Talk:Heritability/Archive 1

Untitled
Quite a bit of work you've added. I'll look at it later in more detail.


 * Is "selection strength" used in psychology? The normal term in agriculture/biology is "selection differential."

Ted 15:09, 29 January 2006 (UTC)

Hmmm. You're probably right, "selection strength" is a casual phrase used in place of a correct term. I'll check with the usual sources, but my sense is that "strength of selection" is the common biology term I was thinking of. Cheers, Pete.Hurd 17:14, 29 January 2006 (UTC)

I tossed the images around, the first large piece of text mentiones several time twin studies, so haveing one of the twin study images there would be a nice illustration. It also solves the problems later with to many cramped up iamges. --KimvdLinde 23:09, 4 February 2006 (UTC)

Estimation methods -> quantitative genetics?
I do not know how other feel about it, but I think the estimation methods might be better served in the Quantative Genetics article. Reason for that is that the methods are equally usefull for the estimation of genetic correlations, which falls outside the scope of this article? What about moveing those section? --KimvdLinde 23:30, 4 February 2006 (UTC)


 * I think, in their current states, it fits more in Heritability than quantitative genetics, but I see no reason to make it either or at this point. I can't tell where the quantitative genetics article is going, it's possible that the two articles are future merge candidates.  I suggest letting them each grow as seems appropriate and see what subject matter gets duplicated down the line. Pete.Hurd 03:14, 5 February 2006 (UTC)
 * Thea reason I asked is that I am busy with QG and I am going to add estimation stuff. Heritability is just a term based on some aspects of the estimation methods. We use more variances and covariances anyway. But I leave it for the moment as is, and we see when it is growing at QG. --KimvdLinde 03:21, 5 February 2006 (UTC)

I have taken away merge into quantitative genetics. It has been over a month with a vote of 1 yes, 2 no -- hardly an enthusiastic response of any kind. Ted 13:25, 30 March 2006 (UTC)

Question
Hi, I have a question regarding the methods of calculating heritability for Kim. I would like to comfirm that in order to calculate the heritability, the data have to have some relaives structure, like parents and offsprings, siblings, etc. Since, recently, someone told me that these relationship is not neccesary. I am confused. So please help me to clarify it. Thanks. Peter Hanon.


 * I'm definitely not Kimvd, but here is the answer anyway. Yes, you need some data from relatives.  Basically, that is the only way to get a handle on genetic variation vs. environmental variation. Ted 00:43, 1 April 2006 (UTC)

Additive Effect and Dominance Deviation
I had hoped to avoid it, but maybe the concepts of additive effects and dominance deviation needs to be made more clear. I just made the illustration, so it is easily changed if need be. The accompanying text might be something like:
 * The simplest genetic model involves a single locus with two alleles that effect some quantitative phenotype, as shown by + in the figure. We can calculate the linear regression of phenotype on the number of B alleles (0, 1, or 2), which is shown as the Linear Effect line.  For any genotype, the phenotype can then be written as the sum of the overall mean, a linear effect (a), and a dominance deviation (d).  The additive genetic variance is the weighted average of the squares of the additive effects:
 * $$\sigma^2_A = f(bb)a^2_{bb}+f(Bb)a^2_{Bb}+f(BB)a^2_{BB}$$
 * with a similar relationship for variance of dominance deviations.

I'm not sure it would be that useful, although there is something to spelling it all out.

Comments? Ted 15:06, 4 May 2006 (UTC)
 * I have one. To me, {bb,Bb,BB} is a categorical variable. It's a bit heavy-handed to treat it as a continuous predictor, and furthermore it's quite arbitrary to suppose Bb should lie halfway between bb and BB in this respect. For example, it could be that Bb is more hearty than both bb and BB. Btyner 14:49, 29 June 2006 (UTC)


 * Thanks for the comment. There is a difference between Bb being halfway between bb and BB and yBb being halfway between ybb and yBB.  In fact, the diagram shows a situation where B is completely dominant (or, nearly so).  The X-axis is really the number of B alleles.  In this way, bb=0, Bb=1, and BB=2, which is the way the additive effect is defined.  I'll see if I can come up with a way to make that more clear. TedTalk/Contributions 15:22, 29 June 2006 (UTC)
 * The model for the additive effect is: yij = ai + aj + dij + error, not the usual statistical model. TedTalk/Contributions 15:25, 29 June 2006 (UTC)

