Hacker News new | past | comments | ask | show | jobs | submit login
The negative Flynn Effect: A systematic literature review [pdf] (gwern.net)
48 points by gwern on Feb 24, 2017 | hide | past | favorite | 16 comments

> We did, however, find the average maternal age per country (N = 7) to negatively correlate to the level of IQ decline (r = −0.88). Despite the small sample size, the correlation reached significance (p < 0.01). Nevertheless, as the correlation was negative this seems to indicate that among the countries showing a negative Flynn Effect, the ones with the highest average maternal age show smaller declines in IQ.

Anyone else see the problem here? They are comparing the change in a variable over time with the value of another variable at a fixed point in time. That means they haven't canceled out the time variable, resulting in a confound with the change in maternal age. Would it be possible to get their result--a negative correlation between greater maternal age at the end of a time period and a drop in IQ over the whole time period--if, in fact, the correlation between maternal age and lower IQ is positive? Yes.

One way this could happen is if the relationship between maternal age and lower IQ is linear but the rate of change in maternal age over time is nonlinear and decreasing. The latter is quite plausible because fertility already decreases with age in a nonlinear fashion. It's easier for a country's average maternal age to go from 25 to 35 than from 35 to 45 without even taking societal factors into account.

The authors admit that their finding is contrary to other research they reviewed which found maternal age to be positively associated with lower IQs. My guess is that the discrepancy is because of the authors' flawed model.

Edit: typo, more quote

Humanity does not like the lense of science to be turned upon itself and there will be people who are offended, dismissive or try to hold this academic work to an impossibly high standard - much higher standard than they would for a study they find more palatable.

The truth of the matter is that a decrease in the latent genetic 'g' is what we should expect given a basic understanding of evolution and the structure of modern society which removes (and reverses) the evolutionary pressure. Of course 'g' as with any system of caterogisation is problematic, but no more so than other less contraversial concepts we regularly use. So let's not allow ourselves be stripped of concepts we need to reason about things - 1984 style.

Evolution is an unfair system, and all attempts to restore (reproductive) fairness undermine evolution. The obvious compromise is to have fairness in other matters but not reproduction. For example possibly you shouldn't be allowed to have many children if you cannot support the first one(s). Of course others would say the solution genetic engineering (a technological get out of jail free card), thats another debate entirely (ok, not entirely).

Oh my god, it's literally the thesis from Idiocracy.

And the thesis from "The Marching Morons" (1951) [1][2]

[1] Spoilers: https://en.wikipedia.org/wiki/The_Marching_Morons

[2] Original publication: https://archive.org/details/Galaxy_v02n01_1951-04

Makes for some interesting bedtime reading in conjunction with the recent GWAS studies showing declines in polygenic education/intelligence scores.

Do you have any recommended links for those recent studies?

Sure. Here's a short collection:

- "Sexual Selection as a Justification for Sex" http://www.unz.com/gnxp/sexual-selection-as-a-justification-...

- "Holocene selection for variants associated with cognitive ability: Comparing ancient and modern genomes" http://biorxiv.org/content/early/2017/02/20/109678 , Woodley et al 2017

- "Rates and Fitness Consequences of New Mutations in Humans" http://www.genetics.org/content/190/2/295.full , Keightley 2012

- "Parent-of-origin-specific signatures of de novo mutations" https://www.gwern.net/docs/genetics/2016-goldmann.pdf , Goldmann et al 2016

- "Older fathers' children have lower evolutionary fitness across four centuries and in four populations" http://biorxiv.org/content/early/2016/03/08/042788 , Arslan et al 2016 (from "The cost of inbreeding in terms of health" http://www.unz.com/gnxp/the-cost-of-inbreeding-in-terms-of-h... )

- "Childhood Autism and Assortative Mating" http://citeseerx.ist.psu.edu/viewdoc/download?doi= , Golden 2013

- "Heritability, Autism, & Fear of Breeding" http://www.unz.com/gnxp/heritability-autism-fear-of-breeding...

- "Estimating the Inbreeding Depression on Cognitive Behavior: A Population Based Study of Child Cohort" http://www.plosone.org/article/info:doi/10.1371/journal.pone... ; see also "Genetic diversity and intellectual disability" http://blogs.discovermagazine.com/gnxp/2013/07/genetic-diver...

