Even though simple logic said “there will be blood”, there were nevertheless long stretches [of WWI] when all was quiet on the Western front. Why? Axelrod said this was due to the power of tit for tat, which we discussed in Chapter 5. German and French soldiers tacitly created their own unwritten peace treaties: if you don’t shoot at our side, we won’t shoot at yours. And if the higher-ups do force us to fire our artillery, we’ll intentionally aim short of your position, or aim long of it. These rules were never written down, and they were rarely even spoken as far as we can tell – but they left their mark. Take this example. One day the German artillery fired toward the British side without doing any damage. A German infantryman climbed up on a parapet just to deliver an apology to the British: “We are very sorry about that; we hope no one was hurt. It’s not our fault, it’s that damned Prussian artillery” The German infantrymen didn’t want to destroy the fragile truce they had created with their alleged enemies. Military higher-ups hated this tacit cooperation across enemy lines. Fortunately for the officers (but not for the enlisted men) there was a simple solution: move troops around. By swapping one division south two kilometers and another north two kilometers, military officers could turn a repeated prisoner’s dilemma into a literal one-shot game.
This seems almost unbelievable, and absent knowledge of the primary source I can’t be 100% sure, but I guess it fits together with what you always hear about the soldiers celebrating Christmas together and so on.
Some skeptics dismiss IQ tests as just measuring whether you’re good at staring at a piece of paper, coming up with an answer, and writing it down. But the comprehensive IQ test used most often today, the Wechsler mentioned earlier – involves little paper-staring and almost no pencils. The person giving the test (a psychologist or other testing expert) asks you why the seasons change or asks you to recite a list of numbers that she reads out to you. You answer verbally. Later you are handed some wooden puzzle blocks and you try to assemble them into something meaningful.
I hadn’t realized this. There’s a lot of talk about IQ being related to “test-taking skills”, and it’s easy to imagine that people who are used to take paper-and-pencil multiple choice exams might be more comfortable with them, but that’s not what a lot of real IQ tests are and we should stop imagining them that way. This also makes theories about education and the Flynn effect more impressive.
The images are simple shapes like a regular C or a backwards one, and your job is to note, for instance, whether the C is open to the left or the right. So at what point would you no longer be able to do better than random at correctly answering “left” or “right”? When the image flashes for just an eighth of a second? A 32nd? A 128th? That’s the key variable the researcher keeps track of in this study…can it possibly be the case that people who only need to see the image for a tiny fraction of a second tend to have higher average IQ scores than people who need to see it for an eighth or sixteenth of a second? Summarizing “dozens of studies” run on “four continents” dating back to the 1970s, psychologist Ian Deary says, “the overall answer is yes, there is a moderate association between how good people are at the inspection time test and how well they score on intelligence tests.”
…speaking of IQ not just being about test-taking. Jones also goes through some studies on reaction time and reverse digit span, not to mention literal brain size as measured by MRI.
IQ tests do about the same as the best kind of job interviews – structured job interviews in which the interviewer carefully designs the questions beforehand and sticks to the same ones with each candidate – and better than most of the methods people use to choose employees…IQ tests are even better at predicting outcomes when the job requires higher skills. Back in the 1960s, the Bell Telephone System gave its entry level management trainees an IQ-type test along with a number of personality tests…when, after two decades, the company looked back to see which tests did the best job of predicting which trainees eventually rose the highest in the company hierarchy, the IQ-type test did the best job, beating out the personality tests. Looking across many studies of IQ in the elite workforce, one review says “general cognitive ability is the best single predictor of executive/professional-level performance, just as it is of performance in the middle to high end range of the general workforce.”
What I get out of this is “apparently people should structure their job interviews more”.
One might contend that it’s harder to measure social or emotional intelligence than it is to measure more conventional intelligence. But psychologists have tried: they’ve checked to see whether people with more social or emotional intelligence tend to have higher IQs, and so far it looks like they do. The relationship often isn’t as strong as the relationship between, say, a person’s vocabulary test scores and their Raven matrices scores, but the results are clear: IQ scores predict practical social skills.
On the other hand, all of these “measures of social intelligence” are things like “can you read this person’s face to determine what emotion they are feeling?”; I wonder how well that correlates with tasks like “work your way up to becoming coolest person in your high school class” or “win the heart of your love interest”. There is a George Washington Social Intelligence Test, but it doesn’t seem to actually involve becoming the universally beloved father of your country.
