That isn't what is typically done in these studies. Rather than using a valid method (statistical correlates of names with race) they instead use forced choice ratios.
They ask a sample of 100 people is __name__ black or white?
Then based on the forced choice percentages they assign a 'race' to the name.
They tend to also use extremely common 'white' names and extremely uncommon 'black' names (The studies that have used common black names such as Ebony; show that the common black name ends up with better response rates than the less common white names).