- "Mutation and Human Exceptionalism: Our Future Genetic Load" http://www.genetics.org/content/202/3/869 , Lynch 2016

- "The Biodemography of Fertility: A Review and Future Research Frontiers" https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4577548/ , Mills & Tropf 2015

- Fertility and intelligence https://en.wikipedia.org/wiki/Fertility_and_intelligence

- "Genetic evidence for natural selection in humans in the contemporary United States" http://biorxiv.org/content/early/2016/05/05/037929 , Beauchamp 2016

- "Assortative mating and differential fertility by phenotype and genotype across the 20th century" http://www.pnas.org/content/early/2016/05/25/1523592113.full , Conley et al 2016a (Dysgenics found in the USA, 1920-1955. Appendix http://www.pnas.org/content/suppl/2016/05/25/1523592113.DCSu... ); see also "Changing Polygenic Penetrance on Phenotypes in the 20th Century Among Adults in the US Population" http://www.nature.com/articles/srep30348 , Conley et al 2016b

- in Iceland: decrease in the education polygenic score 1910-1990, "Selection against variants in the genome associated with educational attainment" https://www.gwern.net/docs/genetics/2017-kong.pdf , Kong et al 2017 (graph https://www.gwern.net/images/genetics/2017-kong-iceland-educ... ; media https://www.theguardian.com/science/2017/jan/16/natural-sele... "Natural selection making 'education genes' rarer, says Icelandic study: Researchers say that while the effect corresponds to a small drop in IQ per decade, over centuries the impact could be profound" )

- in the US: decrease in the education polygenic score 1920-1960, "Mortality Selection in a Genetic Sample and Implications for Association Studies" http://biorxiv.org/content/early/2016/04/21/049635 , Domingue et al 2016 (graph https://www.gwern.net/images/genetics/2016-domingue-usa-educ... )

- "Genome-wide analysis identifies 12 loci influencing human reproductive behavior" https://www.gwern.net/docs/genetics/correlation/2016-barban.... , Barban et al 2016 (supplement http://www.nature.com/ng/journal/v48/n12/extref/ng.3698-S2.x... ; genetic correlations with fewer later offspring: _r~g~_ = -0.236 and 0.712 respectively. Cross-sectional confirmation of Conley et al 2016. )

- "How cognitive genetic factors influence fertility outcomes: A mediational SEM analysis" http://biorxiv.org/content/early/2016/08/18/070128 , Woodley et al 2016

- "The negative Flynn Effect: A systematic literature review" https://www.gwern.net/docs/iq/2016-dutton.pdf , Dutton et al 2016

- "Assortative Mating, Class, and Caste" https://www.gwern.net/docs/genetics/2015-harpending.pdf , Harpending & Cochran 2015

An Iceland study by Kong et al 2017 showed a .1 point decline of a polygenic measure of IQ, per decade http://www.pnas.org/content/114/5/E727.full.pdf

This seems small enough that genomic technology in the not-s-distant future could reverse it

They showed a lower bound of .1 points, not a total .1 point decline; the most reasonable extrapolation would be .3 points, so since the 1910 start of their sample, -3.21 points (and possibly more since the phenotypic correlations driving this decline were noted for decades before then in England, but at the moment, while we have ancient genomes from thousands of years ago and many modern genomes from 1910AD on from 80yos+, 1000-1900AD is a bit of a doughnut hole - not so ancient people are willing to sequence without 'consent', but too ancient to ever get consent).

My name is Mike Flynn and this has been a rough few months to have that name...and now "The Negative Flynn Effect"? Perfect.

Why don't you just go by Mike instead of Michael? ...

Oh come on downvoters, haven't you seen Office Space?

I know, don't explain the joke, but for those who haven't seen it (it did come out ~18 years ago, and I'd guess that its popularity/currency peaked ~10 years ago), the relevant scenes: https://www.youtube.com/watch?v=_BaMx_n2_hM

"Richard Lynn is one of the most unapologetic and raw “scientific” racists operating today, arguing, among other things, that nations with high average IQs must subjugate or eliminate lower-IQ groups, which he associates with particular racial groups, in order to preserve their dominance."


I would very strongly advise taking the paper and findings here with an exceptionally large helping of salt.

How about we just read the paper on its own merits?

More import than related people's political ideology, must surely be the methodology used, and whether others can confirm the data and methods used as correct....

Science is not racist or non-racist, it's just science. And the personalities of the people doing the science should havr no bearing on the correctness of the conclusions reached if the proper scientific method is used, plus some peer review and attempts to replicate the findings.

Because contexts matter.

Because reputation matters.

Because ideologically-driven "research" often shows both conscious and unconscious bias. (For a fascinating study of the latter, look up the case of "Clever Hans".) Forewarned, a closer examination of the piece for flaws is highly advised. The narrator has proved unreliable in past.

And because it seems of late the practice of giving the benefit of the doubt seems to have presented itself as problematic in instances.

Yes, it's possible for even the most unreliable of sources to be correct on occasion. That would be an exception to the rule, and still not provide a basis for trust, reputation, or authority.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact