Talk:Heritability of IQ

I wanted to bring this up, as I am uncertain in its value as a source. WP:SPS suggests that "Self-published sources may be considered reliable if published by an established subject-matter expert, whose work in the relevant field has previously been published by reliable, independent publications" (see also WP:USESPS). I believe that this work by Alexander Gusev is an useful analysis of many of the questions we are addressing in this article . It is definitely a self-published source, as it has not (yet) been published as a book or any journal articles. But it is directly relevant as a secondary source summarizing large swathes of material from molecular genetic research that aren't adequately explained or summarized in other material (or often when such material is explained, it is technical terms that are extremely difficult for the average reader to comprehend). What do others think about using this as a source in this article?

Ping previous discussion participants @PositivelyUncertain @Generalrelative. Katzrockso (talk) 23:08, 12 October 2025 (UTC) Reply

Two thoughts - 1) Gusev definitely qualifies as an expert, but his area of expertise is the genetics of diseases while heritability of IQ appears to be a side interest of his, and 2) he appears to have a very specific point of view on this topic and, when considering the totality of research/thought in the area, we need to consider due weight and not create a false balance. PositivelyUncertain (talk) 23:39, 12 October 2025 (UTC) Reply

His area of expertise as defined by his research output is definitely the genetics of disease, but the broader topic here is quantitative genetics. The facts about quantitative genetics don't change between one application (disease) to another (behavior/psychological phenotypes). I wasn't thinking that he would be cited to say "IQ is not heritable" or one value or another, but as a means of adding valuable information on genetics research that currently isn't included in the article. For example, we might want to talk about how twin studies/other family study designs are affected by particular confounds and that different molecular genetics design may address the question better, and I think his article here is a great source to cite for that kind of content. Moreover, there is a large research literature in statistical/molecular genetics that is relevant to this topic, but we need good secondary source analyses / explanations of it, which this piece acts as a great source for that.

Thanks for your reply and thoughtful comments on the inclusion of this source. Katzrockso (talk) 00:14, 13 October 2025 (UTC) Reply

It would be much easier if this were a properly published source, but all sources are judged in context. What, specifically, would this source be used for in this article? Grayfell (talk) 00:18, 13 October 2025 (UTC) Reply

I was planning on expanding the molecular genetics section of the article; the Gusev piece provides very good analysis of the developments in quantitative statistical genetics. Since many of these studies are highly technical, they are rarely covered in an accessible manner and the quick development of the field means there are scarcely literature reviews or textbooks that include the evolving scientific understanding.

Perhaps also as an additional source alongside others in the heritability section. There's also Kempthorne that goes along with his explanations, so maybe citing this piece in that section is unnecessary.

I do believe he intends on publishing it in some manner when it is done, but he has slowly been adding more to it over time. Katzrockso (talk) 02:35, 13 October 2025 (UTC) Reply

Turns out I didn't end up using the source at all, though it may be helpful as a secondary source to supplement some of the primary sources cited. Katzrockso (talk) 09:35, 15 October 2025 (UTC) Reply

Are the sections on environment even relevant in this article? It seems like a WP:COATRACK of debates that should be covered at the Environment and intelligence article. I do think that broader discussion of the environment (non-shared & shared) variance components are relevant because the heritability of trait is often investigated in these partitioning studies, but the current exposition of the topic is merely to list a bunch of papers that show/don't show some effect and then researchers disputed interpretations of these studies.

I think that we should cut any mention of specific interventions or environmental effects, as that belongs at the Environment and intelligence article (perhaps merging any relevant content over there) and then focus specifically on the research that aims at partitioning the variance components. We can cite larger reviews like the Nisbett et al. piece when we are explaining why the consensus is that environmental impacts exist. Katzrockso (talk) 18:50, 13 October 2025 (UTC) Reply

An alternative may be to merge the two articles. I'd been editing in this topic area for years before I was aware that an article on environment and intelligence even existed (it has far lower readership too: only 541 views in the past month, versus 9,812 for this one). The situation is not quite a WP:POVFORK but it's not exactly helpful to the reader either to present these as two separate articles. Environment and heritability are, after all, two sides of the same coin. Generalrelative (talk) 20:33, 13 October 2025 (UTC) Reply

It depends on what we want the scope of this article to encompass. I think this article should focus on the quantitative / biometrical analysis, rather than any research on specific genetic or environmental causes. To wander off into the specific causes of intelligence would make this article way too lengthy. We can certainly analyze the broader concepts of environment in this article, but a long list of the different types of environmental influences would be just as undue as listing the million different SNPs that have small statistical associations with IQ.

