April 30, 2017

IQ and Socio-Economic Status

In this post the connection between IQ and socio-economic status (SES) is looked at. First, the fact that both traits are heritable is established. Then the causal interaction between the two is looked at. Finally, the idea that part of the association between these traits is explainable by the fact that the same genes are responsible for both traits is discussed.

  1. The Heritability of IQ and Socio-Economic Status

Variation in how high, or low, people’s SES is is largely attributable to genetic differences between them.Such are the results of several twin studies. For instance, Krapohl et al 2014 analyzed data on more than 13,000 twins and found that variation in high school test (GSCE) scores was 62% attribute to genes and 26% attributable to shared environment. Branigan, McCallum, and Freese meta-analyzed 34 twin studies from 9 nations and found that 40% of  variation in educational attainment was  attributable to genes while 36% of the variation was attribute to shared environment. Furthermore, Rietveld et al 2013 looked at data on over 125,000 people and found multiple genes were which associated with educational attainment across multiple samples.

Hyytinen et al 2013 found that life time income had a heritability of 24% for women and 54% for men. Hyytinen also review 19 previous samples from which the heritabiliy of income has been estimated. The typical finding is that about 42% of income variation is caused by genetics while about 9% is explained by the shared environment.

Finally, Trzaskowski et al 2014 analyzed data on roughly 3000 participants and found that around 30% of variation in children’s SES was attributable to genes.

1

That IQ is heritable is among the most consistent findings of behavioral genetics. Haworth et al 2010 utilized data on nearly 11,000 twins and found that the heritability of IQ was 66% at age 17 while shared environment explained only about 20% of IQ variation. Similar heritability figures are generated by studies that look at how genetic similarity predicts IQ similarity among unrelated individuals (Plomin et al 2013 and Trzaskowski et al 2014). Clearly, both SES and IQ are moderately to highly heritable.

  1. IQ and SES Correlate 

IQ test scores correlate with socio-economic status (SES). Smarter people tend to come from higher SES homes.  Sirin 2005 meta-analyzed data on roughly 100,000 students and found the mean correlation between cognitive ability and parental SES to be .28, indicating a weak to moderate relationship. This finding is similar to what was reported by 1996 American Psychological Association task force report on intelligence (Neisser et al. 1996). Given this correlation, the question arises as to whether it is IQ or SES which causes variation in the other variable. The answer is almost certainly both.

  1. IQ Influences SES

 Strenze 2007 meta-analyzed data on well over 100,000 people and found that the mean correlations between a person’s IQ measured at one point and their education, occupation status, and income, measured 10 or more years later, were .49, .41, and .22.

Some people argue that the association between IQ and SES is caused by the fact that people from high SES backgrounds have more cognitevly enriching environments. But there is good reason to think that this is false. For one thing, Strenze 2007 also showed that IQ is a better predictor of someone’s future SES than their parent’s SES is.

Moreover, Murray 1998 analyzed data on 12,686 subjects from the National Longitudinal Survey of Youth and showed that IQ differences within sibling pairs predicted future differences in SES. That is, the sibling with a higher IQ typically ended up having more education, a higher status occupation, and more income. Because siblings have identical parental SES, this analysis shows that IQ leads to higher SES independent of parental SES.

Further still, we know for large literature reviews that IQ is an excellent predictor of job performance. Consider, for example, the findings of Huffcutt and Arthur 1994:

2

Thus, IQ increases SES by, among other things, increasing job performance and this is why IQ predicts SES even within the same family.

  1. SES Influences IQ 

However, SES also influences IQ. This has been shown by multiple adoption studies in which poor children were placed into high SES families and, as a result, received a boost in their IQ. Specifically:

Schiff et al 1978 studied 32 French children who were abandoned by poor parents and adopted into wealthy families. These children were found to have IQs of 110. By contrast, their biological siblings who remained stuck in poverty had IQs of 94.5. Thus, adoption increased IQ scores by 15.5 points.

Skodak and Skeels 1949 studied 100 poor children whose biological mothers averaged an IQ of 85.7. At age 13, the children’s average IQ was 108. Thus, adoption increased IQ scores by 22.3 points. Major limitations of this study include that the IQ of the fathers is unknown and the study is from more than 50 years ago when environmental variation was more extreme than it is today.

Capron and Duyme 1989 studied 38 French children, 14 years old on average, and found the following results:

3

Thus, if you had high SES biological parents, whether you were adopted into a high or low SES family was associated with a 12.1 point difference in your IQ. For people with low SES birth parents, the difference was 11.2 points.

Kendler et al 2015 studied 436 male siblings pairs in which one brother was adopted while the other was not. The mean IQ of adopted siblings (97) was higher than the mean IQ of home reared siblings (92). Being adopted was thus associated with a 5 point advantage in IQ. Moreover, the IQ of adoptive siblings correlated at .2 with the educational level of their biological parents and .18 with the educational level of their adoptive parents. Thus, both variables had similar levels of influence. The IQ of home reared siblings correlated at .34 with their biological parent’s educational level suggesting that both genetic and environmental aspects of their parent’s SES influenced their IQ. Further supporting the causal role of parental education is the fact that adoptive children put into homes with lower levels of parental education than their biological home actually suffered a decrease in IQ.

Kendler et al replicated their findings in an analysis of 2,341 half brothers from the same data sets. The home reared sibling IQ averaged 95.2 and the adopted sibling IQ averaged 98.4. Being adopted was associated with a 3 point advantage over one’s siblings. The IQ scores of adopted half siblings correlated at .18 with both their adopted and biological parent’s education level which one again suggests identical levels of influence for the two variables. In both samples, IQ was measured at 18 years of age.

Kendler et al is recent, uses a large sample size, and tests IQ in adulthood, and thus suffers from none of the major limitations of the other studies in this literature. It, especially in conjunction with the other, more problematic, studies, is strong evidence that parental SES has a causal influence on a person’s IQ.

That being said, how SES influences IQ is unknown. But it is important to realize that money, or class, does not magically make people smarter. The most likely casual pathways by which SES might influence IQ involve nutrition, parenting styles, and how cognitively stimulating the home environment is.

  1. Genetic Mediation

Thus far, we’ve seen that SES and IQ are both partly heritable and that they both causally influence each-other. Since both traits are heritable, it is therefore worth noting the extent to which the same genes explain variation in both traits.

Trzaskowski et al 2014 looked at the correlation between genetic similarity, SES similarity, and IQ similarity in a sample of 3,000 unrelated children. Their genetic sample consisted of roughly 1.7 million SNPs and was therefore large, but still far less than the entire human genome. IQ was measured at ages 7 and 12. Family SES was measured at ages 2 and 7. They found that variation in the portions of the genome they analyzed accounted for about 30% of variation in both IQ and SES. In this sample, IQ and family SES correlated at .31 and 94% of that correlation of mediated by genetics. In other words, the reason why high SES 7 year olds were smarter than average was almost entirely because they had “smart genes”. The correlation between family SES at age 7 and IQ at age 12 was .32 and 56% of that relationship was genetically mediated. In other words, about half of the relationship between intelligence at age 12 and SES at age 7 was explained by people having smart genes.

This is direct evidence showing that a substantial amount, likely more than half, of the correlation between parental SES and IQ in adolescence is explained by the parents having smart genes and passing them on to their kids.

An important take away from this is that the studies looked at earlier showing that changing someone’s SES can change their intelligence doesn’t tell us anything about the ultimate sources of variance in IQ within our population. To better understand this, consider a hypothetical: imagine that differences in the amount of training people did for sports was 100% a function of whether or not they were naturally gifted at sports. In such a scenario, we would find that variation in athletic ability is almost entirely due to genetics even though forcing an unfit person to train would make them better athletes. This would be due to the fact that the people who trait are already the people who are naturally good at sports. As a result, the differences caused by training would really be an effect genes and, so, variation in training would be a proximate, but not ultimate, cause of variation in ability.

To a less extreme extent, the same thing is going on with intelligence and SES. Sure, putting poor people in a high SES environment would increase their intelligence. However, one reason why they aren’t in a high SES environment is because they aren’t very intelligent.

This is similar to how reading probably correlates with IQ, and making low IQ people read would probably boost their IQ, but the reason why they aren’t reading in the first place is because they have lower IQs. This is a little less intuitive in the case of parental SES because your parent’s SES is not your choice. But the same principle is at play.

Given that the relationship between parental SES and IQ was small to begin with, this suggests that SES does not causally explain much in terms of IQ variation within the general population.

 

Facebook Comments
  • Emil Kirkegaard

    Kendler used a young adulthood sample (18-20 at testing), not regular adults, so their finding of some shared environment effect is consistent with the decline of this effect with age to ~0 in adulthood.

    There is also the question of whether the smarter siblings were more likely to be adopted away. Adoption studies generally show that age of adoption has a negative relationship to outcomes, including IQ. This could be a selection effect, not an environmental effect. Maybe both.