You remind me of a very important point: including any more than one race has implications for study validity and could lead to bias in itself. If you include other races, you will need a larger sample size to allow controlling for race in your models, whether through regression or stratification. Limiting to one race…