So... perhaps the introduction can be clearer
Perhaps I am just daft, but does a high heritability correlate to significant or insignificant genetic factors (as opposed to environmental factors)? A clear and plain English statement in the article's introductary paragraph might be helpful to casual "link browsers" like myself. BigNate37T·C 06:38, 17 July 2006 (UTC)


 * That question is unanswerable, in part due to the plasticity of heritability itself. TedTalk/Contributions 11:55, 17 July 2006 (UTC)


 * Uhh... I don't think so... from the article, "In genetics, heritability is the proportion of phenotypic variation in a population that is attributable to genetic variation among individuals." Now, my question could be restated to ask "Does this mean hertiability is proportional to phenotypic variation in a population that is attributable to genetic variation among individuals, and does this imply that heritability is inversly proportional to phenotypic variation in a population that is attributable to environmental variation among individuals?" Now, since I asked this question I've read enough from other articles to tell me that this statement is more of less true, so I'm really only asking whether we can't make the introductory paragraph more meaningful to the lay person. BigNate37T·C 14:36, 17 July 2006 (UTC)


 * My apologies. I think I see what you meant to ask.  In the simplest model, H2=VG/(VG+VE).  The answer to your question is no.  The relationships are not proportional.  However, as VG increases and/or VE decreases, then H2 increases while as VG decreases and/or VE increases, the opposite happens.  Is that what you meant? TedTalk/Contributions 21:22, 17 July 2006 (UTC)


 * Er, perhaps if I knew what VG and VE were. But I think barring that confusion, you've answered my question. BigNate37T·C 21:40, 17 July 2006 (UTC)


 * I think I get it... if this is correct, then I'm satisfied I know what I wanted to know about heritability:
 * as environmental variation's influence increases, heritability approaches zero;
 * as genetic variation's influence increases, heritability approaches one:
 * Thus, in the limiting cases where environmental or genetic variation are the sole factor, heritability approaches zero or one (respectively)? BigNate37T·C 21:46, 17 July 2006 (UTC)

request for comments
On race and intelligence, please Slrubenstein   |  Talk 13:16, 31 January 2007 (UTC)

Neutrality (?)
Is this statement neutral? It seems like the writer has already made the decision that intelligence doesn't have high heritability, regardless of the study: "Much the same goes for intelligence tests. The conclusions from studies involving intelligence tests often conclude that intelligence has high heritability. This is probably due to inherent problems with human twin studies, as well as reflecting a high level of genetic variation for many human traits, and corresponding lower environmental variation within the confines of the test." I can almost hear "Once those are taken into account surely the heritability of intelligence will be shown to be lower than these studies mistakenly conclude." —Preceding unsigned comment added by 69.123.249.121 (talk • contribs)


 * Actually, the definition of high heritability is a large genetic variance compared to a relatively smaller environmental variance. How would you change it?  Possibly, a more expanded version would be better, but this is an article on heritability, not intelligence studies.  Heritability is quite plastic for just about any quantitative trait, which also means it is fairly useless for how some (many?  most?) people would like to use it. Ted 02:41, 22 April 2006 (UTC)

I agree that this wording is highly biased and borders on propaganda. The author of this statement should explain the "inherent problems with human twin studies." Further, the statement about high levels of genetic variation is vague and needs to be explained. If this particular article deals with heritability and not the heritability of intelligence, and therefore it is not appropriate to explain these statements in this article, then these statements should not be included in this article to begin. Finally, to state that anything is "probably" true, without justification, in a serious forum, is scientifically irresponsible. Cheers, BJH 00:37, 29 September 2007 (UTC)

Regarding this sentence: "In the case of scholastic ability, how well one does in the final school exams depends on both what and how well one was taught, how hard one has studied, how ‘naturally’ smart one is and, of course, on a fair bit on luck." The use of the words "of course" is a direct admission of bias on the part of the author. In science, nothing is assumed true, but rather must be proven. If the author believes that luck plays any role at all in final exam scores, then that hypothesis must be proven experimentally. If that hypothesis has already been supported/proven by data, then references should be given. Further, the hypothesis that these four factors are the four determinants of final exam scores also is debatable and without references, purely theoretical. This type of undocumented bias does not belong in a serious discussion of heritability. Cheers, BJH 00:37, 29 September 2007 (UTC)