Heritability is to genetics as "shared environment" / "unshared environment" are to "environment" in this sense. I do think that in this classification of Wikipedia articles, there is a missing article we could write on "Genetics and intelligence", so I understand where your desire to merge is coming from.

On a related note, I added a section Heritability of IQ#Methodology that presents relevant background information. Katzrockso (talk) 20:43, 13 October 2025 (UTC) Reply

@Generalrelative That wasn't the best way I could have explained it.

I think the two articles focus on different levels of analysis. Heritability of IQ focuses on analyzing IQ from a more abstract standpoint, with statistical models that partition variance. Environment and intelligence is analyzing specific causes of intelligence. They are at different levels of analysis and should probably both be reworked/renamed. In an ideal world, we could name this article "Quantitative analysis of IQ" because that is the most accurate description of what this article actually focuses on. The first version of the article makes that clear . The Environment and intelligence article should be renamed "causes of intelligence" and expanded to include genetics as well, or we make a separate article on "genetics and intelligence". I have no interest in working on either of those articles right now, as these are all such a big mess that it would take a long time to sort out all the issues. Katzrockso (talk) 20:54, 13 October 2025 (UTC) Reply

This is the structure I can see happening for "Heritability of IQ"

Heritability and caveats - current section, reworked

Methodology - current section

History - covers the history of this debate - Kamin, Lewontin, Eysenck, etc are all relevant. There are several sources we can use for this, but Panofsky is relevant. There are also several books on Charles Murray's The Bell Curve that are very relevant, I can't think of titles off the top of my head though.

Estimates - covers the estimates produced by different study designs (twin, adoption, path analysis, genomic) and their issues.

Influences - covers how different factors may influence the heritability of IQ as estimated in a specific group (SES, Age, etc)

I'm not sure how "between-group heritability" really fits into this article - "between group heritability" is not a well-formed theoretical concept (Lewontin has specifically argued no such concept exists). Maybe we could retitled and rework that section to be about relevance to race? In large part, heritability is not particularly relevant to the race intelligence "debate" (a "question" pushed by WP:FRINGE racists). But it is commonly understood as relevant to that debate, so I think a section is worth it. Katzrockso (talk) 21:01, 13 October 2025 (UTC) Reply

Fair enough. I don't have a lot of time to devote to this either right now, and I like what you added for Methodology. Just keep on doing what you feel like doing, and if I or anyone else have questions or objections, we can discuss them here. And if there is interest in a merge discussion down the line, I will be happy to join it. Cheers, Generalrelative (talk) 21:05, 13 October 2025 (UTC) Reply

Yes, but what I was thinking was moving some of the sections I have under Heritability of IQ#Environment effects to Environment and intelligence and expanding/reworking the sections on "shared environment" and "non-shared environment". I think the section on "maternal environment" can be mostly cut, maybe small bits incorporated into Estimates#Twin and family research since I added Devlin's paper there. And move the Heritability of IQ#Malnutrtion and diseases to Environment and intelligence. Some of the other sections can be reworked/rewritten to be more within the scope of the article. But I'll keep on editing and see what seems to work. Katzrockso (talk) 23:40, 13 October 2025 (UTC) Reply

I have no objections to any of the content in this section Heritability of IQ#Between-group heritability on factual grounds, I believe it all to be accurate. My concern is that much of this content is not within the article's WP:SCOPE - it belongs more at the cited Race and intelligence article. There are much more topical points to be made about the irrelevance of the value of heritability to the "race and intelligence" question - Lewontin's arguments from "The analysis of variance and the analysis of causes", this recent paper from population geneticists , and some other related literature .

