In case you were curious.

Likelihood of you being FEMALE is 3%

Likelihood of you being MALE is 97%

SiteMale-Female Ratio

google.com0.98

yahoo.com0.9

youtube.com1

wikipedia.org1.08

amazon.com0.9

craigslist.org1.13

facebook.com0.83

walmart.com0.77

paypal.com1.04

cnn.com1.35

blogger.com1.06

imdb.com1.06

flickr.com1.15

weather.com1.08

nytimes.com1.13

typepad.com0.94

washingtonpost.com1.15

orbitz.com0.83

drudgereport.com2.08

indeed.com0.72

slate.com1.11

wired.com1.41

npr.org0.98

politico.com1.7

salon.com1.13

beliefnet.com0.45

salary.com0.77

abebooks.com0.96

poetry.com0.68

gmail.com0.9

slashdot.org1.74

economist.com1.47

haloscan.com1.22

scienceblogs.com1.41

dailykos.com1.56

w3schools.com1.27

pricewatch.com2.77

(some of these must be pop-ups and what not, I don't recall visiting the sites)

- Log in to post comments

According to that program, so am I, apparently: 96% likelihood. "I do not think that word means what it thinks it does...." :)

Hmm... that's weird, it thinks I'm a chick - mostly due to visiting banking sites?!

Likelihood of you being FEMALE is 70%

Likelihood of you being MALE is 30%

SiteMale-Female Ratio

myspace.com0.74

youtube.com1

facebook.com0.83

cnn.com1.35

wellsfargo.com0.87

intelius.com0.75

fedex.com1.06

wamucards.com0.82

zillow.com1.13

megaupload.com1.5

vonage.com0.96

experian.com0.69

amazon.co.uk1.11

menupages.com0.72

familysearch.org0.67

orkut.com1.08

haloscan.com1.22

travelblog.org0.96

americanchronicle.com1.17

There goes mine:

Likelihood of you being FEMALE is 52%

Likelihood of you being MALE is 48%

I am used to misconceptions due to my name, but this analysis totally added another dimension.

Some of the more interesting ratios from that site: moveon.org 0.55, abc.com 0.47, fark.com1.82, bungie.net 1.27 (makers of Halo), barackobama.com 0.68 (vs. johnmccain.com 1.27), ea.com 0.68, snopes.com 0.74 (vs. straightdope.com 1.17), and erowid.org 1.06.

I wonder if someone can find a porn site with a ratio less than 1. It's a shame these stats aren't reliable...

60% female last night, 65% female today with only one new site (checked CitySearch for the name of an eyeglasses store). I only had 14 or 15 sites from the top 10K, so there's obviously lots of variance.

There's probably bias in the sites included -- guys are less conformist, so the typical sites I visit aren't on there, nor is anything like them on there. You'd find the same thing if you took the 10K most frequently used words, estimated gender ratio for each, scanned a person's writings, and estimated gender that way. Guys use 50-cent words more often because we're bigger showoffs.

I'm a dudette, but according to that test, there's a 99% probability that I'm male.

Regarding the women posting *here* about their man-like internet histories, two words: selection bias.

I'm another female with a "Likelihood of you being MALE is 99%."

And I'm not so sure about Quantcast's stats after this. Jezebel.com is 51% male? Color me skeptical.

Ok, I got 100% male. My most female sites were toysrus and tmz (Brooke Hogan tanning, iirc).

Drudge was over 2:1 male. Surprising to me. And I read some articles about new CPU benchmarks a while back. Total lack of female readers not a surprise in that case.

Forgive the question from the statistical ignoramus.

The methodology here according to the site was to compute a ratio of male visitors to female visitors for each site, take the product of all the ratios of the sites you visited, dub this product r, then 1/(1+r) is your probability of being female.

Can someone explain this method to me? I'm decent with math, but not up to speed on statistics.

I think I get what the product of ratios is for, and why its not entirely reliable.

Essentially, the goal is to compute what the ratio of men to women would be in the set of all people who visited exactly the same set of sampled sites as you.

So, supposing people visited sites randomly based only on their gender, then if the male/female ratio for the two sites you visited were each 2:1, then the ratio of men who visited both sites to women who visited both sites would be 4:1.

However, this ignores correlations between sites. If the second site were a popup site that everyone who went to the first site got routed to, then the male/female ratio for people who visited both sites would be the same as the male/female ratio for visitors to the first site alone---and the estimator would have a ratio twice as high as it should be.

The problem is, we know that, even without the issue of popups, the web is broken up into highly correlated clusters of sites with similar viewership (e.g. new york times and washington post; abebooks and amazon). So, the accuracy of the estimator cannot be counted on to improve as the number of sampled sites increases, as correlations begin to have stronger effects as more correlated sites are measured.