Due to a lack of references and proper explanation, I have deleted these sentences from the article. Please feel free to contact me if you would like to discuss this issue further. Cheers, BJH 21:05, 30 September 2007 (UTC)

Zero heritability
Hello most of the time while working on growth traits of synthetic populations, i come across zero heritability estimates for body weights. Traditionally these traits have been defined as the highly heritable traits. My estimates may be due to loss of variability due to continuous selection within a synthetic breed, or may be due to reaching plateau for the trait. There is no scope for introducing variability too. What reason can probably define this phenomenon? 117.196.225.217 (talk) 18:27, 14 June 2009 (UTC)

Source for heritability from twin studies image?
As usual now, when I see an image on Wikipedia, I look at the full image file (which by best practice should be hosted on Wikimedia Commons) to see if there is any source for the image cited. I don't see any source cited for the twin study image for this article, indeed I see a notation on the image file metadata that the image comes "(from unknown source)." On that basis, I am about to delete the image, which disagrees with some sources that I have at hand and hope to add to the article soon. Any sourced image on the issue available to another editor would be much appreciated. -- WeijiBaikeBianji (talk) 01:23, 29 September 2010 (UTC)

Introductory pictures
The titles of the hair color pictures are very imprecise. There is variance in any trait, such that 'a crowd with variance in hair color' is tautological. Indicating 'high variance' would precipitate the question as compared to what? The second picture states an example of low heritability. In case of no variation in the phenotype heritability would be undefined. Indeed for hair colour, heritability was shown to be very high. I would suggest removing the pictures Sboehringer (talk) 07:56, 29 September 2010 (UTC)


 * The pictures are included in a commendable effort to rewrite this article to make it more clear. I am trying a different effort at a rewite (and have dozens of sources at hand) that may do the same sort of clean-up with different illustrations. Heritability is a nonintuitive concept, so I'm trying to check the best sources with the best explanations to add further improvements to the good work already done by other editors here. -- WeijiBaikeBianji (talk) 17:16, 29 September 2010 (UTC)

Addition of a general explanation
I once, while browsing, required only a general understanding of heritability but found no such simple explanation on wikipedia. As such, I would like to add some of these general points about Heritability and correcting misunderstandings, and would love for community input as far as clarifying the ideas and adding sources! Here's what I wanted to add a while ago:

One idea that is key to understanding heritability is that both nature (genotypes) AND nurture (environment) help sculpt the traits (phenotypes) of organisms. That is, if genes are a sort of general building blue-print, then proteins from the environment provide the materials and influence the process of construction; to ask whether a specific trait is the result of either nature OR nurture is incoherent.

The ideas behind heritability:

Organisms in a population often differ from one another in certain phenotypes. One thing we can do is figure out how much variation there is for that phenotype. Another thing we can do is figure out how much of the variation is being caused by differences in genotypes, versus how much is due to differences found in the environment. This is Heritability. It is important to understand that we analyse ONE population of individuals for ONE phenotype (take height, for example): Heritability is not a fixed characteristic of a trait (it varies with environment & across populations). Consider the following:

If all phenotypes (height) are the same, even though we know genes play a large role in height, there is no heritability, since there is no variation to explain! If there IS variation in phenotypes, and the environments were the same, any existing variation must be due to difference in genes. Heritability would be 1.0! If there IS variation in phenotypes, and all genes were the same, it follows that any variation is due to difference in environment. Heritability would be 0.0! Heritability is NOT identifying the source of differences BETWEEN populations. The use of heritability instead helps predict whether a trait will be influenced by evolution or changes in the environment. Tesseract2 (talk) 02:16, 25 December 2009 (UTC)


 * I have been digging into reliable secondary sources on genetics checked out from a medical library for a while, looking for succinct explanations that fit current science. I'll try doing an article edit in a while. -- WeijiBaikeBianji (talk) 17:53, 24 August 2010 (UTC)


 * I'm getting close to catching up with this to-do. I'm broadly surprised at how few articles in the genetics category on Wikipedia are sourced with reliable sources for medical articles, which is a problem I have been repeatedly visiting a medical library to fix. -- WeijiBaikeBianji (talk) 20:16, 29 August 2010 (UTC)


 * Such efforts, of course, make wikipedia a more powerful tool for millions (billions some day, if not already)! I thank you.-Tesseract2 (talk) 18:48, 31 August 2010 (UTC)


