Quoting a twitter reply for an explanation:

"Each row is an outcome. The orange dot signifies how strong the genetic correlation is for that outcome from genes not related to IQ. The blue dot is the correlation with genes that correlated with IQ. The grey dot is the correlation of educational attainment with the outcome. When theres an asterisk, it means there was a significant difference between the IQ based genetic correlation and the non IQ based, so you can infer how much of a role IQ plays for that outcome as compared to non IQ related genes.

So take cigarettes per day for example. You can see that the non IQ genes were more correlated with cigarette usage than IQ genes, which tells us a little but about what mediates that outcome."

What do the negative vs. positive correlations mean here? I mean, you can smoke more or less, so that's clearly a scalar variable, but how are, for example, "non IQ genes" a scalar variable that can have a positive or negative correlation with smoking?

I think, and I am no expert in this.. but a genetic correlation is between two traits.

Negative genetic correlation: if X increases then Y decreases

Positive genetic correlation: if X increases so does Y

I think each trait on the graph should be paired with educational attainment as a second trait.

So: GeneticCorrelation(Cigarettes, EDUCATION) -> Negative

GeneticCorrelation(Longevity, EDUCATION) -> Positive

My take is this: The chart shows the relationship between the genes responsible for higher education X and {insert trait} = Y, where X is further split into IQ and non-IQ..

Here's an explanation in plain english by one of the authors, Dr. Paige Harden:

https://medium.com/@kph3k/investigating-the-genetic-architec...

Thanks. The preprint looks digestible too, if slowly.

https://www.biorxiv.org/content/10.1101/2020.01.14.905794v1....

What does it mean to have a negative correlation?

Positive correlation to me would mean that from the Dna you can predict if someone has attribute X or not. What does a negative correlation mean?

Positive correlation between DNA and X means if DNA then more likely X. (A correlation of 1 is perfect, so if DNA then always X.)

Negative correlation between DNA and X means if DNA then more likely not X. (A correlation of -1 is perfect, so if DNA then never X.)

Positive correlation between x and y means that if someone has x they are more likely to have y. Negative correlation means someone from with x is less likely to have y.

Positive corr would mean, "The more of X we see, the more of Y we see". Negative corr would mean "The more of X we see, the less of Y we see"

Y is the measurable outcome ("has/has not ADHD", household income, whatever), right?

What's confusing GP and me, I think, is what is X here and what does having more or less of it mean?

Both x and y have to be measurable. Here, x could be score on IQ test. Since score on IQ tests are somewhat well correlated with academic achievement, then more x (higher IQ score) means more likely y (higher academic achievement). It’s important to note that it’s only more likely. Correlation is a predicter of a relationship given a population, not a dictator of outcomes for individuals.

Yeah exactly

What is Y in this graph please? Possibility to have correlation?

X is one variable and y is another. The correlation is, given x, how likely can we predict y.

So if x is IQ score, and y is income, and if there is a strong positive correlation, then given high IQ (x) we can confidently predict high income (y).

If there is no correlation between x and y, then given x, we cannot predict y.

y(x) = x

y(x) = -x

This sort of exploration of genetic correlations is really intriguing to me, largely because of the related question of human inequality. This study essentially affirms that certain genes act as a risk factor in life. To simplify my point: if we know the genes that cause some to be more likely handicapped in life, should we give them extra support? Should we give some personal tutoring to everyone lacking x,y and z genes? Free gym membership to anyone lacking the (probably non-existant) gym rat gene to encourage exercise? I'm fully aware that this would be premature, but maybe in 20 years?

Something like extra screenings for anyone with higher genetic chance of cancer isn't controversial, but I wonder if we'll ever be ready to go further than that.

What do you think of as risk factor in terms of inequality? I read the table more in terms that monetary help for the poor would be the most important, since most of The educationally important qualities seem to be non-cognitive, based on this study.

If you meant the higher chance of mental illness correlating with IQ genes, than I think mental illness treatment as whole should substantially change to make it more socially acceptable, and help paying for it for those who need it.

I'm surprised to see that household income correlates relatively strongly with both IQ and non-IQ genes. I would've imagined IQ genes would be dominant there.

Edit: after reading the blog post I'm not surprised anymore. Since non-IQ genes are defined as "...behaviors and abilities that are not measured by traditional IQ tests but are thought to help people be more successful in school, in their jobs, and in life generally." it would be surprising if they didn't have significant correlation.

I don't find that surprising at all. Emotional intelligence plays a strong part in both holding a job and a family. I am assuming here that IQ is disassociated with emotional intelligence.

It's kind of hard to understand that graph without at least part of the corresponding article. Maybe you could paste something that helps in that vein?

It's hard to understand this chart. Also hard to disentangle correlation from causation...

Attempts to correlate genetic traits to IQ smacks close to eugenics to me and gives me a bad case of the weirds.

Can someone describe how would this graph would look like on classical 2D xy scale? Let's say for a major depressive disorder and household income please? This data looks super interesting but they are hard to read. At least for me. Thanks you!

Is this saying that if you have high income and delay having children, then you'll probably achieve more academically?

What's going on with the Schizophrenia line?

Some non IQ genes are strongly correlated with risk of getting this disease.

Reported for trying to start a flamewar.

I'd be happy if anybody would like to respond to the points that I made, if that's what you mean?