tag:blogger.com,1999:blog-9430835.post5766314376767269495..comments2024-03-28T16:22:14.888-07:00Comments on Steve Sailer: iSteve: Regression toward the mean and IQUnknownnoreply@blogger.comBlogger213125tag:blogger.com,1999:blog-9430835.post-7538376884241195122016-12-09T05:21:40.947-08:002016-12-09T05:21:40.947-08:00I am just not familiar with the scientific researc...I am just not familiar with the scientific research you are implying when you say that two parents with an IQ of 160 will produce a child with a lower IQ. It just doesn't make sense to me, and goes against my observations. And then you have the news headlines of the past couple of years that say intelligence is a hundred per cent inherited from the mother, not the father, which I don't believe either. I think intelligence, like other heritable qualities, can skip generations, you get throwbacks to many generations back.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-9430835.post-37846816320739101782014-12-07T20:02:19.479-08:002014-12-07T20:02:19.479-08:00Well I think it is like this. Children's IQ...Well I think it is like this. Children's IQ's regress to the mean. But having a child with a high IQ, is a higher probability with two high IQ parents, because they have a greater chance of hitting that standard deviation.<br /><br />Like two very tall parents. Not every kid they have will be that tall. They will probably be taller than average, but shorter than the parents. But SOME of their kids, MAY be that tall, or taller.<br /><br />And if this tall kid meets a very tall wife later, SOME of HIS kids will be that tall. But not all of them. And on average, their heights will all go down more towards average<br /><br />So I think as a class, higher class people are more likely to have some very smart kids included in there, but the effect can be very uneven. Most of the high class kids will not be that smart. And the average intelligence in each family goes down slowly to average with time.<br /><br />But on the other hand, the IQ of the total population may go up too. Like height.<br /><br />But I think there is a lot of variability, so don't count on your kids being that smart.<br /><br /><br /><br /><br />Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-9430835.post-61957285612664647472013-07-18T05:21:25.650-07:002013-07-18T05:21:25.650-07:00Does this mean we should expect the South Asian an...Does this mean we should expect the South Asian and Southeast Asian populations within the USA, Canada, Singapore, Australia, and other Western countries to continue to drop as time goes by?<br /><br />Right now, Indians, Filipinos, Vietnamese etc. tend to do really well and are net contributors to their new countries in almost every way. Should we expect them to go the way of the South/SE Asian populations at home throughout the generations?<br /><br />Is there any data on this? And if so, how long before they end up falling below the White mean IQ?<br /><br />ThanksAnonymoushttps://www.blogger.com/profile/11770467029102952870noreply@blogger.comtag:blogger.com,1999:blog-9430835.post-89657489894431887482013-02-15T22:23:18.466-08:002013-02-15T22:23:18.466-08:00I come from an Ashkenazi family, very intelligent ...I come from an Ashkenazi family, very intelligent and creative on both sides. Just an anecdote obviously but we have seen no regression over 4 generations in the United States. In the 3rd and 4th generation we have those that surpass the very first highly intelligent generation. The family is increasing in wealth and not from inheritance. I have told my children - and yes this is admittedly terribly racist - if they want smart children, they will marry Ashkenazi Jews. It seems to work. Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-9430835.post-8132964357220080272013-02-02T19:40:17.556-08:002013-02-02T19:40:17.556-08:00Nobody has any idea which genes are responsible fo...<i>Nobody has any idea which genes are responsible for intelligence. Why are you pretending otherwise? And absent such knowledge you cannot speak about what is or is not plausible within a certain time frame.</i><br /><br />But most scientists in the field currently think that the hereditary component of intelligence is the product of many genes of small effect. Under this model, in order to achieve a significantly higher average population intelligence, many genes would have to be involved. That increases the likelihood that the process extends farther back in time, but it is admittedly very circumstantial evidence.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-9430835.post-90010027076351730372013-02-02T12:22:41.935-08:002013-02-02T12:22:41.935-08:00"And since I know that my own ancestors of th...<em>"And since I know that my own ancestors of the 1860's generation came from completely different European countries and not from the same "very small rural community", I can say with near perfect certainty that none of them were third cousins to one another."</em><br /><br />So you not only know all 64 of your ancestors at six generations back, but you know enough about them to be certain that none were third cousins? That's some impressive genealogical knowledge. My grandmother's been studying our family for quite a few years and tracing it back to various parts of Europe, but I doubt she's come anywhere close to that. I didn't know anyone had. I'd think most people would have gaps in a family tree that tall where the records simply didn't exist.Cail Corishevhttp://cailcorishev.wordpress.com/noreply@blogger.comtag:blogger.com,1999:blog-9430835.post-30645703708387879982013-02-02T10:01:34.496-08:002013-02-02T10:01:34.496-08:00The norm changed in Europe but in very small rural...<i>The norm changed in Europe but in very small rural communities - which is where 90% of Europeans were 10 generations ago - even if you're trying to marry as exogamously as possible you're still all going to end up cousins - just not as close cousins as you would otherwise.</i> <br /><br /><br />Yeah, at the level of "forty-second cousins" it is probable that virtually all Englishmen share ancestors, as do virtually all Germans and so so. At a sufficiently large value of "n" all human beings are nth-cousins. <br /><br />But if some hypothetical man and woman who became husband and wife in 1763 were third cousins, and are also in my own family tree, that does not mean that I do not have 1024 distinct ancestors ten generations back. The third cousins would have to marry each other six (or fewer) generations ago from my standpoint. That is, they would have to have married in the 1860's. (Because the common ancestor of third cousins is four generations prior to those third cousins)<br /><br />And since I know that my own ancestors of the 1860's generation came from completely different European countries and <i>not</i> from the same "very small rural community", I can say with near perfect certainty that none of them were third cousins to one another. QED, I have 1024 ancestors ten generations ago<br /><br />Your general principle is correct (there obviously was a great deal of cousin marriage going on for most of human history) but your practical application of it is off.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-9430835.post-88358293556670354852013-02-01T22:52:12.783-08:002013-02-01T22:52:12.783-08:00This sort of calculation is temptingly simple, but...This sort of calculation is temptingly simple, but questionable. The truth is, we don't know *how* intelligence is heritable, only that it is (to a greater or lesser extent: this is still debatable). For example, it's entirely possible that some very intelligent people shouldn't breed with each other, as their cluster of characteristics wouldn't be compatible. It is almost certainly true that some degree of inbreeding might be required. That's how they do it with dog breeds. It might turn out that, say, a very high IQ Frenchman and a very high IQ Chinese woman will produce ... a perfectly average, or even below average set of children, with high probability. <br />On the other hand, a high-IQ man with a hot tomato might actually produce very high IQ offspring, with high probability. Lord knows I've met a few of those among the old elites, and also among my circle of /com and hedge fund types. Most of them did not marry women who overtly display high intelligence, yet their kids are doing things like quadratic equations at 8, and C++ at 10. Millions of years worth of evolutionary instincts cause me to abhor the idea of sticking my genetic code into someone that looks like Emmy Noether, despite the fact that she was a mathematical genius. I find that more compelling than bad gaussian models.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-9430835.post-65411579257037594822013-02-01T10:15:19.765-08:002013-02-01T10:15:19.765-08:00Nor do you have any evidence that average Jewish i...<i>Nor do you have any evidence that average Jewish intelligence was low 250 years ago. It is implausible to me that the whole set of mutations responsible for the increase in Jewish intelligence arose within that timeframe.</i> <br /><br /><br />The "whole set"? Nobody has any idea which genes are responsible for intelligence. Why are you pretending otherwise? And absent such knowledge you cannot speak about what is or is not plausible within a certain time frame.<br /><br />Historically speaking it does appear that the prevalence of genius within a particular population does rise (and fall) relatively quickly. Ancient Greece, for example.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-9430835.post-21934863261905527232013-02-01T04:19:39.109-08:002013-02-01T04:19:39.109-08:00"Nor do you have any evidence that average Je..."Nor do you have any evidence that average Jewish intelligence was low 250 years ago. It is implausible to me that the whole set of mutations responsible for the increase in Jewish intelligence arose within that timeframe."<br /><br />There is a second possibility.<br /><br />1) Direct selection for higher IQ during the banking monopoly phase c. 400 to 1300.<br /><br />2) Burial of (1) under build up of genetic load in little rural villages in eastern europe.<br /><br />3) Shedding of the genetic load in the driftback to western europe via adopting the north euro marriage model which unburies the effects from (1).<br />Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-9430835.post-68905981110929369862013-02-01T04:13:16.367-08:002013-02-01T04:13:16.367-08:00"But unless you're from Africa and barrin..."But unless you're from Africa and barring cases of incest in your family, you certainly do have 1024 ancestors from ten generations ago."<br /><br />Most of the world practises or recently practised cousin marriage as a preference so it's not just Africa.<br /><br />The norm changed in Europe but in very small rural communities - which is where 90% of Europeans were 10 generations ago - even if you're trying to marry as exogamously as possible you're still all going to end up cousins - just not as close cousins as you would otherwise.<br /><br />The icelanders have a giant book to keep track of who's related to who for this very reason.<br /><br />So i don't think it's likely that europeans from even the most assiduous non cousin marrying regions have 1024 *distinct* ancestors.<br /><br />They may have far *more* distinct ancestors than people with other histories and that may have interesting consequences but i don't believe 1024 distinct ancestors is likely to be very common.<br />Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-9430835.post-82058290875724502242013-01-31T20:33:01.895-08:002013-01-31T20:33:01.895-08:00Sure, it could. And the average intelligence of gy...<i>Sure, it could. And the average intelligence of gypsies 250 years ago could have been very high. But there's no actual evidence to support either hypothesis.</i><br /><br />Nor do you have any evidence that average Jewish intelligence was low 250 years ago. It is implausible to me that the whole set of mutations responsible for the increase in Jewish intelligence arose within that timeframe.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-9430835.post-8352062200362263842013-01-31T17:07:18.189-08:002013-01-31T17:07:18.189-08:00@ James B. Shearer
All right. This sounds like a...@ James B. Shearer<br /><br />All right. This sounds like a Bayesian problem with assumed distributions. I have not looked at this stuff for a while. I will try to build a simple model and get some approximate probabilities.<br /><br />Neil TempletonAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-9430835.post-11690869091475551612013-01-31T15:47:46.954-08:002013-01-31T15:47:46.954-08:00But among out-bred populations - the typical white...<em>But among out-bred populations - the typical white American has ancestors from several different European counties who were certainly not related to one another - it is likely.</em><br /><br />But even then, when you go back that many generations, and you get into times when people didn't travel much and didn't marry outside their own faith or much beyond the next village, it seems unlikely that anyone would have a full set of 1024 10-back ancestors without <em>any</em> repeats.Cail Corishevhttp://cailcorishev.wordpress.com/noreply@blogger.comtag:blogger.com,1999:blog-9430835.post-38194253261364053562013-01-31T10:08:04.064-08:002013-01-31T10:08:04.064-08:00what some seem to be saying is that if Jefferson h...<i>what some seem to be saying is that if Jefferson had married an equally intelligent woman, and their children had married people of equal intelligence to themselves, and so on down through the generations, they would still inevitably have dipped down to average by today. That is harder to believe.</i> <br /><br /><br />It's a lot more hard to believe that that sequence of events you describe, in which TJ marries an equally intelligent woman, and their children marry people of equal intelligence to themselves, and so on and so on down through the generations, could ever occur in a world of human beings. The hypothesis sits atop an almost liberal lack of understanding of human nature.<br /><br />But set aside the fact that real people do not behave as you would like them to. The other cause of regression to the mean is that, in addition to <i>people</i> not behaving as you would wish, <i>genes</i> do not behave that way either. It's depressing that so few people commenting on this erstwhile "HBD blog" seem aware of the reality of recessive genes.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-9430835.post-42054911815534804102013-01-31T09:56:17.920-08:002013-01-31T09:56:17.920-08:00You're extremely confused. This has nothing to...<i>You're extremely confused. This has nothing to do with father-daughter or aunt-niece couplings.</i> <br /><br /><br />Of course it does. Those are some of the possible ways in which a person might have less than the theoretical number of ancestors. Cousin marriage is another possible way.<br /><br /><br /><i>The chance that you have 1,024 ancestors 10 generations ago approaches zero. </i> <br /><br />The chance that <i>I</i> have 1,024 ancestors 10 generations ago approaches a certainty.<br /><br /><i>The posited 1024 ancestors at x-10 would have 512 children (x-9) who would form 256 mating pairs. If any of these pairs involve siblings or half-siblings (i.e., if they share a parent), the number at x-10 will be less than 1024.</i> <br /><br />Nothing wrong with your math. It's the "if" which is dubious.<br /><br /><br /><i>Having 1024 ancestors 10 generations back is extremely unlikely.</i> <br /><br />Among inbred populations - Jews, for example - it is extremely unlikely. But among out-bred populations - the typical white American has ancestors from several different European counties who were certainly not related to one another - it <i>is</i> likely.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-9430835.post-37433247640255203252013-01-31T09:42:01.844-08:002013-01-31T09:42:01.844-08:00The average intelligence of Ashkenazi Jews 250 yea...<i>The average intelligence of Ashkenazi Jews 250 years ago could have been very high</i> <br /><br /><br />Sure, it <i>could</i>. And the average intelligence of gypsies 250 years ago <i>could</i> have been very high. But there's no actual evidence to support either hypothesis.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-9430835.post-64190724925044497022013-01-31T08:01:12.291-08:002013-01-31T08:01:12.291-08:00No that isn't what it means. The examples are ...<i>No that isn't what it means. The examples are assuming high IQ and low IQ people have the same number of children to keep things simple. But if high IQ people are having more children than low IQ people than the average IQ will rise over time. And if low IQ people are having more children than high IQ people than the average IQ will fall over time.</i><br /><br />Thank you James, I knew someone would come up with a straight answer eventually.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-9430835.post-50761740618974442092013-01-30T20:38:45.342-08:002013-01-30T20:38:45.342-08:00All the evidence we have - which is far from concl...<i>All the evidence we have - which is far from conclusive, I admit - suggests that Ashkenazi Jews of 250 years ago were not of greater than average IQ.</i><br /><br />The average intelligence of Ashkenazi Jews 250 years ago could have been very high, but it would be difficult to estimate because they were so minimally involved in secular affairs at that time. The bulk of the population had transferred to less-developed Eastern Europe over the previous five centuries. Their occupations were restricted, and the community at that time was realing from the effects of the Khmelnitsky massacres followed by the false messiahs Shabtai Zvi and Jacob Frank. There was little internal impetus within the religous communities to join the secular world, and the secular world was not yet welcoming to Jews at that time.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-9430835.post-68403127089381260872013-01-30T18:49:37.701-08:002013-01-30T18:49:37.701-08:00I'm intrigued by what you are selling and woul...<i>I'm intrigued by what you are selling and would like to hear more. So tell me (a) exactly how many ancestors you had ten generations ago and (b) which of them were marrying their own daughters and/or nieces.</i><br /><br />You're extremely confused. This has nothing to do with father-daughter or aunt-niece couplings.<br /><br />You would have fewer than 1,024 ancestors ten generations ago if, for instance, one of your ancestors married a second cousin seven generations ago or if one ancestor married a sixth cousin three generations ago. The chance that you have 1,024 ancestors 10 generations ago approaches zero. <br /><br />The posited 1024 ancestors at x-10 would have 512 children (x-9) who would form 256 mating pairs. If any of these pairs involve siblings or half-siblings (i.e., if they share a parent), the number at x-10 will be less than 1024.<br /><br />The posited 256 ancestors at x-8 form 128 mating pairs. If any of these pairs involve first cousins or half first cousins, or siblings or half siblings (i.e., if they share a grandparent), the number at x-10 will be less than 1024. That there are no first cousin pairings out of 128 is unlikely (less than 28% if we assume first cousin pairings were just 1% of the total).<br /><br />The posited 128 ancestors at x-7 form 64 mating pairs. If any of these pairs involve second cousins or half second cousins or closer relations (i.e., if they share a great grandparent) the number at x-10 will be less than 1024. That there are no such pairings out of 64 is very unlikely (about 14% if we conservatively assume pairings that shared a great grandparent were just 3% of the total).<br /><br />And so on. As you get closer to the present generation, you have fewer ancestors but more and more distant cousins they must steer clear of to avoid pedigree collapse.<br /><br />Having 1024 ancestors 10 generations back is extremely unlikely.<br />ben tillmannoreply@blogger.comtag:blogger.com,1999:blog-9430835.post-19203070766211639062013-01-30T18:22:22.125-08:002013-01-30T18:22:22.125-08:00Regression to the mean implies no evolution then? ...<i> Regression to the mean implies no evolution then? We have the same IQ as some amoeba a zillion years ago. </i><br /><br />No that isn't what it means. The examples are assuming high IQ and low IQ people have the same number of children to keep things simple. But if high IQ people are having more children than low IQ people than the average IQ will rise over time. And if low IQ people are having more children than high IQ people than the average IQ will fall over time.James B. Shearerhttps://www.blogger.com/profile/13452342984383895221noreply@blogger.comtag:blogger.com,1999:blog-9430835.post-70222946704686717202013-01-30T18:18:24.370-08:002013-01-30T18:18:24.370-08:00Neil Templeton said:
