The percentages you mentioned would be relevant if I, the author, was look to show intent to discriminate but I was not. I am just showing the effect of the pro

The percentages you mentioned would be relevant if I, the author, was look to show intent to discriminate but I was not. I am just showing the effect of the program and for that the percentages you mentioned are entirely irrelevant. The sample size (14%) was large enough statistically. Might also be interesting to see if this trend was larger when white owners did not identify themselves.