A funny thing about the gender gap in the Quinnipiac University poll

According to a Quinnipiac poll released on October 2, 2012, 56% of likely women voters support Obama whereas 38% support Romney:

1. If the election for President were being held today, and the candidates were Barack Obama the Democrat and Mitt Romney the Republican, for whom would you vote? (If undecided) As of today, do you lean more toward Obama or Romney?

LIKELY VOTERS.............................................
                     Tot    Rep    Dem    Ind    Men    Wom    Wht    Blk
                                                                         
Obama                49%     7%    94%    45%    42%    56%    42%    94%
Romney               45     91      5     47     52     38     53      2 
SMONE ELSE(VOL)       2      -      -      4      3      1      2      1 
DK/NA                 4      2      2      5      3      5      3      3 

There were 1,912 respondents. 947 of them were women. Therefore, 530 women surveyed declared a preference for Obama.

According to the same poll, Obama's support among white women was 49%. There were 819 white women responded to the poll. Therefore, 401 white women supported Obama. (The demographic breakdown of the poll is here.)

That means, 129 non-white women voters must have expressed support for Obama.

Subtracting 819 (the number of white women) from 947 (the total number of women) gives us 128 as the total number of non-white women in the sample.

The mismatch between total non-white women voters and non-white women voters supporting Obama is likely due to rounding error.

What we have is a situation where all non-white women likely voters in the poll supported Obama for president.

This indicates to me that the "gender gap" is not really driven by "gender".