John Benning

SAVE THE DATE: the yearly NY Population Genetics meeting will be back on March 9 2026, generously hosted by the
@simonsfoundation.org. Details to follow. Please RT.

We thank our generous colleagues across various disciplines whose feedback helped improve this work.

We are concerned that a publishing culture which rewards sensationalism may drive a decline in standards of rigor. In that respect, everyone has a role to play: it is crucial that researchers, reviewers and editors uphold high standards in their handling of these issues.

The study of the genetics underlying human behavior and social outcomes, with its fraught history and heightened potential for misappropriation, requires rigorous science. The failure to fully reckon with confounding fuels misinterpretation of genetics research and impedes scientific progress.

Confounding among families induces biases in genome wide association studies (GWAS). ‘Population structure confounding’ in genomic data relates to correlations between the structure of genetic relatedness in a GWAS sample (exemplified by the orange-to-purple gradient) and the phenotype studied. Here we show genetic sequences from individuals 1−4 at top left, with their attendant phenotypes (height) at top right. For a given genetic variant, individuals with purple alleles will tend to be taller than those with orange alleles, regardless of the variant’s causal effect on height. This confounding affects any variants that reflect this axis of genetic population structure—typically many millions of variants. While researchers often use methods that adjust for population structure in an attempt to avoid spurious associations, the extent of residual confounding in GWAS remains unclear.

(iii) how confounding is obscured or even leveraged in a broader class of contemporary genetic studies. In particular, we discuss where claims about genetic causality may reach beyond the evidence.

(a) Song & Zhang [32] show that the estimated genetic correlation between BSB (a measure of bisexual behaviour) in males and number of children is significantly different from zero (left diagram). They hypothesized the causal structure shown in the right diagram: genetic variants affecting BSB affect the number of children only through their simultaneous effect on risk-taking behaviour. When adjusting for genetic effects on risk-taking behaviour, the partial (residualized) genetic correlation between BSB and number of children is no longer significantly non-zero. They take this observation as evidence for their hypothesis. (b) When we repeat this analysis but replace risk taking with a variety of other measures (‘measure X,’ blue in causal diagram), 16 out of 18 measures we examined yield a partial genetic correlation between BSB and number of children that does not significantly differ from zero (bars show 95% confidence intervals). Therefore, the data are equally consistent with the hypothesis that genetic variants driving BSB are maintained through evolution as a byproduct of selection on any of these 16 measures. We note that most of these measures—like risk-taking behaviour—have a significant partial genetic correlation with number of children. Four of the measures, denoted with an asterisk, are not. (c) Male participants in the study sample who reported having first had sex before the age of 13 (including victims of childhood sexual assault, many of which would have had same-sex perpetrators) are likelier to be classified as BSB by the criteria used in Song & Zhang [32]. N, total number of males in each group.

(ii) a study by Song & Zhang (2024) purporting to solve a paradox: the evolutionary maintenance of genetic variants that increase same-sex sexual behavior, despite being “reproductively disadvantageous.”

Confounding between genetic and non-genetic factors influencing traits. (a) Confounding within families. Non-genetic transmission can parallel genetic transmission and their respective effects are confounded in observational data. Illustrated is a model where a trait value is the sum of an inherited component from parents and random noise. Under this model, the expected resemblance between relatives depends on transmissibility (t2, the portion of trait variation attributable to the transmitted component) and a rate of decay across genealogical distance (the ‘persistence rate,’ b, which increases with increasing degree of assortative mating). Ignoring the confounding of genetic and non-genetic transmission in the data, Clark [3] misassigns all transmission as genetic heritability and all assortative mating to be on a latent ‘social genotype’.

Reanalyses of data from Clark [3] challenge the paper’s claims. (a) Example of confounding between genetic and non-genetic transmission. The relationship between paternal wealth and a measure of social status suggest at least one source of confounding between genetic and non-genetic transmission. Across relative pairs, correlation in occupational status is highly correlated (Pearson’s r = 0.91) with those relatives’ correlation in paternal wealth. Inset shows that individual occupational status is strongly correlated (Pearson’s r = 0.72) with father’s wealth. (b) Pseudoreplication distorted estimates of familial correlations. Familial correlations (95% confidence interval (CI)) in occupational status (1780−1859) using the approach employed by [3] (in gold) involved pervasive, non-uniform pseudoreplication (electronic supplementary material, note S6). For example, the (1780−1859) occupational status correlation for fourth cousins is calculated from 17 382 pairs, derived from only 1878 unique individuals. In teal we show conservative estimates using only a single relative pair per surname (means and 95% CI over 1000 bootstrap samples are plotted for each familial correlation), which are therefore not susceptible to pseudoreplication. Distant cousins show dramatically higher correlations after adjusting for pseudoreplication. (c) Signals of change in social mobility. Parent-offspring correlations in multiple status measures generally decrease over time in Clark’s [3] data, in contrast to claims of stagnant social mobility made in the original paper. To mitigate pseudoreplication, we calculated correlations using one pair from each surname (as in (b)). Shown are average correlations (95% CI) across 500 bootstrap iterations of correlation estimation. The electronic supplementary material, figure S13 shows two complementary analyses estimating correlations either without accounting for pseudoreplication, or using percentile ranks—both result in similar trends.

