tag:blogger.com,1999:blog-1812431336777691886.post6932450951738346749..comments2024-02-29T03:57:00.088-05:00Comments on The Mermaid's Tale: Statistical Reform.....or Safe-harbor Treadmill Science?Anne Buchananhttp://www.blogger.com/profile/09212151396672651221noreply@blogger.comBlogger4125tag:blogger.com,1999:blog-1812431336777691886.post-36606182083973789872016-04-06T13:10:10.663-04:002016-04-06T13:10:10.663-04:00'Additive' is either a statistical term o...'Additive' is either a statistical term of estimation or a word about biochemical mechanism. The first is obviously 'true' if you fit an additive model:you'll get effect size estimates whether or not the second is true. In fact, 'true' is misleading because there is in principle no way to show all the multi-way mechanisms, nor to show that two things are perfectly additive (that is, interaction term(s) exactly equal to zero). The very definition is a reductio ad absurdum, since without genes there is no brain at all, and some genetic factors must have approximately additive effects. <br /><br />The reason there is so much pressure to see additivity and ignore non-additivity is that without additivity, complex traits cannot be predicted with knowable precision for a new sample, based on retrospective analysis of samples. One might ask whether genes with strong causative effects on traits like autism and many others, that clearly affect 'IQ", are additive or show dominance, etc; I doubt that would be the case. More simple statistically 'additive' effects are easier to show when one is analyzing a plethora of simultaneous, individually small, context-dependent effects only among 'normal' individuals, and when non-genetic factors predominate and are themselves hardly identifiable, much less measurable.<br /><br />And then there is the sociology of this whole area. Just beneath the covers of assertions of genetic predictability of intelligence lie various forms of eugenics, elitism, and/or racism, that is, the search for 'scientific' justification for manipulative or discriminatory social and resource policy. There is a very long and clear history of this, so such objectives are a fair a priori assumption when these issues arise and someone is so insistent that intelligence simply must be genetic, and tractably and predictably so, and environments etc are conveniently ignored or dismissed. Ken Weisshttps://www.blogger.com/profile/02049713123559138421noreply@blogger.comtag:blogger.com,1999:blog-1812431336777691886.post-54488659083932141512016-04-06T12:35:23.717-04:002016-04-06T12:35:23.717-04:00So to be specific, do you believe that when we tal...So to be specific, do you believe that when we talk about the heritability of IQ that it is not justified to say that the heritability is a factor of genes with independent and additive effects? If there were evidence that would convince you of that, what would that be? Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-1812431336777691886.post-16140282174206703802016-04-06T08:38:11.745-04:002016-04-06T08:38:11.745-04:00You are entitled to your view, but just as a brief...You are entitled to your view, but just as a brief response to this sort of comment, genetic variation does affect our traits, all our traits, though always in a context-dependent way (and that includes genomic and external 'environment'). As typical, when many genome regions contribute their effects are approximately additive in a statistical sense. How genetic mechanisms can actually be truly additive is a separate biochemical question. The apparent additivity of results is partly due to the way that analysis is done, such as the statistical modeling used, mainly for convenience because interactions are exceedingly difficult to parameterize convincingly much less to demonstrate statistically with available samples and under realistic assumptions. It is in a sense for the satisfaction of vested interests that some authors stress additivity, because prediction would not be very efficacious otherwise, because this relates to the degree to which retrospective sample-dependent additivity justifies extrapolation of the same results into the future; the record is pretty full and clear about that. Our post here, and those in the past, are about those and other issues, and each person will have his/her opinion about the statistical analysis enterprise as currently practiced. Readers have to judge for themselves what they think about this area.<br /><br />Now, you've made your point (repeatedly), and we've published it. And that's enough.<br />Ken Weisshttps://www.blogger.com/profile/02049713123559138421noreply@blogger.comtag:blogger.com,1999:blog-1812431336777691886.post-91967342878812537582016-04-05T12:02:05.872-04:002016-04-05T12:02:05.872-04:00I want to bring your attention to this paper:
htt...I want to bring your attention to this paper:<br /><br />http://www.nature.com/mp/journal/vaop/ncurrent/pdf/mp201645a.pdf<br /><br />In this GWAS it is finally proven that there do exist genes with additive and independent effects on phenotypes of mental traits like intelligence, memory, educational attainment, and more.<br /><br />In this blog there has been criticism of the idea that there can be additive and independent SNPs on these traits, but here the SNPs have been replicated in independent samples and did shew that the heritability of many phenotype is instantiated by P = G + E model.Anonymousnoreply@blogger.com