Undersampling of women in polls skews results ?

When I opened the newspaper this morning (Washington Post) the headline proclaimed "Obama, Clinton Are Even in Poll".  The accompanying article indicated that in a national poll Hillary is now leading Obama by 47 to 43, within the margin of error.  I went online to look at the crosstabs and one thing immediately jumped out from the numbers:  "sex (in sample): 48% male, 52% female"

Source: http://www.washingtonpost.com/...

The gender breakdown above is way off when comparing with how voters have voted already in the early Democratic primary states:
IA - 57% female
NH - 57% female
MI - 57% female
NV - 59% female
SC - 61% female
FL - 59% female

Avg. 58.33% female

Source: http://www.cnn.com/...

It should be noted above that when I say "way off", a difference between 52% female and 58% female is HUGE under these circumstances.

I could not find the breakdown for the other recent national polls (Fox, Gallup, Rasmussen), and, likewise, there’s a dearth of information in the state polls.  The only recent state poll breakdowns by gender were the following:

Field Poll (CA) – 56% female
SUSA (MO) – 54% female
SUSA (AL) – 56% female
SUSA (CT) -55% female
SUSA (NY) – 59% female
SUSA (NJ) – 55% female

State Poll Avg. (from above) – 55.8% female

Sources: http://www.surveyusa.com/...

http://www.field.com/...

Rasmussen and Zogby have done a lot of state polls recently but they have the crosstabs either available only for paying members (Rasmussen) or not available as far as I can see (Zogby).  In several state polls, they did however provide hints at how the gender vote broke down.  In their recent Alabama poll, which Clinton led Obama by 5 points, Hillary had an 8 point lead among women, and a 2 point lead among men.  In Rasmussen’s poll of California, which Clinton led Obama by 3 points, Hillary led by 10 among women and trailed by 8 among men.  In Zogby’s poll of California, which Obama led by 5 points, Hillary trailed among men by 20 points but led among women by 11 points.  It’s hard, if not impossible, to extrapolate the gender breakdown based on these numbers alone (because other mathematical factors come into play, i.e. don’t know how many are undecided among gender groups, for example), but I played with the numbers a little and my gut feeling is that in most of the Rasmussen and Zogby state numbers, the gender breakdowns are probably much closer to the Washington Post’s low-fifties number and not close to the actual voters’ performance number in the high-fifties.

Sources:  http://www.rasmussenreports.com/

http://zogby.com/...

What this all means is that many national and state polls we are looking at are may be significantly underestimating the numbers of women who will vote on February 5.  (And again, this is based on a small number of actual poll results -- Washington Post national poll; California Field Poll; several SUSA state polls -- combined with my gut feelings re. how the state polls from Rasmussen and Zogby look.)  As it appears that women may break more proportionally for Hillary, we may be in for a surprise on Tuesday evening where Hillary’s numbers are greater than currently anticipated.