Hacker News Reader

Comment by aa143 | Hacker News Reader

I think there is something to this. This is an interesting article showing that on GitHub, female coders have their code approved at a higher rate than men - but only if their gender is not disclosed.

https://www.google.co.uk/amp/s/amp.theguardian.com/technolog...

View on HN

That article has been thoroughly disproven on multiple sites, multiple times.

http://slatestarcodex.com/2016/02/12/before-you-get-too-exci...

Seems like this does still suggest at least that in the aggregate females should have their code approved at roughly the same rate as males, which is far from what FB showed.

Facebook hasn't released any of their data. They have said normalized for experience, the gender bias goes away. So I can't agree with you based on the WSJ article.

Also I personally have a hard time believing a company with Sheryl Sandberg as it's COO would lie about their conclusions. She is a strong positive force for increasing women in tech.

They said normalized for rank (aka promotion level), not experience, which one might reasonably expect to also be influenced by bias. The study that showed a bias normalized for FB experience.

If Sheryl Sandberg personally says she looked into this and agrees that the study is mistaken and there is no statistically significant bias, that would definitely impact by assessment. I agree that she is great.

I don't seem to understand why you refuse to take Jay Parikh at his word. He is a indian minority who stands for diversity, and yes he often works very closely with Sheryl for many years. Jay Parikh is not a suit or corporate hr person at all. http://www.today.com/popculture/sheryl-sandberg-speaks-new-g...

Has he made a public comment on this? The only reporting I have seen has been third hand. I am taking the reporting at its word, which is that he did an analysis using alternative dependent variables that showed no significant difference. I'm also asserting that his analysis is more likely to be flawed than the original study, since promotion level is not clearly an independent variable whereas Facebook tenure is.

If the facebook engineering department is biased against women, they are likely to both promote deserving women less and reject their code more. If you do an analysis using promotion level to predict code rejection, you may not see any evidence of bias in this scenario, no matter how severe it actually is. Do you disagree with this possibility? Can you point me to anything that Facebook has said or done that explains why they are not concerned about it?

The research you are trying to dismiss several times in this thread as "suits" was done by Parikh.

After reading your replies to myself and other users in this thread,You clearly have an issue with him, by the rather extreme lengths you've gone to discredit him in this thread.

I don't want to be part of your crusade against an ally of diversity.

I had never heard of him before this story. I think it is unfair of you to accuse me of some vendetta merely because I don't take the convenient, proprietary, flawed company study results over the better designed and more open engineering results.

Reply on news.ycombinator.com