US bloggers and gender genie

OK. The results are in.
US male bloggers: 20 out of 26 correct (78%) and 6 out of 26 wrong (22%).
US female bloggers: 8 out of correct (32%) and 17 out of 25 wrong (68%).
Bear in mind that not of all of the surveyed blogs were used - those which are mainly photos, for example, were not useful for this exercise since it needs a good amount of text to work with.
However, the finding appears to be that gender genie is a lot more accurate in identifying male bloggers than female ones!

Using Gender Genie

For a bit of fun I have taken the latest posting of all of the UK blogs that I am studying and have put them through Gender Genie. Obviously this is more of test of Gender Genie than it is a test of the bloggers! The results so far are mixed. Gender Genie uses a simplified version of an algorithm developed by Moshe Koppel, Bar-Ilan University in Israel, and Shlomo Argamon, Illinois Institute of Technology, to predict the gender of an author. You put a piece of text in (they recommend that it is over 500 words for the most accurate results - sometimes not possible with my bloggers), choose whether it is fiction, non-fiction or a blog entry, and then ask it to analyse the words and indicate the gender it thinks the writer is.
How well did it do with my UK bloggers?
For the male bloggers, it spotted 23 out of 33 correctly (70%), got 9 out of 33 wrong (27%) and 1 unknown.
For the female bloggers, it spotted 12 out of 26 correctly(46% - not so good) and got 14 out of 26 wrong (54%).
So is it better with men than women? I will see what the US bloggers do for it.
Funnily enough, I know that it does not analyse the subjects of the posts, but it was funny to see it decide that the blogger who had just posted about going to have her first scan because she is 3 months pregnant was male!