Hacker News Reader

Comment by cm2012 | Hacker News Reader

cm2012

I usually love your moderation, dang, and agree with you 99.99% of the time. But I think the parent here's criticism of the article does add something - the idea of genetics as a powerful confounding variable. He could have expanded on it, but it didn't add nothing.

View on HN

Criticizing something that could be wrong with the methodology of a study is always 100% inappropriate. Either you've read the study and understand what it was trying to show and the methodology the authors used, in which case by all means demonstrate that and then state your criticism, or else you haven't and shouldn't comment.

But trying to get others to not read someone else's work or take it seriously based on something you dreamt up that could be wrong with the study design is a completely inappropriate.

"Accounting for genetic confounds" is extremely difficult--so difficult that no one in the world knows how to do it reliably--for basically any aspect of human behavior.

So the dismissal reads like "there's no excuse for not doing an extremely difficult thing that no one knows how to do reliably." I don't find that to be a useful or informative comment, so I agree with dang's approach here.

I give the post credit in that it provoked your comment and some potentially enlightening discussion for someone not entirely familiar with the study of human behavior (like me). For instance, I would have thought that twin studies might be considered reliable for purposes of accounting for genetic confounds with respect to aspects of human behavior. It sounds from your comment that they are not.

On the OP getting warned, it wasn't immediately clear to me if the objection was more about a) knowledge that Matthew Yglesias didn't read the study and therefore was not in a position to summarize his analysis with a pithy analogy and be quoted, b) the use of any pithy analogy itself, or c) the use of an analogy that seemed to explain a confounding issue? There seemed to be some additional, non-obvious context missing (at least to me) that led to the warning.

There's a gazillion twin studies out there. There are huge databases of twins to facilitate their creation. It's really not that difficult. They might be imperfect but I would not call them unreliable.

We already know from earlier twin studies that genetics play a large role in literacy and numeracy.[0][1] So it's not like this is a shot in the dark, everyone already knows it's a confounder that must be taken into account.

[0] https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1469-7610.... [1] http://journals.sagepub.com/doi/abs/10.1177/0956797613486982

Reply on news.ycombinator.com