Difference between demographic Sex and Gender. Which one is more accurate?


#1

So, there is at least two ways to tag/see the gender of the user:

demographic.gender (Datasift own gender augmentation?)
demographic.sex (demographics pro augmentation?)

So, in a case you want the gender for Twitter users, there is already some difference in the results of these two. We would rather use just one of these across the our system.

My question is which one should be more accurate and how the gender is determined in both cases. Is there some brilliant way already to combine the results to make it super accurate :slight_smile:

Thanks,

  • Juha

#2

Correct; demographic.gender is DataSift's own augmentation, while demographic.sex is provided by our partner, Demographics pro. 

The value of the demographic.gender augmentation is calculated by analyzing the name of the user, and the country they state they are from (if that value exists). i.e. Someone called "Jan" is likely to be female if they are from the US, but more likely to be male if they are from a Scandinavian country.

I can not comment on exactly how the demographic.sex value is calculated; you will need to bring this up with Demographics Pro.


#3

Thanks Jason!

I think it would be highly valuable for your clients I you could run some kind of test set about this with both augmentations and make a comparison? Based on that, would it be possible to make some kind of scoring CSDL that would be more accurate than just one of the augmentations alone.
Just my 2 cents

  • Juha