a majority of modern studies on race rely on sampling. The sample size here (14%) was large enough statistically
This is not a sample in the statistical sense. This is a self-selected subset. Any bias in the groups in question towards answering or not answering this question is not assessed by the study, and the assumption that the sample is representative of the population is entirely unfounded.
A vast majority of studies in the US on race use self-reported data.
Which lets us know we should be taking the results of such so called studies with large grains of salt