Here is a language log post which may demsytify some of the statistics surrounding this issue. And here is Mark Liberman, Oct. 6 2007.
OK, so imagine coming into a door labeled "the room of unhappy people". You enter, and find yourself in a hall with 51 to 54 women, and 46 to 49 men. Do you think that you could decide which sex predominated, without lining everyone up and doing an explicit count?
Now imagine that you walk through two such rooms, where the first one is around 51-to-49 female, and the second is around 54-to-46 female. Do you think that you would notice the direction of difference in the sex ratios, without another pair of line-ups?
More to the point, do you think that you could spin differences like these into today's second-most-emailed NYT story?
If your answer is "yes", then you may have a future as a science writer. (Or, perhaps, as an economist...)