Older construction on the east coast might be shorter than on the west coast. Also, weirdly carved up apartments can have tight squeezes, even if they started as very tall industrial spaces; and these are probably more common in NY than LA.

Here is a
graph (p 58 aka 59) with fine-grained religiosity categories, matching Otis's numbers. It gives breakdown of the population into these categories on p 60 aka 61. The source by Hleihel is in Hebrew.

Here are some other sources, obsolete in light of the above.

A paper with more recent fertility data (p 10 aka 112). It distinguishes between 7% ultra-orthodox and 72% other Jews. It puts the ultra at 7.5 in 1999 and 6.74 in 2007; and the non-ultra dropped from 2.22 to 2.20 (1%).

If national-religious are 7% of the population and have a TFR of 4, this gives secular 2.0, so the numbers are generally compatible.

Some details: this paper includes as ultra-orthodox those who have lapsed, but have lots of ultra relatives (ie, no military). How many of the national-religious it includes I don't know. In any event, there is a discrepancy between the two papers in the total percentage of Jews.

Here is a paper that distinguishes secular, orthodox, and ultra-orthodox, but I don't see it actually list TFR, only TFR for the richest 10% of each demographic.

Here's a 10 year old paper that lists the secular Israeli TFR as 2.0-2.2 (p 7 aka 446). This excludes ultra-orthodox and national-religious, but isn't as fine-grained as Otis's numbers. Also, I think it might include secular Arabs. Also, it says that this population is 70%.

Reducing government support for children has been going on for 10 years. I think it has brought ultra-orthodox down from 7.5 to 6.5.

Conor Cruise O'Brien's "What Can Become of South Africa?" seems to have been reprinted in a couple of books scanned by google, so I don't think the phrasing is in that article.

fascism for the blacks: let me spell out my sources. It is widely reported that Arthur Miller coined the basic phrase "state socialism for the whites…and fascism for the blacks," probably in 1991 in the Nation. The only place I can find the longer phrase, other than you, is Paul Moorcraft's book African Nemesis. I'm not sure when it was published. Moorcraft says "there is said to be socialism for the Afrikaners, capitalism for the English-speaking whites and Indians, and fascism for the blacks.

What is the history of the quote "Fascism for the blacks, capitalism for the English and Jews, and socialism for the Boers"? In 1991, Arthur Miller said "state socialism for the whites…and fascism for the blacks," but who added the other details? and when? Maybe Moorcraft in 1994?

in 1453, the renamed Istanbul

Actually, it was only officially renamed in the 20s. The Ottoman government wrote Constantinople on the coins (well, actually, Kostantiniyye, translating "city" into Arabic, but leaving Constantine). "Istanbul" is Greek, not Turkish or Arabic, meaning the City (over the centuries the P in polis turned into a B). The Greeks were using it before the conquest. It was a nickname for a millennium before it became official.

Identical twins are sufficiently rare and, moreover, uncertain, that no one has a grasp on racial difference in their rates. I think there were some papers a few years ago claiming certain small populations to have high rates, though.

Forget whether modern tests are comparable to Galton. Are modern tests comparable to each other? NO.

Here is the scatterplot of the tests in the meta-analysis. What is increasing is the range of averages across papers. In 1900, there were 2 tests, both about 200ms. Today there are lots of papers, some with reaction times of 200ms, some with times of 300ms, some in between. And the situation today is hardly different from the situation in 1945, when there were two tests with 200ms and one with 300ms.

So I conclude that the tests in the modern day are not comparable, so it's no surprise that they aren't comparable to the Victorians. Why aren't they comparable? I assume varying researcher methodology. If you think these numbers are comparable across researchers, which you must to do the meta-analysis, then you can compare the modern tests and conclude differences between the populations studied. You must conclude that Australians and Finns are much faster than Americans and Scots. Is that more or less surprising than the claimed result of the paper?

But if you do believe it, you can go test them with uniform methodology. If it turns out that Australians and Finns really are faster than Americans and Scots, you can come back to this paper. If it turns out that they aren't, this paper is nonsense.

Here's an anti-scalping story: Apple has an annual conference. This year, tickets sold out in 2 minutes. Given the trajectory over previous years, this was not surprising, so they have anti-scalping measures. They already had name tags for re-admittance, so it's not much to add a serious id at check-in.

