Thanks for the answer. It does makes sense but looking at the actual percentage of black voters in the 2004 presidential election it was 11%. This poll has a guaranteed 15% blacks by adding 212 and has probably more in the 1,114 sample. that would put them past 15% in this sample and they only represent 11% in the 2004 presidential election.
The polled is skewed with an oversample of blacks.
The results are adjusted to make up for this. When all is said and done, the responses of the group oversampled should be in line with their actual share of the population.