06 April 2011

Beyond Heritable IQ in Socioeconomic Success

IQ is one of the stronger predictors of socio-economic success in life, and success on a wide variety of other measures, and a lot of a person's IQ is heritable.

Personality, IQ and Success

In a result, largely contrary to the claims of Malcolm Gladwell in his book, "Outliers" in which he claimed that the famous Terman study which looked at life outcomes for a large cohort of California students with an  IQ of 135 or higher on the Stanford-Binet in 1921-1922, over their entire lifetimes, a recent analysis of that data has shown that IQ is a significant predictor socio-economic success at all levels within this elite subgroup.  Male teman study members with IQs in the bottom of the range for inclusion, IQ 135 (roughly the 99th percentile), had lifetime earnings (in 2008 dollars from ages 18 to 75) of about $2,350,000.  Those in the top 10% of the study group (roughly the 99.9th percentile), had lifetime earnings of about $2,700,000, a statistically significant 15% difference of $350,000 that was stronger at the end of their careers than at the beginning.  Averaged over 40 years, the average annual income of those at the bottom of the IQ range was $58,750 (in 2008 dollars) and the average annual income of those at the top of the IQ range was $67,500.

Thus, there were lifetime economic benefits to being a genius, rather than merely "gifted" but the benefits weren't huge either.

More interestingly, the study also had data on the "Big Five" personality traits (conscientiousness, extraversion, agreeableness, openness and neuroticism) of the the study participants, the most widely uses psychological description of general temperament.  Wikipedia notes regarding the Big Five that:


A number of meta-analyses have confirmed the predictive value of the Big Five across a wide range of behaviors. Saulsman and Page examined the relationships between the Big Five personality dimensions and each of the 10 personality disorder categories in the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV). Across 15 independent samples, the researchers found that each disorder displayed a unique and predictable five-factor profile. The most prominent and consistent personality predictors underlying the disorders were positive associations with Neuroticism and negative associations with Agreeableness.

In the area of job performance, Barrick and Mount reviewed 117 studies utilizing 162 samples with 23,994 participants. They found that conscientiousness showed consistent relations with all performance criteria for all occupational groups. Extraversion was a valid predictor for occupations involving social interaction (e.g. management and sales). Furthermore, extraversion and openness to experience were valid predictors of training proficiency criteria.


The personality impacts of Big Five personality traits in the general population were largely confirmed in high IQ populations.  Moreover, Big Five personality traits are independent of IQ in high IQ populations:


[E]ven though the Terman sample has a restricted range of IQ, there is substantial variation in personality. In fact, the Terman men do not differ from the general population in terms of personality.


Evolutionary pressures appear to quite consistently favor higher IQ in the long run, but some models suggest that evolutionary pressures on personality traits favor an ecological mix of personality types rather than a single personality, so no one personality trait has translated into "fitness" in the long run (of course, if it had, we wouldn't notice it, since it would mathematically fold into IQ by virtue of the way that IQ is constructed by psychologists).

Big Five traits are heritable, but not as strongly as IQ, although some of the weaker association may be a product of a less refined measurement instrument, as the Big Five personality traits have been consensus personality measurements only since the early 1980s, while the concepts of "g" and IQ are among the oldest in psychology, which has allowed more time to refine the measurement of the concept.


Of four recent twin studies, the mean estimated broad heritabilities on self-report measures for the Big Five traits were as follows:

Openness: 57%

Conscientiousness: 49%

Extraversion: 54%

Agreeableness: 42%

Neuroticism: 48% . . .

Many studies of longitudinal data, which correlate people's test scores over time, and cross-sectional data, which compare personality levels across different age groups, show a high degree of stability in personality traits during adulthood. More recent research and meta-analyses of previous studies, however, indicate that change occurs in all five traits at various points in the lifespan. The new research shows evidence for a maturation effect. On average, levels of Agreeableness and Conscientiousness typically increase with time, whereas Extraversion, Neuroticism, and Openness tend to decrease.  . . .

Cross-cultural research from 26 nations (N = 23,031 subjects) and again in 55 nations (N = 17,637 subjects) has shown a universal pattern of sex differences on responses to the Big Five Inventory. Women consistently report higher Neuroticism and Agreeableness, and men often report higher Extraversion and Conscientiousness. Sex differences in personality traits are largest in prosperous, healthy, and egalitarian cultures in which women have more opportunities that are equal to those of men. Both men and women tend to grow more extraverted and conscientious and less neurotic and agreeable as cultures grow more prosperous and egalitarian, but the effect is stronger for men.
Conscientiousness overlaps with ADD inattentive symptoms, and both of these traits are also correlated with digit ratio which is an easy to measure physical trait associated with natal testosterone level exposures.  Specifically: "More masculinized finger-length ratios show associations with ADHD symptoms, possibly acting through the trait mechanism of conscientiousness."  This is in line with the cross-cultural gender discrepencies observed in onscientiousness.