 * I'm close now to having a couple dozen sources arrayed in my office with detailed explanations of heritability in a variety of contexts. When I can next devote time to off-line writing, I'll digest and cite all the sources in a new block of article text in Wikipedia mark-up that I can then insert into this article and into related articles. Thanks to all the editors who have been keeping an eye on this article meanwhile. Feel free to suggest sources about genetics for the source list on that subject shared with all Wikipedians in my user space. -- WeijiBaikeBianji (talk) 15:14, 15 October 2010 (UTC)

Figure
I don't find the figure in the Definition section particularly pretty. and it doesn't fit well on Firefox 3.0.18 with KDE 3.5.4-25.el5_4.1 (on a redhat linux). 212.126.224.100 (talk) 19:21, 15 October 2010 (UTC)

Criticisms?
It is my understanding that heritability estimates have been highly criticised, even to the extent that they have been called useless. Shouldn't we have some discussion of the criticisms of heritability estimates, especially regarding genetic causation, as I understand it a high heritability doesn't indicate genetic causaion, or even necessarily a predominant genetic contribution to the phenotype. Alun (talk) 13:12, 1 July 2008 (UTC)

Under twin studies: Re the comment: "The classical twin study has been severely criticized and is used less and less frequently nowadays." Says who? As a psychologist I could name dozens of ongoing twin studies, including some with many thousands of participants. This comment smacks of bias. Let's have some justification or else delete this unreferenced assertion. T0mpr1c3 (talk) 00:07, 31 July 2008 (UTC)


 * Says me.... Not a WP:RS I fear, you're right. I have revised the comment, have a look if you like it better now. Given the extensive improvements made to the "criticisms" section since I put in that comment, it probably also can be deleted without much problem. BTW, with "classical twin study" I meant those studies where one simply looks at MZ/DZ correlations and then calculates h2 from the difference. I don't think anyone still does that. --Crusio (talk) 10:42, 31 July 2008 (UTC)


 * Looks good to me, thanks, I can get off my high horse now. T0mpr1c3 (talk) 08:46, 7 August 2008 (UTC)
 * Still tells us nothing useful though, the criticisms of heritability are still absent from the article. Heritability is a quite useless concept when applied to humans, it tells us nothing about genetic causation. Alun (talk) 11:47, 7 August 2008 (UTC)

I have added a section that obeys WP:NPOV, particularly WP:DUE. Tijfo098 (talk) 15:52, 25 October 2010 (UTC)

Source for definition?
I'm wondering what sources editors are referring to for the definition of the term that this article is about. Some sources I have at hand criticize other sources for incomplete definitions. -- WeijiBaikeBianji (talk, how I edit) 23:38, 4 November 2010 (UTC)

Estimation methods: bad example?
I don't understand what the example of unrelated army recruits is doing here. Heritability measures the proportion of variation due to genetics, and nothing in this example enables you to do this. If you were working with pairs of brothers (better yet, trios of brothers) you could do this. I think a better example is needed. Bar fly high (talk) 17:07, 29 February 2008 (UTC)

agree this example made no sense - no comment for 3 years, so deleted. Tim bates (talk) 21:25, 31 March 2011 (UTC)

Books
"the footnotes above refer the 1st ed. from 1960"

Usually, if a book or paper is cited, the citation refers to the edition/version actually used as a reference not some later version which would be useless for determining the accuracy and context of the actual work cited. Perhaps someone could repair the oversight (or perhaps I mean undersite).154.5.33.29 (talk) 02:53, 16 November 2011 (UTC)

heritability definition
The text under one of the pictures says: "An example of low heritability: a population with genotypes coding for only one hair colour". In my opinion this is solely due to low variance in that particular population, and not due to the heritability of the hair color. I assume someone wanted to explain the heritability as the proportion of genetic variance, but this should be viewed from a broader perspective of the trait, not on so narrow examples. In this case you are saying: left side picture: there is variation in the hair color i.e. hair color is heritable, right side picture: there is NO variation in the hair color i.e. hair color is NOT heritable. Better examples would be e.g. conception rate low heritability (some genetic background, but major influence of the environment), body height highly heritable (some influence of the enviroment, but major genetic background). Gabor 141.244.234.218 (talk) 09:54, 31 January 2012 (UTC)

