So this is pretty typical of the low-information content you get from the genetic racists. The majority of this post is just blather but there is one (1) specific claim about genetics: that the molecular genetic contribution to IQ keeps going up every year. This is false. A 🧵:
The first study in 2011 into the heritability of IQ using molecular genetic methods found moderately high estimates 40-51%. But this approach was flawed technically (estimator bounds and population structure) and conceptually (environmental confounding).
Fast forward to 2023, using hundreds of thousands of people from the UK Biobank, Williams et al. [] ran a battery of analyses to refine a high-quality IQ estimate. The heritability ... 0.20 (with very precise error). pubmed.ncbi.nlm.nih.gov/36378351/
But this doesn't address the conceptual issue of environmental confounding. For that, Howe et al. used a large-scale within-family analysis, which does a much better job of isolating the genetic component from shared environment. Their estimate of the heritability ... 0.14!
So we've gone from 51% to 20% to 14% as the field has learned how to apply these methods more precisely and address confounding. Researchers that raised concerns of environmental confounds and stratification were proven right, and it's unlikely that we've resolved all the issues.
This trend is even more extreme for Educational Attainment, an easier to study trait with more practical relevance. Initial studies had estimates of 22% heritability which decreased to ~15% as better methods and more representative cohorts were applied.
When Howe et al. finally employed a proper within-family analysis their estimate of the heritability was ... just 4% (with a tight error bound). That's right, the *entire* common genetic contribution to educational attainment, a major status-driving factor, is a rounding error.
So the technical point is flat out wrong. And this style of argument mirrors a general trend. Charles Murray has been promising that his views will be vindicated in "just five years" since 1994!
Steve Hsu has been arguing that a 60% accurate genetic predictor is just around the corner since 2011, and last year declared himself vindicated! (As we just saw, the GWAS heritability of IQ is <14%, it cannot possibly reach 60%)
The shell game they play is: (1) claim that there's *lots* of evidence out there; (2) present one or two sketchy correlations based on bad methods; (3) promise that the better results are just around the corner. In the real world, their position keeps losing. /fin
Sailer tends to disappear whenever he is addressed directly so I'm not hoping for a response. Maybe in a day we'll get another story about sports or movies. But here are a few more unaddressed errors in the whole race/genes/IQ/outcomes project:
I wrote about how population stratification in genetic analyses led to a decade of false findings and almost certainly continues to bias emerging results. But we are starting to have statistical tools to sniff it out. A 🧵:
First, stratification = genetic structure + environmental structure. If two populations have some genetic variation (e.g. due to drift) and differing environmental influences on a trait, that will induce a false/non-causal correlation between genes and the trait.
When such false correlations are further aggregated into polygenic scores, they can accumulate into very large *apparent* genetic differences between even closely related populations. And these false differences will mirror the environment: environment looking like genes.
This is a good example of how pointless a lot of the "data oriented" conversations on X are. DataRepublican, a DOGE analyst, makes a bold claim that 0/60,000 sampled government contracts had outlays < potential award ...
Judd Legum, a journalist, points out that having outlays lower than the potential award amount happens frequently, explains why, and highlights a number of specific examples. Seems like a pretty basic error, should be easy to acknowledge right?
Wrong. DataRepublican first responds with a bizarre claim that they hadn't sampled enough contracts because "hard drive overheated", but that the methodology is sound. Then notes in passing that there was a bug, but follows it up with a brand *new* analysis.
So it turns out the person running this account and accusing mainstream behavioral geneticists of fraud was actually one of the authors of the discredited Pesta at al. paper that was being criticized. Pretending to be an objective third party so they could sling mud.
FWIW I don't have a problem with anon accounts and enjoy interacting with many on here. I understand that people may want to partition their on-line/IRL lives. But setting up a sock puppet persona so you can aggro out on colleagues that disagree with you is pathetic.
And using a pseudonym so you can self-cite and email your own preprints to other researchers for them to cite is just sad.
It's been interesting seeing Murray become an Ibram X. Kendi figure but for the right. Everyone knows his "analyses" in Human Diversity -- like comparing non-causal allele frequencies between populations -- are completely bogus. Razib knows this too.
But Murray says the things that are politically correct and pleasing to that audience's ego so he regularly gets trotted out for softball interviews and never needs to exhibit any rigor.
This happens over and over. Here's AEI hosting a debate between Murray and Princeton Professor Dalton Conley. Conley explains that the claims made by Murray about genetic differences are unsupported by the data and often gross misinterpretations.
This thread and especially the underlying LessWrong post are a good demonstration of the IQ super-baby conspiracy theory that seems to be gripping Silicon Valley. Here's how it works ...
First, claim that we already have the knowledge of how DNA affects college graduation rates but no one is interested in applying it. This is false, we almost never know *which* genetic variant is actually causal nor *how* it actually influences the associated trait.
This is also a challenge the field is very interested in understanding, including large-scale NIH-funded consortia efforts like IGVF (). Claiming that we already have the knowledge also undermines such efforts.genome.gov/Funded-Program…
How population stratification makes environments look like genes. A short 🧵:
Start with two populations undergoing neutral drift but with no frequency differences on the alleles that influence the trait (i.e. no genetically causal population differences).
Generate a phenotype that differs slightly between the populations for entirely non-genetic reasons (i.e. a difference in the environmental means). Drift + environmental differences = population stratification.