(i) a study by Clark (2023) suggesting that patterns of similarity in social status between relatives indicate that social status is largely determined by one’s DNA

We detail how multiple modes of confounding are incompletely understood and underappreciated in many contemporary papers. We illustrate these issues through reanalysis of data from case studies and discussion of the literature. We focus on…

Confounding fuels misinterpretation in human genetics | Proceedings of the Royal Society B: Biological Sciences The scientific literature has seen a resurgence of interest in genetic influences on human behaviour and socioeconomic outcomes. Such studies face the central difficulty of distinguishing possible cau...

Full OA paper here: royalsocietypublishing.org/doi/10.1098/...

We're excited to share the final version of our paper, where we demonstrate how confounding remains a thorny problem for claims about causal genetic influences on human behavioral and socioeconomic outcomes. (w/ @jedidiahcarlson.com @oliviarxiv.bsky.social Ruth Shaw @arbelharpak.bsky.social) 🧵👇

incredible opportunity!!

I maybe did bury the lede a bit 😂

I sent the letter above to WSJ as a Letter to the Editor a few days after Dr. Wright's piece came out. They declined to publish it, and declined to comment on why. They did publish other letters.

These facts should be considered when evaluating Dr. Wright’s claims of oppression.

These views are my own and do not represent Cornell University.

To be clear, Dr. Wright was not implicated in any wrongdoing. But the assumption that this association was a non-issue for hiring committees reviewing Dr. Wright's application in 2020 seems implausible.

In early 2020, Pruitt — Wright’s PhD advisor — was accused of scientific fraud and later found by his university to have engaged in multiple instances of data falsification and fabrication. “Spidergate” was the biggest scandal our field has weathered.

A notable omission from Dr. Wright’s commentary is the fact that 95% of his publications at the time of his applications were co-authored with Jonathan Pruitt.

His CV certainly demonstrates potential. But “top in [his] field”? You would be hard pressed to find a domain expert who agrees with that claim.

Most folks spend years applying to dozens or hundreds of faculty jobs, and most PhDs do not end up in faculty positions. There are many potential reasons Dr. Wright failed to land a job.

I personally know two other white male evolutionary biologists hired by Cornell within the last two years.

People unfamiliar with EEB often do not realize how competitive the job market is.

Dr. Wright suggests that being a white male prevented him from securing a faculty position in our field (EEB) and that in particular, Cornell University’s doors were closed to him because of his identity. However, I was hired by Cornell this year.

I am the most recent white male evolutionary biologist hired at Cornell. I’d like to provide essential context for readers mulling over Colin Wright’s commentary, “Cornell University discriminated against me.” 🧵

I say we start by making it less financially appealing.

We need a ruthless anti-scam administration. A total crackdown on robocalls, MLMs, crytpo scams, get rich quick influencers, and fake supplement companies.

Steep penalties and enforcement structures with teeth.

had not seen this; very cool

I need to write a response, as the latest white male evolutionary biologist hired by Cornell!

h/t @jbyoder.org for inspiring!

Part of the Science Homecoming initiative: sciencehomecoming.com @sciencehomecoming.bsky.social

America leads the world in science and innovation. Let’s not throw that away. Published July 2025 in my hometown paper, The Franklin Times. Part of the Science Homecoming initiative. KJ Muldoon was only a few days old when his parents received devastating news: their son was…

Published an op-ed in my hometown paper, advocating for the continued funding of American Science.
johnbenning.net/2025/08/01/a...

Very stoked to tell y'all that I just started as an assistant prof in Cornell EEB!

More to come later, but for now, if you know folks looking to do a PhD in evolutionary ecology, they can chat with me about this awesome fellowship. 2-pg LOI is due July 31!
www.simonsfoundation.org/grant/simons...

They are not. We have maybe the worst political system in the developed world, one that makes an unrepresentative sliver of voters with no principles who aren't really paying attention decisive. Every few months, they get profiled like a set of jangling keys and folks want to throw democracy away.