Estimation based on unrelated individuals in the population?
Anyone think a section should be added that discusses recent methods that use SNPs to estimate heritability? Where similarity of phenotype is predicted by net genetic similarity across all loci in individuals that are not closely related. This recent method has confirmed many of the heritability estimates derived from twin and family based models and is not biased by the assumption of equal environments made by family based studies. The method is discussed here http://gettinggeneticsdone.blogspot.com/2011/08/estimating-trait-heritability-from-gwas.html — Preceding unsigned comment added by 67.190.100.34 (talk) 23:49, 30 March 2012 (UTC)

Random chance
From the article: "Factors including genetics, environment and random chance can all contribute to the variation between individuals in their observable characteristics." However, we cannot conclude that random chance actually contributes to this variation. There is no evidence for randomness, only evidence of specific causation to nonrandom factors. It is true that we do not know all of the factors contributing to this variation, but we can attribute that to pure randomness than we can to the hand of god. I propose the edit: "Factors including genetics, environment, and others can all contribute to the variation between individuals in their observable characteristics." — Preceding unsigned comment added by 76.105.173.201 (talk) 23:28, 24 December 2012 (UTC)

High importance but C class article
I've been a page watcher of this article for a few years, and I'm amazed how slowly this article improves, and indeed how often it seems to regress. Does anyone reading this article talk page have suggestions for more reliable sources on genetics and related issues? It seems high time to dig into the quality secondary sources and work through a top-to-bottom improvement of this article. Let's discuss what a good improvement process would be. -- WeijiBaikeBianji (talk, how I edit) 02:03, 2 June 2013 (UTC)


 * Just to collaboratively let all of you know, I plan to work on revisions of this article in a user sandbox, based on the sources mentioned on this article talk page (and on many other sources) and gradually add updated sections to this article as I have time to do that. Feel free to let me know how I'm doing as the edits proceed, and feel free to suggest aspects of this article that can be updated and how to bring this article up to good article standards of quality, at least. -- WeijiBaikeBianji (talk, how I edit) 16:38, 30 May 2014 (UTC)

Some sources
I've been looking up sources. I'll try to revise the article soon. Meanwhile, feel free to have a look at these sources.






 * This authoritative textbook includes chapters by Arno G. Motulsky, Stylianos E. Antonarakis, Michael R. Speicher, Michael Dean, Jon F. Robinson, Nicholas Katsanis, Andrew G. Clark, Jacques S. Beckmann, Bernhard Horsthemke, David N. Cooper, George P. Patrinos, Alexandre Alcaïs, Laurent Abel, Jean-Laurent Casanova, Stefan Mundlos, Ian Tomlinson, Romulo Martin Brena, Joseph F. Costello, Emmanouil T. Dermitzakis, Alan H. Bittles, Michael Hofreiter, Ross C. Hardison, Sohini Ramachandran, Hua Tang, Ryan N. Gutenkunst, Carlos D. Bustamante, Sophia S. Wang, Terri H. Beaty, Muin J. Khoury, Nicole M. Walley, Paola Nicoletti, David B. Goldstein, Jonathan Flint, Saffron Willis-Owen, David L. Nelson, Thomas D. Bird, Brett S. Abrahams Daniel H. Geschwind, David Goldman, Francesca Ducci, Michael R. Speicher, Markus M. Nöthen, Sven Cichon, Christine Schmael, Marcella Rietschel, Antonio Baldini, Morgan Tucker, Min Han, Ruth Johnson, Ross Cagan, Heidi G. Parker, Elaine A. Ostrander, Siew Hong Lam, Zhiyuan Gong, Tiemo Grimm, Klaus Zerres, Vivian W. Choi, R. Jude Samulski, Ian Wilmut, Jane Taylor, Paul de Sousa, Richard Anderson, Christopher Shaw, David J. Weatherall, Rachel A. Harte, Donna Karolchik, Robert M. Kuhn, W. James Kent, David Haussler, Xosé M. Fernández, Ewan Birney, Roberta A. Pagon, Ada Hamosh, Johan den Dunnen, Helen V. Firth, Donna R. Maglott, Stephen T. Sherry, Michael Feolo, David Cooper, and Peter Stenson.



