It's a little mysterious to me that if Apple wants a lottery, they don't just hold an explicit lottery, rather than this internet congestion lottery.

To answer my question, I think that the message that Ticketmaster is trying to send by this article is that if you act like Springsteen, you'll share his fate. It takes effort for Ticketmaster to sell your tickets to the people you want to, rather than the scalpers, so why should they bother? The choice is to let the scalpers do the price discrimination or to let Ticketmaster do it, so just go with flow.

oops, I meant to say that Jagger claims that it's impossible to eliminate scalping, right after the article explains how he did it.

Why is this article in the paper? Is Ticketmaster intentionally trying to look like the underdog? If so, why? I see two potential audiences for this act: the public and the musicians. The public to not look like a monopoly and so avoid antitrust action. The musicians to make it look impossible to do better than ticketmaster, to discourage them from negotiating or leaving.

The Gulf News article linked in the comments quotes Mick Jagger on how it's possible to eliminating scalping in the sentence immediately following one in which they explain how they held a scalping-free show (id everyone).

Incidentally, before Live Nation bought Ticketmaster, Live National already ate all their profit because they had a better monopoly.

That’s a good point that the cost is $3000 per family, not per individual. The only quote I could get for single site brca is $600, though.

I don’t think any other of your points are relevant.

Also, I don't think Buckley was a flight instructor during the war, unless flight is a euphemism for sexual hygiene.

For those interested in the stories of Buckley's flights, they are in his essay "Learning to Fly." However, Steve has conflated two different flights. Buckley got lost when sober. During the flight after staying up on speed for a couple of days, he fell asleep.

The 1906 Strachey quote about hundred-year planning is yet another form evidence that they cared about the long run.

Sure, pleiotropy is a plausible explanation for the correlation. Indeed, the greater the number of genes that control IQ and height, the more reason there is to expect pleiotropy. If that’s the only role that load is playing, sure. But the quoted sentence seems to claim load as an alternative explanation for correlation, rather than as evidence for the existence of pleiotropy.

Presumably individuals with a higher mutational load will have lower
intelligence and be shorter, all things equal, because these traits have
extensive genome-wide coverage and are big targets.

I don’t mean to put words in your mouth, but it seems to me that a naive reading of that sentence is incorrect. The error I see is reifying mutational load. If you compare people from different populations with different mutational loads, then, yes, maybe IQ predicts population load predicts height. But if you pull a sample from a single population, the particular mutations are independent. The individual’s height load tells you nothing about the individual’s IQ load (excluding pleiotropy). Inside a single family, the situation might be different yet again because of linkage. But maybe Geoffrey Miller says something more sophisticated to salvage this.

Beware the classification of homicides as stranger vs acquaintance. 40% of homicides are not classified. Moreover, the number of unclassified homicides varies from year to year much more than any other aspect (eg, number), so the time series is probably meaningless.

Does anyone know why the percent of homicides not so classified varies so much?

qualified [probably something like "helped" would be a better translation] him as a papal candidate

I think Der Spiegel is pretty precise with English, but if you think that's a translation error, look at the original: "Bis heute hat Bergoglio nicht nur die argentinische, sondern auch die italienische Staatsangehörigkeit – eine Tatsache, die ihn für die Wahl zum Papst qualifizierte."

Christmas vs Memorial Day: drugs go generic in two steps. For the first 6 months, only one generic version is available and the price doesn't change much. Then any generic is allowed in.

Steve, I think that you're confusing Roswell with some other incident.

The green font is fine on the water, but not on the hill to the left. But I think that the black font was fine in the first place.

No, Bob Arctor is right. If you exclude the bottom 10%, the median of those who remain, the 10-100 precentiles is the the 55th percentile. In an extreme case, if you exclude the bottom 90%, the median of what remains is not the meaningless 90+50=140th percentile, but the 95th percentile.

I found the original Simon Price article in Word. It is quoted in its entirety here and here. Here are the relevant paragraphs:

"The perception that poshos are colonising the charts isn’t an illusion. It’s demonstrable fact. The Official UK Top 40 of the week ending 20th October 1990 contained 21 British acts. Of these, 16 and a half – ranging from PiL to Paul McCartney to Pop Will Eat Itself – went to their local state school as children. Another four could not be verified by forensic Googling, but it‘s safe to assume they did too. Just half of one act was educated privately: the Pet Shop Boys’ Chris Lowe, who attended Blackpool’s independent Arnold School.