You are assuming that the e...Neil Templeton said:<br /><br /><i> You are assuming that the error "y" is normally distributed with mean 0 and s.d. = 5. So the distribution of error is symmetric with mean = 0, yet you argue that it is probable that he erred "high" on his first iteration of the test. That is, given his first result of 120, we should assume that his actual IQ is less than 120 and that therefore his "error" was biased high on the first test. This contradicts your assumption of unbiased error. </i><br /><br />No it doesn't because we have additional information. We know he scored above average on the test and that tells us the random error was more likely positive than negative for him. If he had scored below average than we would know the random error was more likely negative than positive. <br /><br />Suppose all the people we tested had true IQ 100. Then the test would just be measuring the random error and we could tell exactly what it was but the random error would still be unbiased.James B. Shearerhttps://www.blogger.com/profile/13452342984383895221noreply@blogger.comtag:blogger.com,1999:blog-9430835.post-89901267867150586402013-01-30T18:03:01.483-08:002013-01-30T18:03:01.483-08:00@ James B. Shearer:
Given an observed score of 12...@ James B. Shearer:<br /><br />Given an observed score of 120 on the first test, I agree it is more probable that the true mean equals 115 than that it equals 125. However, it is also more probable that the true mean equals 116 than that it equals 115, and it is more probable that the true mean equals 124 than that it equals 125 (ignoring the fact that points technically do not have probability in a continuous distribution).<br /><br />As we home in on 120 from below and from above, the probability that we are capturing the true mean increases. And the difference in probability between above and below decreases.<br /><br />For instance, it is more probable that the true mean lies between 119.98 and 119.99 is greater than the probability that the true mean lies between 120.01 and 120.02, but the difference in probability is very small. And the probability that the true mean lies in the interval 119.995 and 120.005 is greater than for either of the other two intervals.<br /><br />Neil TempletonAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-9430835.post-9788279331213319362013-01-30T17:34:25.660-08:002013-01-30T17:34:25.660-08:00[Thomas] Jefferson has "thousands upon thousa...<em>[Thomas] Jefferson has "thousands upon thousands" of descendants alive today. I've seen the figure 5,000 mentioned, though its hard to know exactly how many there are as some are always dying and new ones being born. But let's go with 5,000. If you were to examine those 5,000 people there is a high probability that their average IQ would be very close to 100 - in spite of the fact that TJ's IQ was well above that.</em><br /><br />True, because most of the time they (including Jefferson himself) would have procreated with less intelligent people over the years. But what some seem to be saying is that if Jefferson had married an equally intelligent woman, and their children had married people of equal intelligence to themselves, and so on down through the generations, they would still inevitably have dipped down to average by today. <em>That</em> is harder to believe.Cail Corishevhttp://cailcorishev.wordpress.com/noreply@blogger.comtag:blogger.com,1999:blog-9430835.post-89568889884260171372013-01-30T17:32:42.057-08:002013-01-30T17:32:42.057-08:00@ James B. Shearer:
I believe the term "regr...@ James B. Shearer:<br /><br />I believe the term "regression to the mean" comes from examination of a simple regression on one predictor variable, where the constant b0 is greater than one, and the slope coefficient b1 is between 0 and 1. <br /><br />For instance, assume that regressing Child IQ on Parent IQ results in the equation: Child IQ = 40 + 0.60 * Parent IQ. Here the slope coefficient is "heritability". So if parent IQ = 160, then expected child IQ = 136, and if parent IQ = 60, then expected child IQ = 76. This is regression to the mean.<br /><br />If one observed the scatterplot of Child IQ (y-axis) on Parent IQ (x-axis), it would look like an elongated cloud of points sloping upward with slope equal to 0.6.<br /><br />However, if one were to observe the scatterplot of points representing our example agent's current IQ score on his previous IQ score, the plot would simply be a cloud of points with no trend, centered at x-120, y-120 (assuming the unknown mean actually equals 120). A simple regression of current score on previous score would give a result very close to: Y = 120 + 0.0 * X. The value of the slope coefficient would be essentially zero, because there is no trend. Again, the expected value of any score is 120, the unknown mean, and the sample mean is an unbiased estimator of this unknown.<br /><br />You are assuming that the error "y" is normally distributed with mean 0 and s.d. = 5. So the distribution of error is symmetric with mean = 0, yet you argue that it is probable that he erred "high" on his first iteration of the test. That is, given his first result of 120, we should assume that his actual IQ is less than 120 and that therefore his "error" was biased high on the first test. This contradicts your assumption of unbiased error.<br /><br />Neil TempletonAnonymousnoreply@blogger.com