beginner's specification problems

Hi, I am doing an assignment on unemployment (DV) and racial discrimination - specifically african-american population - and I am in doubt whether to use percent of resident population black resident alone, or black alone resident population. If I use the latter, would I need to switch from an OLS model to a count model such as NBR?

So far I used percent, but had a very low R2. Also, since I was trying to avoid mixing percentages and total, if I keep on running a regression with percentages I will not be able to assess the effects of gender (the dataset only has these in terms of total numbers and not percentages). Are there any suggestions for what I could do please?!

thank you a million!