against country of origins and native language), looking at where identified "minority" editors are contributing article wise, looking at those experiences in terms of say total edit reversions on articles related to that topic. (I can remember an issue about African American women in professions categories that led to the creation of the article "African American women in politics" on English Wikipedia because there was a rationale that the failure of an article existed showed the lack of necessity for this specific category grouping.) I would also do sampling on photos from meetups. I would go through any chapter reports and see what chapters have written about this internally inside their own research or based on their own membership details. I would look at the representation of who was selected by the WMF in fundraising banners. (Indians from India is different than Indians from say the USA or Australia because of the bias inside their own country.) Then I would compile that all together. While it would not be as simple or pretty as one methodology, a mixed methods approach would more likely give a better idea of the picture when the population boundaries are clearly unknown.