This is an interesting post on the topic by Virginia Rutter, especially interesting given the concerns Chris Matthews has aired about women possibly voting on their gender than on the issues. What Rutter argues is that men were more likely to have voted their gender in Iowa than women, though perhaps both did:
Given that distribution, here's what we know about "voting one's gender" in Iowa: Out of all the men who voted for either Obama or Clinton, 60% of those were for Obama--that is men voting for a man. Men voted their gender 5 percentage points more than we would expect if their voting weren't influenced by their gender. Meanwhile, out of the women who voted for either Obama or Clinton, 46% of those were Clinton supporters--that is women voting for a woman. This means that women voted their gender more than we would expect—but only by 1 or 2 percentage points.
Her argument goes like this: About 61% of the votes went to either Obama or Clinton, and if we now regard that 61% going to the two candidates we have singled out as 100%, then the inside split of those votes was roughly 55% to Obama and 45% to Clinton.
Now, do the same exercise as above, but look at how men and women voted in two separate goes. For example, 58% of all men in the Iowa Democratic caucus voted for either Clinton or Obama. Now set that 58% as the new 100% and see how the male votes were split between the two candidates. The answer is that roughly 60% of those votes went to Obama and 40% to Clinton. So compared to the overall split, men were more likely to vote for Obama and less likely to vote for Clinton.
The percentage of women who voted for either Obama or Clinton in the Iowa Democratic caucus was 65. Out of that total 54% of the votes went to Obama and 46% to Clinton. Compared to the overall split women were very slightly more likely to vote for Clinton and less likely to vote for Obama.