Fast-forward two decades, and a very different picture emerges. Lifting the lid on the corresponding week’s chart in 2010 reveals a far more complex can of wriggling worms. Of 17 British acts, two attended top private schools (Battle Abbey boy Taio Cruz and the aforementioned Ms Doolittle) and three went to fee-paying stage schools (BRIT alumni Adele and Katy B, and Italia Conti pupil Pixie Lott). A further two were groups with mixed educational backgrounds: The Saturdays (of whom two emerged via stage school, including another Italia Conti product, and one Surbiton High), and The Wanted (at least one of whom attended Sylvia Young). Three were TV talent show creations, and one unverifiable. Only seven were ordinary comp kids, unassisted by privilege or patronage, and even that figure includes Syco-signed producer Labrinth, who is Simon Cowell’s nephew."

In particular, if you summarize this as saying that 60% are posh, you are counting Adele (and, even more bizarrely "tv talent show creations"). Price does not seem to have used the numbers 1% and 60%.

Steyn's source, that he named but did not link, was Julie Burchill, whose source was a 12/2010 Word magazine article by Simon Price that may never have been on the web. He compared a week in 10/2010 to a week in 10/1990.

Mankiw could have said "economists like my students." If he has as much loyalty to his students as most academics, this is a real cost to him.

In addition to writing Shakespeare, the Earl of Oxford was the son of Elizabeth I.

I was recently talking to a California principal who had retired to minor administrator/teacher on the east coast. She held that isolating middle grades was terrible. I'm not sure I remember correctly, but I don't think she saw so much difference between 1-8 and 7-12, but thought that 6-8 was unadulterated evil (and probably saw 7-9 the same way). Her opinion was so strong that I felt uncomfortable asking for details.

If you think rhythm might be relevant to drawing, why not talk about it?

Rhythm is obviously important to political rhetoric.

Some kind of rhythm is important to prose meant to be read silently, but is it the same skill? I have heard descriptions of pictures that talk about pacing, but this seems likely to be even further afield.

Surely it is New York that is the exception, not Chicago.

For the first few months of the year, the murder rate was up 50% compared to 2011, which is quite mysterious, but for the whole year it is only up 15%.

Actually, I mangled the Los Angeles numbers in a way that effectively removed it from the data, so the curve didn’t change much. Here is the corrected table and the corrected graph.

Here’s the data as a table. Here is a loess, weighted by vote total, which is also the opacity of the scatterplot.

Japanese toilets: forget America, what about France? Are high-tech bidets pushing out low-tech ones? I searched on amazon.fr and it didn’t look to me like it was selling many, but it did produce 3 or 4 ads taking me off of amazon.

It's weird that Tom House seems to say that it was pitchers using steroids in the early 70s. I guess they might want to use them to recover from injuries, but he also says that he was trying to improve his fastball. Or maybe he just means that he talked to other pitchers, but didn't know what batters were doing.

I think you are mistaken about height.

Yes, the Dutch have gained an inch on the Germans and Swedes, but they were already the tallest in Europe before you were born. Discarding the Dutch, the gaps between all the other European countries are surprisingly stable, even as everyone gets taller.

My assessment is based on tables in wikipedia. Does you have better data?

I got it by reading the article Razib linked to.

@5,14: The reason that the school already has more than one CF student is that they are siblings, so there’s no point separating them.

Actually, since most of this is driven by generational differences, not individual change, you don’t need a time series. You just need to know that young people exaggerate attendance less than old people. That could be measured directly, but assessing age of real attendees is harder than just counting.

As more people leave their nominal Christian affiliation, they are no longer lying.

What leads you to believe that this is the correct explanation of the data? The Pew data says that in the past 5 years, the unaffiliated have grown a lot faster than the people who admit to rare attendance. Do you have data over a longer period?

I know there is some direct measure of exaggeration of attendance, but is there a time series showing it decreasing?

I don’t know how contagious most parasites are, but I would expect them to show up as shared environment, not residual unshared. Certainly, exposure to house cats is shared environment.

I am confused by the second paragraph, about lumping together all non-western cultures, which does not seem connected to the rest of the post. My best guess is that the connection is that the subjectivists of the first paragraph implicitly believe that non-western standards are natural and uniform, while western standards are artificial and subjective. Of course, it contradicts their claim in the first paragraph, but lots of people have inconsistent beliefs. Do you endorse this claim? Was that the point? Should I not guess?