I think we should keep the first paragraph that clarifies that hereditarian ideology is not supported by science, as well as the first sentence of he second paragraph, but cut the sentences on g-loadings, adoption studies, admixture, or closing gaps. I think that stuff belongs more on the Race and intelligence article, since it is only barely related to heritability. In its place I propose an expansion of the vacuous "between-group heritability" concept, how it is logically unrelated to within-group heritability and give an explanation as to why heritability does not provide an answer to the "race and intelligence" debate.

Tagging @Generalrelative as a recent commenter and large contributor to this article, and also the person that seems to have added this material from the other page.

I am also hoping to get comments on my recent changes to the article, which I know have dramatically changed the content. I have rewritten sections Heritability of IQ#Shared environment, Heritability of IQ#Heritability and age, Heritability of IQ#Heritability and socioeconomic status, Heritability of IQ#Twin and family research, and added the sections Heritability of IQ#Methodology, Heritability of IQ#Molecular genetics.

I plan on rewriting the remaining sections as well. I was thinking of folding/consolidating the maternal environment section into the section on twin and family research, since it largely just discusses the Devlin et al paper, which is but one model. I was also thinking of reworking the "Influence of parents genes that are not inherited" section into a section on the broader molecular genetic literature on the concept of indirect genetic effects and its many names (indirect genetic effects, dynastic effects, non-transmitted coefficient, interpersonal genetic effect). Katzrockso (talk) 08:15, 28 October 2025 (UTC) Reply

I appreciate the ping. No objections from me to shortening the section to its essentials. Just as long as we make it clear where folks can learn more if that's the reason they came to this article in the first place. Generalrelative (talk) 12:57, 28 October 2025 (UTC) Reply

The part in "Molecular genetics" and/or the "Methodology" about the missing heritability gap for IQ (citing older sources saying molecular estimates are 10-20% vs. traditional 40-80%) feels a bit outdated now.

Recent 2025 papers using big UK Biobank whole-genome sequencing (WGS) data have closed a lot of that gap across complex traits generally; WGS captures about 90% of pedigree-based heritability on average, with "rare" variants explaining around 20-22% of it (Wainschtein et al. - doi :10.1038/s41586-025-09720-6 ).

For cognitive stuff specifically, things like educational attainment (generally considered a close IQ proxy) hit about 0.35 heritability via WGS, and fluid intelligence around 0.32-0.34 after adjustments. There's also a within-sibling analysis (Yengo et al., medRxiv 10.1101/2025.09.17.25336022 ) pointing to an average 30% (narrow-sense) for behavioral traits, hinting some traditional estimates might have a bit of upward bias from confounds (this study did not specifically measure cognitive variables).

That is, given that Wainschtein et al. report WGS captures 88% of pedigree heritability on average, they found rare variants contribute about 22%, and cognitive proxies like educational attainment and fluid intelligence reach around 32-35%. Within-family designs also suggest approximately 30% narrow-sense heritability for behavioral phenotypes. Much of the remaining gap may stem from ultra-rare variants or confounds in older methods.

Something incorporating this info may keep the article more accurate/up to date. Granted, I'm not a scientist by trade. Feedback welcome. Bluelight2000 (talk) 07:13, 24 December 2025 (UTC) Reply

We can add the Wainschtein paper and its estimate, but the whole genome sequencing paper is being misunderstood in much of the discourse. The missing heritability gap they are closing is not between twin studies and their WGS estimates, but between pedigree estimates and WGS estimates. The pedigree h2 estimates are already significantly lower than estimates from twin studies, making the 'missing heritability' problem well still a question. At least the paper doesn't explicitly provide an explanation for the gap between twin studies and other estimates, several commentators online have opined based on it though.

Another problem with WGS estimates is that they are very sensitive to subtle population stratification and they don't have a particular good way of remedying that in their paper. The Wainschtein paper doesn't use a method like sibling analysis that can estimate the parameter without this issue.

Either way, we have to be careful about SYNTH until better secondary sources come out. That's already something I have been worried about and have been planning on reworking some of the material here to be clearer.

@Bluelight2000 Katzrockso (talk) 07:39, 24 December 2025 (UTC) Reply