Also, the fact that LW itself isn’t smiley friendly. An interesting project would be to gather data from the real life facebook pages of both males and females on LW and see if a discrepancy shows up there. People would have to volunteer their facebooks for you to look at which might cause a bit of a selection effect. (The less trusting/interpersonal types might be less likely to both volunteer their fb, and to use smileys)
The reason I say this, is because I severely limit my smiley usage on LW.
One user who’s part of the female dataset has already reported cutting out the smileys deliberately. As I say, I don’t put much faith in the results.
I did consider scraping lesswrong.com for data, but a) I wasn’t sure of the etiquette b) I don’t have a list of female users (maybe I can get them from the survey?) c) it’s a lot more coding.
Okay, after threatening, I had a go at hacking up a smiley gender detector for lesswrong irc.
Looking at the counts of smileys-per-message by nick, no obvious pattern.
Looking at averages:
male avg 0.015764359871 female avg 0.0194180023583
The dataset I’m using is so male dominated I don’t think the results can be particularly meaningful.
Also, the fact that LW itself isn’t smiley friendly. An interesting project would be to gather data from the real life facebook pages of both males and females on LW and see if a discrepancy shows up there. People would have to volunteer their facebooks for you to look at which might cause a bit of a selection effect. (The less trusting/interpersonal types might be less likely to both volunteer their fb, and to use smileys)
The reason I say this, is because I severely limit my smiley usage on LW.
One user who’s part of the female dataset has already reported cutting out the smileys deliberately. As I say, I don’t put much faith in the results.
I did consider scraping lesswrong.com for data, but a) I wasn’t sure of the etiquette b) I don’t have a list of female users (maybe I can get them from the survey?) c) it’s a lot more coding.
How are you determining gender of users?
The number of female users is so small I just hardcoded known female nicks.
As I say, I don’t think the results are particularly meaningful.