But what did the apparently more cautious, more careful Wicherts report? He said that the average IQ in sub-Saharan Africa was about 82 – corresponding to the 12th percentile in the United Kingdom…that’s an improvement from [Lynn’s estimate of] 70, and it’s an improvement that arose partly because Wicherts chose to throw out samples of students who came from families with nutrition problems and low socioeconomic status. The Wicherts average of 82 only includes samples of apparently healthy students from families that have typical socioeconomic status. And in a region of the world with as much poverty and disease as sub-Saharan Africa, that decision is quite likely to leave out substantial portions of the population. To further test the data and get the best average, Wicherts wrote a separate paper that looked at onl the best test samples…Wicherts’ best samples of students have an average IQ score of 76. That’s at the 5th percentile within the United Kingdom.
I keep hearing wildly different numbers for sub-Saharan African IQ, but 76 seems pretty plausible. Note that if the higher 82 number which I keep seeing around is true, it would throw a wrench in a lot of things. Remember that African-Americans are at about 85, so if going from Africa to America only gains you at most 3 IQ points, everything we know about the Flynn Effect falls apart. Even 76 -> 85 is a kind of low Flynn Effect estimate, but I’ll allow it if we speculate that these Africans probably have been at least a little Flynnified, and maybe African Americans are still impoverished enough not to have been fully Flynnified.
Even in rural Pakistan, higher Raven’s IQ predicts higher wages. One might think that thousands of miles away from the Western universities where the tsts were designed, abstract IQ tests would have no power to predict which workers earned more and which earned less – but an IQ test made up of boxes and lines and circles had a modest ability to predict a person’s wages across rural Pakistan, just like in the United States.
Also relevant to IQ tests and culture.
Country-level results go back decades. As early as the 1960s, studies in Taiwan and Hong Kong found average IQ scores slightly above the European average. This happened at a time when these countries’ economies were growing fast, but were still poor by US and UK standards. Any simple story that “wealth causes IQ” has to account for the puzzlingly high average scores found in Taiwan and Hong Kong decades ago…a healthy environment helps to boost IQ, but it can’t be the whole story. The high average IQs of East Asia and Singapore have yet to be fully explained.
See, this is another thing that confuses me about the Flynn Effect. Taiwan etc are still slightly above the European average. So either they had fully caught up to Europe in Flynnification by 1960 (how?!) or Asians have some sort of magic anti-Flynn-effect armor. Didn’t Ron Unz speculate something like that once?
The key evidence comes from the horrific experience of famine in the Netherlands during World War II. Some Dutch towns were cut off from regular food supplies toward the end of the war – daily calories were down to about one third of recommended levels at one point – and the children in gestation during that period had low birth weight. But when these children grew up, their IQs were nearly identical to those of children born a few years earlier or a few years later: massive in utero food deficiencies had no long-term effect on child IQ…famines lasting a few months might not matter for a child’s brain development. But what about chronic malnutrition among the young?…Again the experimental method offers an answer. In one study of Guatemalan villagers, some malnourished children where given a protein supplement and some were given a nonprotein supplement, and then their cognitive skills were measured in few different ways. Of the two interventions, the protein supplement gave a bigger boost to student scores.
I was expecting something awful but on first glance the paper actually looks really good. I’ve uploaded it here if you want to see. They don’t measure the results in IQ so I’m not sure exactly what the magnitude is. It only worked on people who were already poor and probably not well-nourished. Possibly another argument for vegetarians taking protein supplements?
So far it appears that schooling clearly boosts crystallized intelligence, but the typical efect of school on fluid intelligence may be small or even nonexistent…[but] here are two examples of apparently successful increases in fluid intelligence. First, in a study in Sudan, a few months of training on the abacus appeared to boost Raven’s matrix scores dramatically. Second, a study of Israeli schoolchildren, comparing students of similar ages who were born just before or just after the school’s age cutoff: because students born before the cutoff get an extra year of formal schooling, the age cutoff let the researchers identify the effect of an extra year of schooling on student IQ scores. The researchers found a bigger effect on vocabulary scores than on matrix scores, but nevertheless matrix scores indeed rose for students who had received an extra year of schooling.
Right now I am pretty confused on what affects fluid vs. crystalline intelligence. Also, really want to know if the school year results hold up 10 or 20 years later.