What are your “metaphysical opinions”?

15 people claimed to be exactly centrist. Since that was the default setting, they should probably be interpreted as not answering and be excluded. For each axis, another 20 changed that axis, but left the other at 50. I’m uncertain as to whether to exclude them. Anyhow, it doesn’t make a difference to the regression line, R^2, or lowess to exclude them.

One danger of scatterplots is that you can’t see people with identical responses, which happen with default values, extreme values, and, to a lesser extent, round values. One answer is to add noise. Another is transparency, here done to your data.

I’m surprised by the large number of people who put themselves exactly on the diagonal.

For age the values are straightforward. Older people have higher values. There isn’t a strong trend. Similarly with socieconomic index. The magnitudes for the beta are not high because intelligence and education probably is what is really driving the socioeconomic differences.

It appears to me that there is a strong trend for age, though I may be wrong about the meaning of these numbers or this may be quibbling over words.

Are these regression coefficients standardized? I couldn’t find the answer in the documentation, but recoding them seems to show that they are not standardized in the default settings. (If it is possible to switch to standardized coefficients, please tell me how.)

I think it is misleading to say “the magnitudes for the beta are not high,” or otherwise draw attention to the magnitudes of raw coefficients. (Also, “beta” usually means standardized.) The displayed raw coefficient for age is small because the units are logit/year. For most of the categories a sex change is worth an age change of 25 years, excepting homosexuals, where it is only worth 10 years.

As a proxy for strength of these coefficients, we could use statistical significance. In the case of communists, the T-statistic and p-value for age are strong, though not as strong as for degree and wordsum. But in the case of homosexuals, age is stronger than degree, though weaker than wordsum.

You are right to say that degree and wordsum are the most important, but I think you are wrong to discount age. Maybe I’m misunderstanding how much you discount it, but still I think you are wrong to draw attention to the magnitude of coefficients and to call them “beta.”

If 23andme has tested 15x as many people as ancestry.com, they should find 15x as many 4th cousins, 5th cousins, etc. We certainly know that 23andme has sequenced many more people than ancestry.com, though I don’t know the actual ratio.

(I don’t know the criteria either company uses for identifying relatives. I got the numbers 7th & 10th by assuming that you have twice as many nth cousins as n-1-th cousins. And I ignore everyone not on the same generation. The point is that different thresholds could easily produce this situation.)

pconroy, it’s probably just that 23andme only counts 7th cousins as relatives, while ancestry.com counts 10th cousins.

Another possibility stems from the fact that Ancestry.com’s main product is not DNA testing, but software for making family trees. They may be counting as distant relatives people in the trees of those who have been tested.

Steve, you sound so certain when you make these assertions. But previously, you asserted that speeches are after the fact bribes, to encourage current office holders.

If you believe Hillary that she will retire from government after 2012, it doesn't make sense as an on-going bribe.

Finally, what happened in Libya? Was it the work of Hillary, or was she just trying to avoid the appearance of the UK acting alone?

How might it be possible to determine if the general public distinguishes between “genetic,” and “biological,” and “not a choice”?

Should I have said “mechanism” rather than “hypothesis”?

There are lots of explanations of why paternal age could have a correlation. How much weight should we put on the particular mechanism of de novo mutations?

Given an observed correlation between paternal age and a particular condition in the child, say, autism or schizophrenia, how much credence should we put on the hypothesis that de novo mutations are a significant cause?

Actually, that’s my point. If the gap consists of 10 factors that have as simple interventions as breastfeeding, how will we find them? The breastfeeding hypothesis has been floating around for decades, and we don’t know. It should be easy to test, if people really cared, but finding other factors sounds very difficult. How many have been identified that are anywhere near as plausible as breastfeeding?

The near future is going to yield lots of genetic information. There’s little reason to expect better understanding of environmental effects. That’s true in general, not just about IQ.

I don’t actually believe breastfeeding increases IQ, but there is a cluster-randomized (n=36) study showing 5 points.

Yes, I have not exhausted justifications for AA, but my comments apply just as well to those two examples.

