Thread by @antonioregalado

Holy crud, my current two favorite words “polygenic score” make 1st appearance in New York Times, in an editiorial about the genetics of intelligence. (!).

Lots to unpack ...I’ll (thread) the highlights.

nytimes.com/2018/07/24/opi…

The headline. Nice. Trying to head off, parlay, knee jerk reaction by the further left to anything to do with IQ and genetics.

Author is a geneticist who wants to create space for science...but only so much space, as we shall see

Here’s what prompts the editorial. A news study linking genetic factors to length of education.

This succinct summary is interesting for a couple of reasons.

Everyone involved in this research, and all the reporters writing about it, had to make decisions about how to handle it. Sensitive stuff.

The crux of of the sensitivity boiled down to one thing: are the genetic predictions strong or weak?

If genetic predictions of educational success are strong, then temptation/danger of using IRL.

authors of original report skinned cat by saying predictions really bad, but very good for further science stuff they want to do.

Ed yong here in Atlantic: theatlantic.com/science/archiv…

Ed showed the take on the data that makes it look bad. From this cloud of dots (education vs. gene score) were supposed to judge individuals? Nah thanks

But there is another view of the predictions that’s kinda startling. Which is what happens when you bin people up by their scores into 5 groups low to high. Then it really looks powerful.

College grad rate on left axis. Polygenic score quintile on bottom.

The author of the New York Times editorial, even though she’s taking a position similar to Ed, interestingly, chooses to highlight the more dramatic data.

I can’t say why she did. But in my own coverage I did the same. It was not, I think, the preferred message of the authors of the report, which appeared in Nature Genetics technologyreview.com/s/611680/milli…

Are these predictions from DNA of how far a person gets in school strong or weak? Depends on how you look at it.

In technical terms they explain 11% “of the variance” in the trait.

Authors of original report insist this has no practical value in policy or indiv prediction.

I am not so convinced. Some reasons:

-polygenic scores in the market, like 23andMe prediction of BMI to consumers, are worse or equal to this one (I’m pretty sure)

-medical researchers are stoked if they can predict 5% of variance in say heart failure and think it’s a product

What’s more, one of the authors of the Nature Genetics article, Dalton Conley, wrote in his book, The Genome Factor, that as soon as the score top 10%, which they just did, it’s gonna be time to talk about “personal eugenics”

Whereas in a lengthy FAQ accompanying the research the authors dispute any practical or policy implications whatsoever.

Since we know polygenic scores as strong or even weaker than this one do have utility, even only as 23andMe infotainment, I cant see the authors are correct that their education one is useless “because science”

More like its off limits for social or political reasons.

The NYTimes editorial argues polygenic scores for education have a behind scene role as a control variable in studies of social outcome, which is true, and, as author writes, lacks “dark allure” of eugenics or scarlet letters.

I’m a dark allure guy so I think I just got shaded by the New York Times!

Some past coverage:

Eugenics style embryo scoring;
technologyreview.com/s/609204/eugen…

Individual prefiction(scarlet letter): technologyreview.com/s/610339/dna-t…

This thread is to be continued. But if you have reactions to the NYTimes editorial on polygenic scores—-something of a landmark for this technology—-especially in regards to its subtextual elements, please post them here👇🏻👇🏻