Before You Get Too Excited About That GitHub Study | Hacker News Reader

Before You Get Too Excited About That GitHub Study

slatestarcodex.com

144 pointslladnar14 comments

This study, and the news articles about it, say a lot more about attitudes towards software developers, than it does about software developers themselves.

And those attitudes are disturbing. They describe the industry as a terrible, sexist place, even though the actual study shows the exact opposite.

This isn't just insulting to us software engineers, which it definitely is. It's also making the problem worse. Why should women be interested to join a field that is described as horrible to them?

1. "Our analysis (..) shows that women are harder on other women than they are on men. Men are harder on other men than they are on women.”

2. Not peer reviewed.

3. The confidence intervals may overlap.

4. Graphs all have non zero base lines.

I'm usually the first person fawning over any new SSC post, but by his own admission here he's just regurgitating the points from the original HN thread (https://news.ycombinator.com/item?id=11074587). Putting too much stock in this post is kind of "training on your validation data", as it were.

He's also doing the yeoman job of pointing out how rabidly the media will seize upon this sort of study (which a lot of people have identified as having serious flaws) and put up broadly general headlines that "fit the narrative", like this one:

"Suck It Dudes. Science Proves Women Are Better Coders." (http://gizmodo.com/suck-it-dudes-science-proves-women-are-be...)

I mean, I know that Gizmodo is not exactly the pinnacle of responsible journalism, but as mentioned in the article, the Guardian, the BBC, and Business Insider are not doing much better.

It's at least a useful service to pull all of the good points regarding the validity of the statistics and what they really say into one place and drop the outrage.

As seems typical with this sort of study, it's easy to reinterpret and cherry-pick things to say that it proves whatever you want it to say, but it seems that it doesn't really say much of anything conclusively.

Hmm... odd. Despite a decent number of upvotes, my earlier submission of this link[1] never made it to the front page. I thought it was flagged down by users, but maybe not?

Edit (after downvotes): Jeez, I was just curious, not accusing.

1. https://news.ycombinator.com/item?id=11092049

What the world really needs is more dudes taking the devil's advocate position on sexism issues.

What a lazy comment. I recommend looking at the actual study [1]. It is reported as showing that male programmers are incompetent and misogynist. However, when you look at the data (especially Fig. 5 on page 15) you'll find that this claim is baseless. Male outsiders showed the exact same drop in acceptance rates when their avatar was gendered as women outsiders did. No evidence for gender bias.

Looking at the data for insiders (left part of Fig. 5), you have evidence that male insiders have lower PR acceptance rates when their avatar was gendered but women insiders don't. Applying the authors' logic you would have to conclude that tech is biased against men, which is nonsense and which illustrates that the reasoning used in this study isn't sound.

[1] https://peerj.com/preprints/1733.pdf

Edit: Down voting instead of responding is lazy, too.

What the world needs is less journalism that picks any available thread out of ambiguous research to push their pet agendas. It's not just the "SJW-types", it's an across-the-board problem. It generates clicks and grabs eyeballs...

Calling that bullshit out is a worthy cause.

Scott (author) critiques bad statistics; that's all there is to it. He admits all the time that critiquing studies that are in favor of his beliefs makes him feel like a bad person, but altogether Scott is "in favor of niceness" and it shows in his writing. Scott ends up defending the fringe simply because they get the worst arguments thrown at them the most often.

He's not playing devils advocate, he's accurately interpreting the study.

What the world really needs is more truth being spoken regardless of any physical or social characteristic of the speakers. Any action based on lies or misinterpretations damages the world.

What the world needs is less people calling out others for who they are as opposed to the what they said.

My sarcasm detector isn't operational; if serious, why more dudes as opposed to dudettes?

Yes, that would work out really well for them.

Reply on news.ycombinator.com