When I say “magically different” I don’t mean “magically untreatable,” but different in the reactions they are predicted to elicit. Maybe this is a correct prediction of human psychology, but no one ever justifies it, or even acknowledges it. It’s not just AA, but, to use your other example, tribalism. Why would a genetic basis for an IQ gap lead to more salience of race than the current situation of psychometrics, of a gap that we don’t know how to change? Society deals with it by only sometimes endorsing IQ. Why can’t we have a society that understands the genetics of IQ without endorsing IQ more than it does now?

You seem to be saying that we’ll know how to change all causes, but that genetics is special because it will be the most expensive intervention. But we currently don’t understand how to change any substantial environmental causes that we don’t fix (unless you count breastfeeding, of which I’m skeptical). Why do you expect change in understanding of environmental effects?

As to peer effects, I don’t think your intervention would work. I don’t think that precise intervention has been tried, but if it were so easy, I think we’d see better results from the many interventions that have been tried. KIPP is pretty close to what you suggest, but is selective in the parents and students they admit (and keep). It’s also pretty expensive and its cost is labor, unlikely to fall. I don’t find it hard to imagine a point in the future where it is more expensive than genetic engineering.

Karl Zimmerman@19, your position is extremely common and seems to be based on treating genetic causes as magically different from all other causes. You contrast genetics with breastfeeding, but either way it is fixed in adulthood, which is generally the important thing. The worse problem is that genetics and simple interventions are not the only possibilities; other environmental causes might be hard to identify or hard to change. You do mention peer effects. Is there much of a policy difference between peer effects and genetics?

Even in the case of simple interventions like breastfeeding, I don’t think any justifications given for affirmative action distinguish them from genetics. The three justifications I know of are (1) denial of the reliability of IQ (racial bias, adult stability, utility to employer); (2) equality of outcome; and (3) attribution of the IQ gap to wealth and hope that AA will redistribute wealth, thus IQ and eliminate the gap. (The breastfeeding gap is partly caused by wealth, but if it were well established to have its putative 5 point value, I think the breastfeeding gap would narrow.)

Razib@20, IQ tests could save people from stereotypes, but employers don’t use them. If AA were eliminated, maybe they would become legal, but I don’t think their illegality in the US is the cause of their disuse. Other details of the hypothetical might lead to their use, though.

When reviewers delay papers to scoop them, are they stealing the idea, or are they creating a marginal delay so that an existing paper gets published first?

I redid some of your graphs.

Your post Verbal Intelligence by Demographic consisted of graphs of interpolations of wordsum distributions. The data wasn’t very smooth and the interpolation produced ugly curves. I redid religion and bible as points + loess curves.
Mainly I did it to be a cheerleader for R and its Hadley Wickham dialect. They aren’t any better for the color-blind, a topic discussed on that thread.

Here’s my code:

library(reshape)
library(ggplot2)
wordsum <- read.csv("generalsocialsurvey.csv")

wordsum.temp <- wordsum[c("Score", "Catholic", "Protestant", "Jewish", "No.Religion")]
wordsum.temp <- melt(wordsum.temp, id.vars="Score", variable_name="Religion")
wordsum.religion <- rename(wordsum.temp, c(value="Percent"))
#qplot(Score, Percent, data=wordsum.religion, color=Religion, geom=c("point","smooth"))
qplot(Score, Percent, data=wordsum.religion, color=Religion, geom=c("point","smooth"), se=F)+opts(legend.position="top")
ggsave("wordsum_religion.png",dpi=100,width=5.67,height=5.67)

wordsum.temp <- wordsum[c("Score", "Bible.is.Word.of.God", "Bible.is.Inspired.by.God", "Bible.is.Book.of.Fables")]
wordsum.temp <- melt(wordsum.temp, id.vars="Score", variable_name="Bible")
wordsum.bible <- rename(wordsum.temp, c(value="Percent"))
qplot(Score, Percent, data=wordsum.bible, color=Bible, geom=c("point","smooth"), se=F)+opts(legend.position="top")
ggsave("wordsum_bible.png",dpi=100,width=5.67,height=5.67)

ohwilleke: The percentage of all births that are fourth or higher ordinal births in 1998 was on the order of 7.1%… This is less than a quarter of the GSS number.

Those don’t sound like comparable numbers. If everyone has exactly 4 births, the GSS number will be 100%, but only 25% of births will be 4th births. From the GSS numbers, 100 mothers produce 272 births, of which 55 are 4th or later. 55/272=20%. I don’t see how you get 7.1% out of your link. It looks like 14% to me: (5.7+2.6+1.7+0.6)/71.