In the case of classroom learning, the evidence is mixed on whether your child’s standardized test scores will likely rise if she’s in a class room of high achievers or tend to fall if she’s placed in a classroom of weaker students. This is one question that’s been tested and retested in numerous ways in classrooms around the world, and perhaps the best way to summarize the vast literature is to say that some signs point to positive peer effects and some point to no peer effects. Amid the ambiguous findings, the clearest peer effect is that disruptive students hurt learning.
I could have told you that in third grade.
Sub-Saharan Africa was the last major region of the world to eliminate lead from its gasoline, a goal that apparently was reached in 2006.
That is exciting and will hopefully cause some progress for the region.
Did your neighbor win the lottery and buy a nicer car? That means that you’re more likely to buy a nicer car too. At least that’s what economist Peter Kuhn and his coauthors found when studying the results of lotteries in the Netherlands: people who won big lottery prizes tended to buy nicer cars (no surprise), but people who lived near the winners tended to buy nicer cars too. Buying is a social activity: most of us try at least a little to fit in with our neighbors.
This section was titled “Staying Frugal Like The Joneses”. Stop tooting your own horn, Garrett.
One study had twins play a repeated prisoner’s dilemma game against each other for a hundred rounds, and they found that higher-IQ pairs of twins did cooperate more often than lower-IQ pairs of twins.
I feel like making twins play the prisoners’ dilemma against each other is the sort of thing where you risk accidentally making the machinery of the universe divide by zero and explode.
Here’s the most exciting result from my experiments with Omar and Jaap: on average, over the course of the entire experiment, higher-IQ pairs were five times more cooperative than higher-IQ individuals. The link between IQ and cooperation was an emergent phenomenon: it arose not from smart individual players but from smart pairs of players.
There was a similar result in a dataset taken by looking at all the prisoners’ dilemmas done in different colleges’ psych programs and such, then sorting them by the SAT score of the college. This is a scarier lesson about WEIRD samples than anything I’ve read explicitly on that subject.
So far, two studies in the United Kingdom find that higher [IQ] individuals are more likely to vote, regardless of other things known about the person such as his social class, his education, and some personality traits…however, a study in the United States drawing on three different surveys finds no substantial evidence that IQ predicts voting behavior.
Thanks a lot, Get Out The Vote people.
A team of Notre Dame political scientists ran a series of deliberative polling studies to investigate these questions. They started by surveying participants’ politics attitudes before they were put into a room with other participants. Once they were in the room they were asked to discuss a prescribed set of topics such as the then-ongoing war in Iraq. After the discussion, they were surveyed again. Participants were randomly put into different rooms, some with more Iraq war supporters, some with more opponents, and so on. That made it possible to check whether their post-discussion opinion moved toward the pre-discussin opinion of the average person in their room: it was possible to see if they were conforming. The researchers ran 330 groups of about 10 people each, so if conformity was even a modestly strong force it should have shown up. Here’s what they conclude: “After several years of experimentation, we have found little evidence that group composition influences postdiscussion attitude. This persistent finding, which holds across a range of experimental variations and for subgroups defined by political knowledge, suggests that the expectations of the Asch literature do not apply to the Deliberative Poll setting”
I’m not sure how surprised to be about this. I guess it makes sense that political issues like the Iraq War are so polarizing that you can’t change people’s minds about them with a short discussion. It’s still surprising to find something without an Asch conformity effect at all.
Psychologist Christopher Chabris and his coauthors looked at team efforts another way: they checked to see if there was a “da Vinci effect” for teams, a tendency for teams that did well on one kind of task such as a team game of checkers to do well on another task such as taking an IQ test as a team. Indeed, they found such an efect, which they called a c facor, presumably as homage to the general or g factor of human intelligence. And what predicted which teams did better overall? The strongest two factors were whether team members routinely took turns speaking, and how well team members did on a test of reading the emotions of others…but nevertheless, the IQ of the average group member and the IQ of the highest-scoring group member were weak but still notable predictors of group intelligence, and differences in individual IQ explained over one third of differences in actual team performance.
I guess we should care a lot more about people taking turns speaking. Also, I bet that last statistic is basically entirely dependent on the IQ range in the study.
And I’ll take this space at the bottom here to point out that author Garrett Jones very kindly replied to some of my criticisms in the review here. I reply to his reply here, followed by a long conversation with Pseudoerasmus that continues here. I am not quite to the point of sticking this on the Mistakes page, especially since Arnold Kling had the same concern even before me, but I admit it some of this is more complicated than I originally expected and I might end up going either way on it.