The second digit to fourth digit finger ration has been shown to be a reliable indicator of testosterone exposure in utero. If your fourth finger (your ring finger) is longer than your second finger (the one next to your thumb) you had a high exposure to testosterone when your mother was pregnant with you. This ratio has also been associated with genital size, aggression, ADHD and now conscientiousness.


Thus, a significant cause of conscientious personality and ADHD is congenital, but not hereditary.  Digit ratio and IQ over 130 are also associated.


Conscientiousness and neuroticism are both strongly correlated with ADHD inattentive type, in that order, and agreeableness, neuroticism, and conscientiousness are strongly correlated with ADHD hyperactive type in that order).  ADHD, with the strength of the relationship being strongest between conscientiousness and ADD inattentive type diagnosis (r=-0.59), is the strongest of the Big Five/ADHD correlations.

The Big Five personality traits aren't perfectly independent of each other, but they are sufficiently independent of each other than a four factor model captures significantly less of the variation in personality from person to person, and the Big Five measure was created with the express purpose of identifying maximally independent factors.

Four of the Big Five personality traits had a measurable impact on lifetime earning of men with IQs of 135 of more.  Neuroticism, surprisingly, was completely unrelated to lifetime income.  In 2008 dollar terms, comparing the lowest decile to the highest decile in each of the other four traits, the lifetime impact of being high in each of the traits was approximately as follows (reading from a line chart):

Conscientiousness:  $800,000.
Extraversion: $650,000
Agreeableness: -$350,000
Openness: -150,000

Each of these traits showed a more or less linear relationship between decile in a personality trait and its economic impact.

Thus, an extremely conscientious, extroverted, disagreeable, and close minded individual with an IQ of 135 of more would earn an average of $1,950,000 more than an extremely non-conscientious, introverted, agreeable and open minded individual.  The combined effects of these four personality traits was about five and a half times as great as differences in IQ between the 99th and 99.9th percentile.

Even the difference between being in the sixth and tenth decile for conscientiousness has as much of an economic impact as the difference between being in the 99th and 99.9th percentile for IQ.

These four out of five personality factors and IQ influenced both the amount of education that a person received and the amount that a person earned independent of education:

Our results highlight the importance of personality and intelligence on our outcome variables. We find that personality traits similar to the Big Five personality traits are significant factors that help determine educational attainment and lifetime earnings. Even holding the level of education constant, measures of personality traits have significant effects on earnings. Similarly, IQ is rewarded in the labor market, independently of education. Most of the effect of personality and IQ on life-time earnings arise late in life, during the prime working years.
This adds to a growing literature on traits other than IQ that are important to lifetime successes.

Grit and Success

Another recent study looked at "grit" as a factor in spelling bee success.  The famous rule of thumb, also discussed by Gladwell, is that it take "10,000 hours of intense training (plus or minus a few thousand hours)" to reach the pinnacle of expertise in a pursuit.  But, Gladwell glosses over the question of what traits are necessary to be able to stick to an intense training regime for that quantity of time. 

The spelling bee study first found that the key to performance in spelling bees was the amount of deliberate practice that a kid put into it, which "involved studying and memorizing words while alone, often on note cards" as opposed to alternatives like being quizzed by others or leisure reading:

With each year of additional preparation, spellers devoted an increasing proportion of their preparation time to deliberate practice, despite rating the experience of such activities as more effortful and less enjoyable than the alternative preparation activities.
But:
Why were some kids better at drilling themselves with note cards? What explained this variation in hours devoted to deliberate practice? After analyzing the data, Duckworth discovered the importance of a psychological trait known as grit. In previous papers, Duckworth has demonstrated that grit can be reliably measured with a short survey that measures consistency of passions (e.g., ‘‘I have been obsessed with a certain idea or project for a short time but later lost interest’’) and consistency of effort (e.g., ‘‘Setbacks don’t discourage me’’) over time using a 5-point scale. Not surprisingly, those with grit are more single-minded about their goals – they tend to get obsessed with certain activities – and also more likely to persist in the face of struggle and failure. . . . Here are the scientists: "Grittier spellers engaged in deliberate practice more so than their less gritty counterparts, and hours of deliberate practice fully mediated the prospective association between grit and spelling performance."
In other words, practice is what produces results, but "grit" (a.k.a. "tenaciousness" a.k.a.  “perseverance and passion for long-term goals.”) is the key psychological trait that determines how much someone practices.  And, like the Big Five personality traits, grit is largely unrelated to intelligence:  "Grit was found to be either orthogonal to or slightly inversely correlated with intelligence."  It is not independent of the Big Five personality traits, however, although it doesn't completely overlap:
Grit was strongly correlated with conscientiousness (r = .77, p <.001 and r = .73, p <.001) (2009). While Grit is related to conscientiousness measures, it also differs from conscientiousness in important ways. For example, while both Grit and conscientiousness are often associated with short term accomplishments, Grit is also associated with longer term and multi-year goals. This long-term persistence and dependability are important aspects that make Grit unique from conscientiousness.
Grit has been validated in settings other than spelling bees.
Grit provided incremental predictive validity . . . above and beyond the Big 5 personality traits; . . . higher levels of Grit were more highly associated with cumulative grade point average (GPA) in an Ivy league sample when compared to those with lower Grit levels (r = .25, p<.01); [and] that Grit predicted retention after their first summer in two classes of cadets at the United States Military Academy[.] . . . Grit predicts beyond the typical and unrelated cognitive construct of IQ and can account for variance over and above what is observed in the Big 5 personality construct of conscientiousness.
Grit also predicts the performance of Teach For America teachers in "under resourced public schools.", as does life satisfaction prior to entering the program.

A refined version of the test that Duckworth and Quinn used in the spelling bee study from 2009, with four fewer items and a different means of using answers to produce a scaled score strengthened  "the positive relationships between Grit and educational attainment, GPA, retention in college, and success in a national spelling bee competition."

As Jonah Leher notes in a blog post as Wired, one take away lesson from these studies is that there are personality traits that are largely unrelated to IQ which are important to socio-economic sucess other than IQ, which Steve Hsu sums as as the "W" factor, and which we are increasingly starting to be able to describe with psychological testing; like IQ, a meaningful part of these traits are genetic, congenital, or arise early in life, although there isn't good research on the relatively recently well defined trait "grit."

The other is a bit more subtle and I'll reproduce his observation in his words:


[T]here’s a major contradiction between how we measure talent and the causes of talent. In general, we measure talent using tests of maximal performance.

Think, for instance, of the NFL Combine: Players perform in short bursts (40 yard dash, short IQ test, catching drills, etc.) under conditions of high motivation. The purpose of the event is to see what players are capable of, to determine the scope of their potential. The problem with these tests, however, is that the real world doesn’t resemble the NFL Combine.

Instead, success in the real world depends on sustained performance, on being able to work hard at practice, and spend the weekend studying the playbook, and reviewing hours of game tape. Those are all versions of deliberate practice, and our ability to engage in such useful exercises largely depends on levels of grit.

The problem, of course, is that grit can’t be measured in a single afternoon on a single field. (By definition, it’s a metric of personality that involves long periods of time.)

The end result is that our flawed beliefs about talent have led to flawed tests of talent. Perhaps that explains why there is no “consistent statistical relationship between combine tests and professional football performance.” We need  . . . a test that measures how likely people are to show up, not just how they perform once there.


Mostly, we measure traits like "grit" or "W" indirectly, by looking at how well people perform academically or otherwise, after controlling for traits like IQ.

Poverty and IQ

So, we know that IQ is important to socio-economic success, and that there are independent contributions from at least a handful of personality traits that are independent of IQ.

We also know that IQ is related to poverty in a manner more complex than a Bell Curve model with highly hereditary IQ in which the stupid tend to be both poor and prone to have low IQ children, while the smart tend t obe both rich and prone to have high IQ children is involved.  This is because the measured heritability of IQ is lower, while the importance of environmental impacts is higher, in the poor than in the well to do.

Also, a large subset of mental retardation cases are not genetic and instead depended on enviromental factors like fetal alcohol syndrome, lead exposure and inadequate pre-natal nutrition, and labor and delivery medical care.  While high IQ appears to be rooted in a large number of small, cumulative factors, low IQ frequently has a single dominant cause that is different in kind from the causes of IQ levels within the normal range.  Alleviation of these mental retardation drivers is an important factor in the Flynn Effect (secular long term trends towards higher IQ in the same population from generation to generation over time).

Put another way, you are born with a maximum potential IQ, but seriously adverse environmental conditions, mostly associated with poverty, can prevent you from reaching that potential.  These environmental effects appear to be particularly strong during pregnancy and in the first few years of life. 