Your comment perfectly complements Silver's plan: Beane (or whoever) could follow a well-worn playbook under the cover of "moneyball."

You may be interested in the change over time of the bimodal distribution of first-year lawyer salaries.

“Biglaw” associates used to make somewhat more than ordinary first-year lawyers. Kind of like a medical residency. But in 2006 they made 4x. I found that link a couple of years ago and my memory is that other posts on the blog compared the salary time series with the tuition time series. I think that the total cost of law school has always been about the salary of a first-year biglaw associate. So when the biglaw salary shot up, law school tuition shot up. But why didn’t this rather abrupt change break the linkage between the price of good and bad law schools? There are many gaps between good and bad law schools. Starker than the gap in getting law jobs or passing the bar is the gap in getting a biglaw job, yet those are what set tuition.

One thing that confuses me is why associates’ salaries got bid up, and especially when. The blog post attributes it to the internet boom, which fits the timing. But I’d think these sectors would be competing for rather different people. Wall Street seems like a better candidate. Michael Lewis talks in his book of seeing in 1986 lawyers jump ship for Wall Street (specifically lawyers at firms hired by Wall Street). That makes more sense to me, but seem to predict an earlier shift in salaries. (Maybe the 1991 data, the earliest in the post, shows the trend already in motion.)

The other thing that post says is that all biglaw firm pay the same salary. Even if the partners are making 2x at the 75th percentile as at the 25th percentile, the associates have identical salaries. I suspect this is a similar phenomenon to all law schools having the same tuition, or the fairly narrow spread of private undergrad tuition, though I don’t think I understand any of these phenomena.

TGGP, does Karl Smith define “adult” as 12 years old? or does he not look at the data?

A video of the Sunset Strip Riots. I can't make out the hair very well, but much of it looks as long as the '66 Beatles. At 3:26, the announcer says "You can be sure of seeing a goodly representation of the Now Generation. Long hair and bare feet are sprinkled liberally about the area."

I'm not sure whether that agrees or disagrees with Steve's description. How many longhairs before a crowd stops looking clean-cut?

I found this article very frustrating. It seemed to me to be two good parts pushed together in an unnatural way. I think the need for a conclusion to a column encourages journalists to make stuff up and to amp up their confidence. It is better when you just trail off, as you usually do in Taki's. (or is that just the movie reviews?)

The topics of the origin of hippies and the importance and rate of social change are interesting and important. But what does the existence of proto-hippies have to do with the explosion? By the time people are pointing at punks and predicting their explosion, they already exist. Maybe the long history of proto-hippies means that the culture has evolved to be good at accommodating new people, compared to the new punks, but if that's what you mean, come out and say it.

Yes, German nature-worship is an important predecessor to the hippies, but there are other sources, like the beatniks. In particular, the German proto-hippies were anti-drug, right? eden ahbez was still anti-drug, but otherwise sounds a lot more like the hippies than his predecessors in Germany.

Yes, but it seems dishonest to me to bold all those instances of "student" but not the the two instances of "he" in the ninth paragraph. Let the actual frequency speak.

The change in language on the second page is interesting.

The fact that we are this smart and not smarter means that smarter people are not as "fit" – worse overall at life and reproduction in the ancestral environment.

But what, exactly is the cost? Some candidates: (1) more calories; (2) more genes to keep from mutating; (3) longer maturation period. Under those scenarios, intelligence need not detract from any particular activity to reduce fitness. Or maybe it does detract from some activities.

Thanks!

I, for one, don't mind if you write about topics you're not an expert on, especially if you say so and list your sources.

If you think, as you said at MR, that Hayes is secretly talking about Russian Jews, why don't you write something about German vs Russian Jews?

That sounds like a non sequitur.

Maybe Paris encouraged gentrification, but lots of major cities are gentrifying with increasing population. Why is Chicago different in population growth?

The slang usage of bread meaning money was popular in the 40s. Dough goes back to the 19th century.

the european immigrants who arrived earlier were perceived as inferior and marginally white

Did people really frame ethnic superiority in terms being white?

I agree with Peter Ellis. Yet another way of presenting the data would to make the bins powers of 2. If you took the current graph and aggregated 2-3 and 4-7, that would be pretty close to this. I think the 4-7 bucket would be smaller than the 8-15 bucket, which you could then call bimodality, but I think it would be a pretty small jump. I haven’t looked at other people’s data, so my expectations are hardly justified, but I am surprised at how smooth the curve is and how the regulars don’t stick out.

In any event, the mode of people who visit 4+ times per year are hardly regulars. They’re probably getting here via google, too. They just ask google a lot of the type of question that brings them here.

What is the normal rate of crossover? It looks to me like you have a rate of about once per chromosome. Is that correct? Or is it an artifact of the limited data? But the mother appears to have twice that rate. Is that the normal variation between instances of meiosis? between people?

What are you using to draw the curves? cubic spline?

The ones where there is little data (Jews, atheists) jump out by not being unimodal. In some ways, that’s a good feature of the graph, warning not to take too much from that particular line. But the Jewish data is actually unimodal on the high end and the non-monotone curve is the result of the smoothing method. I guess that’s a sign that the data is really weird, due to small numbers, even if it isn’t so weird as to not be unimodal.

It seems to me that it is a big jump from Sikh and Bollywood imagery to the assassination. Don't all of his films have Sikh imagery?

It doesn't seem like a crazy hypothesis, but virtually everything claimed as evidence supports the simpler hypothesis that he just used Sikh fairy tales to spice up a western fairy tale. Tyler claims that the Queen is made up and dressed differently, more like Gandhi, in the final scene. If so, that draws attention to her death. But I'd have to see what she looks like.

One use of the list price is that if you're uninsured and poor and convince the hospital to write off the bill, they can count it as charitable work.

A better version of that link to HBD Chick on Todd.

Has there been any attempt to construct a racial tree purely based on genetics?

My understanding is that Cavalli-Sforza did not do this, but took the leaves of the tree as given and described how they coalesce, and how fast. This gives some reality checks, eg, if two populations coalesce very early, maybe the distinction was artificial, but it fails to discover unexpected substructure.

Cavalli-Sforza did offer a bit of this, such as PCA plots, which are objective representations of genetic data. A human can look at such a plot and divide it into clusters without being affected by the labels; in practice this usually reaffirms previous beliefs. I’d prefer a computer do the clustering, but my bigger complaint with this is that it is only one level, rather than trying to construct the whole tree.

Another problem with Cavalli-Sforza is that he assumed a tree structure. But I don’t think anyone has good algorithms for more complicated structure, even today. The admixture algorithm you like to run shows that people are working on it, but it doesn’t sound ready for prime time, let alone combination with divergence in time.

Finally, I should note that sampling procedures introduce some bias. For example, if you assume that particular population is homogeneous and don’t sample it much, you can’t discover substructure there. But this problem should probably wait for all the above to be addressed.

Spreadsheets:

There were 3 or 4 really fancy spreadsheets, going further in exactly the direction you're talking about. They were all for the NeXT operating system, which no one used. Weirdly, this includes a Lotus product, which another commenter mentions above. Of course, since no one used NeXT and it was hard to port programs from it, so they all died. Except that the current Mac OS, starting around 2000, is essentially NeXT, so they could have easily reached a large audience there, but it was too late or something.

My theory of spreadsheets is that their purpose is to be sufficiently opaque and jury-rigged that the user can obtain whatever result he wants. Anything that reduces bugginess, like the feature you want, makes them less useful in the common use case.

I am disappointed by their treatment of breastfeeding. The effect goes away when you control for mother's IQ. That's all you need to know.

Maybe there's enough doubt that more study is useful, but the provisional conclusion should be that it doesn't matter.

Kramer's study is not "large"; it has n=31. Also, it cherry-picked different IQ measures. I might be confusing it with a different paper, but Cipla sounds like worthless data-mining.

Americans think that the 60s started on that day, but the French know better.

I did mess up that last step. But first: “24 of which are common” by which I meant “IN common.” Also, I should have 76/4=19, not 24.

So there are 19 rare variants that the cousins are both carriers for. For each variant, 1/4 of the embryos will be homozygous, so we have to apply -log(3/4) = 0.4 bits of selection pressure to avoid that particular homozygosity. If we have 6 bits of selection pressure, we can avoid 15 homozygosities, leaving 4 uncontrolled, of which we expect 1. We can probably do a little better by not choosing the 15 loci ahead of time.

Also, if the shared grandparent comes from a cousin marriage culture, the grandparent will be inbred and have novel homozygosities that one might like to screen for, though these will not be fatal or probably even crippling, since they have already been observed.

Could you elaborate on cousin marriage? Does cousin marriage produce diseases common enough that they can specifically be screened for? I would expect it to produce unique diseases.

Actually, you can screen for the unique diseases. If the grandparent has 304 recent mutations [or two grandparents each have 152], then the grandchildren each have 76, 24 of which are common, so an embryo has an expected to have 6 homozygosities, which 64 embryos are enough to eliminate. [I think I’m overestimating the selection power needed here.]

So it does seem like the power is comparable to the effects of a single cousin marriage, though I’m not sure I’m correctly using the figure of 150-300 that I think I got from Leroi.

Intrade agrees that the presidential election is close. In fact, it currently puts Obama > 50%.

mihirchander@13 made the very interesting claim that Gleick reports the subscores on the IQ test and they were all the same. However, google books says that Gleick says practically nothing.

A similar hypothesis is that he cherry-picked the lowest score for the purpose of discrediting IQ testing.

The theory that the conspiracy is 100% post hoc fails to explain the claim that one of the celebrants entered the room at the same time the phone was stolen. It could have been opportunistic looting, but looting a crime scene seems to me like a bad idea. And maybe we should not trust EJE on the importance of the phone, or even the evidence that it was stolen.

I think the simplest theory is that there was a conspiracy to bug his phone, which may have played a post hoc role with Diallo.

Edward Jay Epstein himself has a page with hotel videos (on youtube)

just the dance:

http://www.youtube.com/watch?v=QAJH2VBb0pY

The article seems badly organized to me, I suspect to avoid drawing conclusions and being sued for libel.

There are three suspicious thing here: the stolen phone; the celebration; Diallo lying about the other room.

The stolen phone seems pretty strong evidence that the hotel staff was spying on DSK, probably about the Carlton scandal. The same people leaped on the Diallo opportunity. The celebration could just be that they did a good job leaping on it, without it being a set-up.

Diallo lying about the other room is not news and doesn't seem like much to me. But if the staff were already working against DSK, as suggested by the phone, that makes it much more plausible that they conspired with Diallo ahead of time.

It is hard to extract technical details from NYRB articles. Removing the battery from the phone would explain all the details, except for the "forensic expert" claim that it's a big deal. The claim implies that the GPS was disabled while the rest of the phone continued to work, which would be very suspicious. But that would still allow crude location tracking by cell tower, so what's the point?

How did he get Sheehan's phone records?

Here’s a hypothesis for the difference: pictures. The quote even says that he knew what her older brother looked like. That implies that he would have seen a picture of her even through ordinary gossip, but I suspect facebook inundated him with pictures of her. It made her like a celebrity.

Is this any different than for height?

Silicon Valley: In Steve Hsu’s comment thread, Yan Shen quotes the book “The Chinese in Silicon Valley.” It says that of the top 150 companies of SV, 10% had Asian CEOs (including Indians) in 1999. But the percentage of high tech firms in general is much higher and much more Chinese than Indian. I find implausible the claim of 20k Chinese engineers and 2-3k Chinese high tech CEOs.

The book has a lot of statistics, but they are badly organized, alternating lumping and splitting, absolute numbers and percentages, different years, etc. I don’t trust that the author correctly describes them, but his sources might be useful.

Steve Hsu @41, I think it is important to distinguish between unconscious discrimination and quotas. They could both be true, but unconscious discrimination doesn’t seem relevant once there is a quota. The claim in that quote is that Stanford was surprised to discover that it was holding Asians to a higher standard. That seems to require it not to have a quota. Of course, there are complicated possibilities, such as some admissions officers being unconscious of a quota enacted by others.

Bigger houses, especially when mandated by developers and / or zoning,

Developers and zoning are completely different. Zoning usually mandates large lots, not large houses. This produces artificial scarcity to drive up the price and keep the poor out. Developers want large houses, close together. In CA, the cost of the house is small compared to the cost of the land. And, as you always say, a large house is cheap if you're willing to put lots of people in it.

While the WSJ doesn’t mention the fast decline of IE, its graph agrees with yours, though sandwiching FF makes it hard to read.

I’ve seen a lot of these graphs and they are usually quite sensitive to the particular website. I’m surprised your data agrees so well with StatCounter.

Your experience at the dealership is an example of salesmen not profiling. Why did they apply these tactics to a gringo? Do they not have enough white customers to bother learning other techniques?