Enjoy. -- WeijiBaikeBianji (talk, how I edit) 02:28, 16 December 2013 (UTC)


 * I just updated one of the references above. Use of these sources would help a lot for improving the article. -- WeijiBaikeBianji (talk, how I edit) 15:18, 16 September 2014 (UTC)

What source gives hair color as an example of heritability?
I think the photographs of people with this or that hair color are ill-chosen examples for this article. What source gives hair color as an example of heritability (high or low)? -- WeijiBaikeBianji (Watch my talk, How I edit) 10:47, 17 August 2015 (UTC)
 * There are plenty: https://www.google.com/?q=heritability%20of%20hair%20color. I think these two photos are attempting to show that heritability depends on the amount of variation in the population in question. I think it would make more sense to move them down to "common misunderstandings". The text should start out talking, not about the edge cases, but about the common use-cases of heritability and how it is applied. Klortho (talk) 14:49, 17 August 2015 (UTC)
 * Let me rephrase my question to focus on the point at issue. If a review article or a handbook chapter or a general encyclopedia article on the topic of heritability gives examples of human traits while discussing the meaning of heritability, how often is hair color the first, the second, or even the third example trait mentioned? Can you cite a specific source that fits the secondary source description given in the Wikipedia content guideline on reliable sources that uses hair color as the example trait, rather than (for example) height or some other trait? -- WeijiBaikeBianji (Watch my talk, How I edit) 22:34, 17 August 2015 (UTC)
 * No, I don't know of any. Still, I'd suggest not removing the hair color images before having something else to replace them with. Klortho (talk) 01:25, 18 August 2015 (UTC)
 * Given the dispute here and the myriad of possible examples we should definitely use another example. I'd like to see us use a trait that can demonstrate biological fitness between members of a species in a specific population, such as human intelligence. Hair colour does not demonstrate fitness. Height does though. Etc. ♫ SqueakBox talk contribs 02:54, 18 August 2015 (UTC)
 * Thanks, SqueakBox, for those comments. In view of the massive lack of source support of hair color as a paradigmatic example of heritability (the age of the sources found by the suggested Google searches suggests that they date back to outmoded understandings of heritability), I will chop out the photographs now. Hair color is not a quantitative polygenic trait in the sense in which the heritability estimate was designed to quantify. -- WeijiBaikeBianji (Watch my talk, How I edit) 03:34, 18 August 2015 (UTC)

Distinction between "inherited" and "heritability"
Klortho recently edited the lead, which I reverted so there could be discussion here. Based on an article in Nature Education, Klortho added this: "High heritability of a trait means that, within a population, offspring will tend to resemble their parents with regards to that trait.  Conversely, if a trait has low heritability, that would mean that there would, in general, be a low correlation between parents and offspring with regards to that trait."

That is definition of "inherited" not "heritability."

Heritability refers only to the variance between individuals on the trait. A trait with a Heritability Index of close to 0 (such as whether people are born with two eyes) can, given normal environmental conditions, be entirely "inherited." As noted in the lead itself: "Geoffrey Miller, an evolutionary psychologist, has said "The concept of heritability applies only to traits that differ between individuals. If a trait exists in precisely the same form across all individuals, it may be inherited, but it cannot be heritable."

The distinction between "inherited" not "heritability" is an important one, and this article should get it right. Memills (talk) 16:13, 24 July 2015 (UTC)


 * Hi, sorry, I guess our edits crossed. Give me a minute to respond. Klortho (talk) 16:16, 24 July 2015 (UTC)


 * Danielle Dick's review article in Annual Reviews is a good article to turn to for definition issues like this. I'll put the citation in an edit to this comment once I look it up again. -- WeijiBaikeBianji (talk, how I edit) 16:19, 24 July 2015 (UTC)


 * Here's the reference I added: Estimating Trait Heritability. I know what the definition of heritability is, but the way the article was, was way too technical. It's very important to get these articles more accessible. Also the previous text didn't scan at all: "Heritability is the proportion of observed differences on a phenotypic trait (observable characteristics) among individuals of a population that are due to genetic differences though environment factor and genetic drift can all contribute to the variation between individuals." The second half of that sentence is tacked on with no punctuation. So, I'd submit that what is on the page now is better than it was.


 * I also agree that it's important to get it right. So, a trait that has no variation in the population has a heritability of zero? Or, is it that the concept of heritability does not apply?  If the latter, then the text I wrote still applies to those traits that have variation -- is that right? Klortho (talk) 16:26, 24 July 2015 (UTC)


 * The distinction between "heritability" and "inherited" is extremely tricky, and deserves more elaboration in the subsection titled "common misunderstandings."


 * Heritability, and the Heritibility Index (HI), refers only to the variance between people on a trait. That varies over time and cultures depending on the hetero -vs - homo-geneity of the environment.   If everyone was a clone, the HI would be 0 for all traits.  If we could create the same environmental conditions for everyone, the HI would approach 1 for all traits.  (Note: nothing to do here with "inherited from parents".)  Here is the kicker:  As the environment becomes more homogeneous, the HI goes up (just the opposite of what most folks would intuitively expect).


 * However, research can assess the probability that one will have a trait if their parent(s) had it. For example, the probability of developing schizophrenia is about 12% if one parent had it, and about 40% if both parents did.


 * Probability estimates need to be differentiated from heritability estimates -- they are quite different. Memills (talk) 16:46, 24 July 2015 (UTC)


 * Yes, it's tricky -- I thought I understood it, but I've gotten myself a bit confused. How about if we take out the sentence starting, "Conversely", and add a qualifier to the second sentence. So, "... due to genetic differences in that population. For traits that vary within a population, high heritability means that offspring will tend to resemble their parents with regards to that trait. Other causes of variation...."? I will read more about it this evening, and try to keep working on it then. Klortho (talk) 17:04, 24 July 2015 (UTC)


 * Indeed, very tricky. So much so that the authors of the Nature Education article got it wrong.  They are in good company though -- most people do.
 * Imagine a hypothetical situation where everyone had the same environment. The HI would be 1, but that doesn't mean that those people inherited their traits from their parents any more than if everyone was a clone and the HI was 0.  The HI only has to do with the amount of variability between people due to genetic factors.  It is a statistical measure of dubious value.  I'm not sure how it can be usefully interpreted or practically used.  It doesn't refer to the degree to which one inherited traits from parents.  That is impossible because the interactive effects of nature and nurture cannot be disentangled. So I suggest dropping the "inherited from parents" language.
 * Instead, I suggest that the lead, and the "misunderstandings" section, should include a discussion of "probability estimates" and how that is different from heritability, the HI, and "inherited."  Memills (talk) 18:15, 24 July 2015 (UTC)


 * I took out the language about parents and offspring. I suspect that you might be overstating your case, though. In you first hypothetical, clearly there is never a possibility that environment is constant, even in principle, if you consider that "noise" is part of the environment. If it were, and there were still variation, then that would mean that, indeed, that variation would be because of genetic differences, and in that case, offspring would tend to resemble their parents. Klortho (talk) 18:43, 24 July 2015 (UTC)


 * Thanks! This is how WP should work: constructive collaboration to get it correct. Again, I wonder if we should contrast these terms ("heritability" / "heritable"), and, include a mention of "probability estimates" in the lead.  Probability estimates are useful and readily and correctly interpretable by most folks.   I can make sense of "because you had one parent with schizophrenia, your risk of schizophrenia is about 12%."  I am not sure anyone can make much practical sense of the proportion of variability on a trait due to genetic factors.
 * And, thanks to WeijiBaikeBianji for the references below. It would be a good idea for all of us to peruse these. Memills (talk) 19:34, 24 July 2015 (UTC)


 * Thanks for the interesting source discussion here. I see Memills has commented here on how the Nature article by Wray and Visscher (which is a standard source on this topic) is still subject to misunderstanding by popular readers. The review articles (the one I mentioned above that I would look up),  (mentioned previously on this talk page, and woefully neglected all over Wikipedia),  (a classic review article), and  (specifically written to correct misunderstandings that even scientists have) would all be useful to guide further improvements to this article. -- WeijiBaikeBianji (talk, how I edit) 18:25, 24 July 2015 (UTC)

I don't think that I misunderstood anything, and that the text I added is correct and supported by the references. I added, "High heritability of a trait means that, within a population, offspring will tend to resemble their parents with regards to that trait." This was paraphrased from Wray and Visscher, and I'd like to put it back it.

Memillis, you wrote: 'That is definition of "inherited" not "heritability."', and you give an example trait of having two eyes. But for quanitative traits that have variation in a population (which is implicit in the article, I think, but could be made explicit) the sentence I added is true.

