As an aside in a fascinating City Journal piece on educational policy, A Wealth of Words:
Vocabulary doesn’t just help children do well on verbal exams. Studies have solidly established the correlation between vocabulary and real-world ability. Many of these studies examine the Armed Forces Qualification Test (AFQT), which the military devised in 1950 as an entrance requirement and a job-allocating device. The exam consists of two verbal sections (on vocabulary size and paragraph comprehension) and two math sections.
The military has determined that the test predicts real-world job performance most accurately when you double the verbal score and add it to the math score.
Once you perform that adjustment, according to a 1999 study by Christopher Winship and Sanders Korenman, a gain of one standard deviation on the AFQT raises one’s annual income by nearly $10,000 (in 2012 dollars). Other studies show that much of the disparity in the black-white wage gap disappears when you take AFQT scores—again, weighted toward the verbal side—into account.
Are we surprised that high verbals can talk themselves into more generous remuneration? But in any case the power of vocabulary is why I believe that the GSS and IQ correlation is probably robust. And speaking of vocabulary, the author alludes to the now well known phenomenon that children from low socioeconomic status backgrounds tend have a much smaller vocabulary than than those from higher socioeconomic status backgrounds, and how that leads to a positive feedback loop that determines life trajectory. The main confound that comes to mind is that those from low vocab households are probably also from less intelligent households, and intelligence is heritable. But, proactive social engineering probably can break apart the gene-environment correlation at least, and dampen the variance in phenotypic outcomes. And this is where the policy prescriptions may not be to anyone's liking. On the one hand this social engineering is social engineering, and probably will cost money. Conservatives will not like that. But, I also suspect that much of the positive value of a non-home environment is going to be abolished as the child matures and begins to self-select peer groups from their own socioeconomic milieu. In other words you need to attack the milieu, the culture of poverty and anti-intellectualism. And I suspect many liberals will not be comfortable with the aggressive paternalism that that implies. So nothing will get done. Addendum: Again, the best thing you can do to have smart well behaved children is to select a spouse with those characteristics.