What crap.
60% of the sample makes up <20% of the voting public...and this isn't even REGISTERED voters.
I believe they only used the oversampling of blacks and hispanics to make their subcategory results statistically valid—but did not include them all in their regular survey totals.