Tuesday, March 08, 2011

Social Animals

The WSJ reviews David Brooks' new book The Social Animal. I discussed an excerpt which appeared in the New Yorker in this earlier post. David Brooks is in danger of becoming the new Malcolm Gladwell: watch out for sloppy thinking, confirmation bias, and pandering to populist sentiments :-(

WSJ: ... In "The Social Animal," Mr. Brooks avoids the challenge of imitating the master by reviving an older form: the 18th-century didactic narrative, epitomized by Rousseau's "Emile." Rather than dig up true stories, he writes one from scratch, and it's not bad. His main characters are Harold and his wife and business partner, Erica. The plot extends all the way from the day Harold's parents meet to the day Harold dies, yet it takes place in a perpetual present-day America.

The chapters are at once chronological and thematic, tackling an array of topics, like learning (a high-school teacher sparks Harold's intellectual curiosity), social norms (Erica has youthful troubles with authority figures), morality (Erica sleeps with a business leader whom she admires), politics (Harold and Erica work for a presidential candidate), and retirement (they move to Aspen and start a tour business).

[See earlier post for more on Harold, Erica, and their Aspen encounter with the Caltech neuroscientist.]

Mr. Brooks is at his best as a social observer, documenting the changing patterns of contemporary life. He showed that talent in "Bobos in Paradise" (2000)—a modern classic that deconstructed every signifier of millennial upper-class life, from Restoration Hardware catalogs to New York Times wedding announcements. Several passages in "The Social Animal" made me think that Mr. Brooks had been glancing through my own windows or fishing from the banks of my own stream of consciousness.

As in "Bobos," he shows genius in sketching archetypes and coining phrases. Here we learn of the "composure class" (who earn their money "by climbing the meritocratic ladder of success"), "sanctimommies" (mothers who critique their peers' parenting skills) and "extracurricular sluts" (kids who participate in too many organized after-school activities).

... If this story is meant to illustrate a broader point, it must be that "cognitive intelligence" and "emotional intelligence" have an inverse relationship: The brilliant are more likely than the average to be socially awkward. But this is nothing but the stereotype that forms the premise of "The Big Bang Theory" and other pop-culture narratives. In reality, tests of emotional intelligence correlate positively with IQ tests.

But Mr. Brooks makes an even bigger claim: Not only is intelligence not connected to social skill; it's not connected to much of anything. "Once you get past some pretty obvious correlations (smart people make better mathematicians), there is a very loose relationship between IQ and life outcomes." This conclusion misstates the science. To reach it Mr. Brooks has to commit a variety of statistical errors and tiptoe through a minefield of contradictory evidence.

First, he notes that studies selecting a group of high-IQ kids and tracking them over decades have failed to identify future Pulitzer or MacArthur winners. He also cites Mr. Gladwell's "Outliers" to note that American science Nobelists "did not mostly go to Harvard and MIT."

[Never rely on Gladwell in statistical matters. Instead, see here and here.]

The problems here include the tiny sample sizes, the shaky assumption that prize juries (and elite universities) make decisions based only on merit, and the focus on the tails of a distribution (here, the highest extremes of intelligence and academic achievement), which is the method guaranteed to tell you the least about the characteristics that matter across the whole range of human ability. To dismiss IQ testing as invalid because it can't pick out the minuscule minority that will attain world-wide fame is to confuse a positive correlation with a perfect one. Only oracles have perfect records of prophecy, and surely no one desires a world in which IQ tests are that good.

[Using a quick IQ screen at age 17 probably allows you to identify future scientific leaders with 100 times better accuracy than chance, and future business leaders with 5-10 times better accuracy than chance. This follows immediately from (soft) 99+ and 90th percentile IQ thresholds. That is, the fraction of people in each of the two categories that is below the given threshold is very small. See links above.]

Next Mr. Brooks cites a study that found "no correlation between accumulating large wealth and high IQ." But in fact, IQ (measured in high school) was almost the best predictor of wealth (measured in early middle age) in this study's own data set. Its author did show that if you consider only people matched in education, income and other variables, then intelligence is unrelated to wealth. But this is hardly a surprise, since education and income themselves both correlate with IQ. In other words, if you look only at, say, high earners with graduate degrees, you will see that the smartest ones are no richer than the dumbest ones, but you will have learned nothing about the true relationship between intelligence and success.

Similarly, Mr. Brooks describes research showing that, during the dot-com bubble, the average investor in many mutual funds lost money by trading in and out too much, when the investor could have made money by just sitting tight. Such people acted "self-destructively because of their excessive faith in their intelligence," he says. Surely intelligent people can do stupid things. But where is the evidence that intelligence or intellectual arrogance was correlated with bad decisions, let alone caused them? The mutual-fund study said nothing about whether more intelligent investors were more overconfident, or traded more, than less intelligent investors.

In fact, studies that do measure both intelligence and economic outcomes find positive relationships. When you consider that IQ can be measured well in half an hour, these effects are surprisingly large: A random person with above-average intelligence has about a two-thirds chance of being above average in income, compared with a one-third chance for a random person of below-average intelligence.

The research that Mr. Brooks minimizes or ignores does not, of course, prove that intelligence is the only relevant trait for success. A host of "noncognitive" skills, many of which Mr. Brooks explains well, are undoubtedly important. But there is no need to tear down intelligence in order to build up the rest.

Even if differences in intelligence explain 25% of the differences among people in how well they perform at work (a much better estimate than the low-ball 4% cited by Mr. Brooks), there is still three times as much territory left to be mapped out. Surely that's plenty of space for researchers to investigate the role of social acumen, mindset, culture, self-control and much else. A thousand flowers can bloom.

At one point, Mr. Brooks veers into odd territory. Economists and political scientists, he writes, assume that "human beings are pretty much the same. . . . This assumption makes social science a science. If behavior is not governed by immutable laws and regularities," he writes, "then quantitative models become impossible. The discipline loses its predictive value." Perhaps this is an allusion to traditional theories in these fields that neglected differences between people for the sake of mathematical simplicity. But the claim is badly out of date. In fact, human homogeneity is neither an assumption in today's social science nor a requirement of the scientific method. It's ironic that Mr. Brooks neglects the abundant evidence for a key source of differences among people—their genes—which influence their personalities, decisions and achievements.

Mr. Brooks would have also done well to suspend his disbelief when his sources talked about brain research. "When stock traders experience a series of good days, the dopamine released into their brains creates a surge of overconfidence," he writes. He cites only a blog post for this claim. Most of the sources in "The Social Animal" are secondary ones, usually books for a general audience. Without information about the primary sources—like journal articles that report methods and results—it is hard to trust that the science is being communicated and interpreted fairly. In the case of the traders hopped up on dopamine, the story sounds good but the necessary experiments just haven't been done.

In "The Social Animal" Mr. Brooks surveys a stunning amount of research and cleverly connects it to everyday experience. The lessons he draws are often insightful, but they are not reliably correct. Perhaps experiencing his own surges of dopamine and overconfidence, he too often abandons his stance of "epistemological modesty" and instead peddles frothy notions that probably won't last long. But in observing the broader trends of social science—and of contemporary life—he gets a lot right. His own achievement here signals a plateau in the market for social science, not a peak.

6 comments:

Jorge_Videla said...

The reviewer gives no evidence either. Is Brooks supposed to be an expert? Why did the APA claim that only 1/6th of the variance in income is explained by IQ? Could they have been right? That's a correlation of ca. .4. Not that high.

"It's ironic that Mr. Brooks neglects the abundant evidence for a key source of differences among people—their genes—which influence their personalities, decisions and achievements."

Now the reviewer shows his hand. So far as personality traits can be measured their heritability is very modest, with the exception of IQ. So modest that the particular gene phenotype correlation may be nonexistent in other societies.

Even the existence of personality traits is questionable as individuals display very different traits from time to time and place to place. The personality trait is a great example of assuming that because there is word or phrase there is some THING to which that word or phrase refers, a great example of the model treated as if it were the reality.

The reviewer can't think, and can't recognize ideology when he sees it.

Catperson said...

"The reviewer gives no evidence either. Is Brooks supposed to be an expert? Why did the APA claim that only 1/6th of the variance in income is explained by IQ? Could they have been right? That's a correlation of ca. .4. Not that high."

1/6th of the variation is very high. When you consider all the other factors that can influence income (work habits, drive, looks, special talents, luck, health, family connections, ambition, personality, charisma) and when you consider how much income can change, how unpredictable it is, and how hard it is to reliably report, it's amazing that 16% of the variation in income can be traced to IQ. No other single trait that we know of seems to be as important as IQ.

Jorge_Videla said...

"No other single trait that we know of..."

Who is "we"?

Botti said...

***So far as personality traits can be measured their heritability is very modest***

Some interesting research on this:

"If the findings are confirmed by other scientists, they suggest that children are not only born with a given personality type, but they develop anatomically different brains as a result of being that sort of person. It raises the prospect of being able to test a young child's future personality by viewing the anatomy of their brain with a hospital scanner."

http://www.nzherald.co.nz/health/news/article.cfm?c_id=204&objectid=10566320

Jorge_Videla said...

There is a link, but as far as I could tell there was no indication of how strong the link was. Look at the data for identical twins.

Some people will speak of a very modest link as the author of this article does, that is as if it weren't merely a correlation but an absolute. I have tried to take personality tests and found the questions totally impossible to answer. They are 100% bull shit.

animal information said...

Visitors who want to gift their pets and animals can find the animal names and animal poems link very suggestive.

Blog Archive

Labels