You also wrote: "... the probability of developing schizophrenia is about 12% if one parent had it, and about 40% if both parents did. Probability estimates need to be differentiated from heritability estimates." Yes, and if the heritability of schizophrenia were higher than it is, the probabilities you gave would be higher -- meaning that the offspring will tend to resemble their parents more. The statement itself has nothing to do with this interpretation of heritability.

You also wrote, "the authors of the Nature Education article got it wrong." What, specifically, did they get wrong? I've reread it now, and I can't find anything.

There are three degenerate cases that you've brought up:
 * A trait has no variation, like "has two eyes". In this case, heritability does not apply
 * A trait has variation, and everyone is a clone. In this case, heritability = 0. In this case, parents and offspring would tend not to resemble each other.
 * A trait has variation, and the environment is constant for everyone. Note that this is really impossible, since there are always effects of noise and non-linearity in development (butterfly effect) that would be ascribed to environment. But, if it were, then heritability = 1. In this case, parents would tend to resemble their offspring.

So, I'd like to put my text back in, perhaps with some caveats. Klortho (talk) 19:49, 26 July 2015 (UTC)


 * Thanks for the analysis. This is really quite difficult / confusing stuff.  I would like to read up on some of the references noted above.   But, a few comments in the interim.  You mentioned:
 * * A trait has no variation, like "has two eyes". In this case, heritability does not apply
 * Heritability (or HI) would be zero, but clearly having two eyes is a trait that is inherited from one's parents.
 * * A trait has variation, and everyone is a clone. In this case, heritability = 0. In this case, parents and offspring would tend not to resemble each other.
 * If everyone was a clone parents and offspring would very much tend to resemble each other. In fact, they would be the equivalent of identical twins.
 * * A trait has variation, and the environment is constant for everyone (a hypothetical) ...heritability = 1. In this case, parents would tend to resemble their offspring.
 * They would resemble their parents, but not as much as if they were clones.
 * The problem as I see it is that heritability does not refer to whether parents and children resemble each other -- it refers only to the amount of variability between people (not necessarily kin) in a population sample.
 * Anyway, maybe I am confused. Let's both read up and see what we can find.  Memills (talk) 20:57, 26 July 2015 (UTC)


 * Yes, I intend to keep studying up on this topic, and hopefully we can make a lot of improvements to this article. I think it needs some TLC. One response: you wrote, with regards to the case where everyone is a clone and H^2=0, that parents would tend to resemble their offspring, because they would be the equivalent of MZ twins. But, *for traits that have variation*, they wouldn't tend to resemble each other any more than any other two people picked from the pop., who would also be clones. (If they did, it would be because of shared environment.) Klortho (talk) 23:04, 26 July 2015 (UTC)


 * I think I'll turn to some of the standard textbooks on the topic (I've been collecting quotations from those for a while) to look for consensus wording and examples found in the literature written for readers learning about this subject with little previous background (the kind of readers we would expect be visiting Wikipedia, in the usual case). It will be a useful editing exercise to practice expressing what those kinds of sources express in words we agree upon jointly here. -- WeijiBaikeBianji (talk, how I edit) 02:14, 27 July 2015 (UTC)

I've removed that quote from Geoffrey Miller because I think it is extremely likely to not just confuse a reader, but to cause them to throw up both hands in despair. The principal problem, I think, is that it was put into the lead without context, so it would seem to be referring to "trait" in the usual sense. As Rieger, Michaelis and Green, the old standard textbook, Glossary of Genetics and Cytogenetics explains Bateson's 1907 definition, a character, trait, or phene is "any observable (phenotypic) feature of the developing or fully developed individual ...". The trait would therefore seem to be something like tallness or hair colour, well known to many people to be polygenic, involving at least some non-additive factors. To say that if the trait (e.g., hair colour in mice) doesn't vary in a population then heritability is zero is unhelpful. A quote like that one requires some rather heavy setting up of clear definitions before it has much hope of being understood by a student looking to wikipedia for help. Sminthopsis84 (talk) 19:26, 2 December 2016 (UTC)

Major change to the lead
I've reverted the edit by Everymorning adding a bunch of criticism to the lead. I am very well familiar with the literature, and these seemingly randomly picked articles do not generally reflect opinion in the field. Please explain why you want to add these to the lead. None of them are written by major researchers in the field nor are they published in journals for the field. Deleet (talk) 00:13, 29 June 2018 (UTC)