With that key observation in mind, the relationship between American performance on international tests of academic ability and poverty can be put in context in the nature v. nuture debate:


[T]he brutal reality is that U.S. schools that contain more than 25% poor children do abysmally--they are literally pulling the scores down:

If we subdivide the U.S. data in a very obvious way, we observe something, well, rather obvious:
But data available now tells us that poverty, as usual, had a huge impact on PISA reading test scores for American students. American students in schools with less than 10% of students on free and reduced lunch averaged 551, higher than the overall average of any OECD country. Those in schools with 10 to 25% of students qualifying for free and reduced lunch averaged 527, which was behind only Korea and Finland.

In contrast, American students in schools with 75% of more of children in poverty averaged 446, second to last among the 34 OECD countries.
We might not be living in a nation where one-third of a nation is ill-housed, ill-clad, ill-nourished. It's only one in five, give or take. . . . Until we get serious about reducing poverty, as well as breaking up large geographic concentrations of poverty, our average test scores will be poor.
This poverty effect is not simply a product of poor students having lower maximum potential IQ at birth because their parents are below average in IQ and poor as a result.  The conditions of poverty prevent thse children from achieving their potential.

You don't have to go far afield to see this effect, even within different schools that are mostly poor in the same district.  A child identified as "gifted" who attends North High School in the Denver Public Schools is far more likely to go to college and not require remedial work than a child identified as "gifted" who attends West High School, with an even more impoverished student body, a mile or two away.

In a nutshell

IQ,  which has a strong hereditary component in the absence of adverse environmental factors, has an impact on socio-economic success at all levels, but among the smartest, personality factors, which can mostly be summed up in a quite small number of measures like extraversion, conscientiousness, grit, agreeableness, and openness, are even more important with the high IQ population.  These personality traits also have some hereditary component, although probably significantly less than for IQ.  Envirommental conditions strongly assocaited with poverty, particularly during pregnancy and early childhood, prevents children from reaching their genetic IQ potential and also has a measurable impact (generally negative) on personality traits associated with socio-economic success.

Implications

This has several implications of importance:

1.  If we want to measure likelihood of success most accurately in a wide range of academic and occupational fields, a multi-dimension measure that captures both IQ and "W" factors is necessary. 

2.  While many "g" intensive activities that are proxies of IQ have been identified, less systematic work has been done to identify "W" intensive activites, like spelling bees, that are proxies for factors like "grit" and can be used to identify children who excel in these non-IQ dependent traits.

3. We should also devote research to ascertaining which occupational and academic pursuits are more or less "W" heavy, and whether there are pursuits like math and physics in the IQ demain within academics, where there appear to be theshold effects, i.e. where a minimum level of certain "W" factors are a minimum for real success, rather than making a linear contribution to outcomes.  Does one need a minimum level of "W" factors, for example, to make it as a special forces soldier or a medical doctor or an accomplished musician?  Similarly, are there occupational or academic pursuits to which individuals who are high in IQ but low in W factors, or high in W factors but low in IQ, are likely to be best suited?

4.  Much but not all socio-economic benefits of both IQ and W factors are mediated through practice and educational opportunities that must be available for people to take full advantage of what these traits offer.  Identifying IQ and W rich individuals is a beginning, not an end point in developing human potential.

5.  Research into the existent to which "W" factors are malleable and the environmental impacts that influence these factors is worthwhile because research on this for IQ has been exhaustive, but research for other factors has not been, because W factors as important to societal success as IQ, and because W factors appear to be less strong hereditary than IQ.  Indentification of enviromental factors that enhance "W" factors should be a priority for researchers.  What parenting and educational approaches best foster the non-environmental components of "W" factors and to a lesser extent IQ. 

6.  Long term success in activities from probation and parole to high school retention to college retention to adherence of mental health medications may be strongly related to W factors that vary from person to person.  This makes these factors attractive as components of validated statistical prediction tools to determine what management approach should be supervising people in these situations.  Individuals who have low conscientiousness or little grit may require more intensive supervision than those who lack it.

7.  The single most promising way to improve aggregate IQ and socio-economic performance in American society is to identify and address ways that poverty inhibits them.  Particularly attention should be devoted to reducing hardship and improving medical care for pregnant women and young children.  Another area where there is an already identified problem that can be addressed is to improve the quality of pre-college educational opportunities, college counseling and collegiate financial aid for children with established strong academic ability from less affluent families.

8.  To the extent that we can psychometric for selected "W" factors that are at least as good as those for IQ, there should be an active and open ended inquiry into what kind of patterns are seen in the residuals of performance in these combined measures.  This inquiry is useful even if the residual factors turn out to be matters such as pure luck or cohort timing that demonstrate that psychological factors related to socio-economic success have been exhausted.

No comments: