If it's true at the county-level, it's true at the national level (which is also stated in the abstract). By specifying it's additionally true at the county level, it makes a much stronger claim -- that high racial bias does not exist only in high-crime areas, and as such, racial bias is not explainable by crime rates.