10 January 2011

Achievement Gap Starts Very Early (Updated)

Achievement Gaps Are Huge At Age Four

In the mid-1960s, Betty Hart was a graduate student in child development working at a preschool in Kansas City, Kan. The preschool was for poor kids — really poor kids. Many came from troubled housing projects nearby. . . . Hart's job was to teach these underprivileged kids how to speak like the children of her professors at the University of Kansas.

For years, she and university professor Todd Risley worked tirelessly toward this goal, doing everything they could think of to expand the vocabularies of these 4-year-olds. The idea was that if the kids could speak with the fluency of their wealthier peers across town, they might go on to similar academic achievements.

But success was elusive. "We tried everything," Hart says. "Everything we could. ... We couldn't do it."

The problem, they realized, was that they weren't getting to the kids early enough. Which led to this question: If age 4 was too late, when was early enough?


From here

Forget achievement gaps in third grade CSAP scores (where they are already stark), let alone achievement gaps in high school. According to these researchers who deeply wanted to reach a different result, the achievement gap is already deep and almost insurmountable at age 4!

Talking To Babies Seems To Make A Big Difference

Their detailed study of forty families from poor, middle class and rich backgrounds found that the affluent talked to their babies and toddlers three times as much as the poor.

According to their research, the average child in a welfare home heard about 600 words an hour while a child in a professional home heard 2,100.

"Children in professional families are talked to three times as much as the average child in a welfare family," Hart says.

And that adds up. Hart and Risley estimated that by the age of 4, children of professional parents had heard on average 48 million words addressed to them while children in poor welfare families had heard only 13 million.


A study by Alan Mendelsohn in the journal the Archives of Pediatrics and Adolescent Medicine this month says that early intervention to have poor parents interact verbally more with their children (in fifteen sessions of 45 minutes each) works.

Skeptics question how long the benefits of this kind of intervention will last, although studies showing that quality pre-schools of very long lasting positive impacts give some hope that the short term success of Mendelsohn's inteventions could persist, and there is a body of evidence that seems to show that catching up gets harder and harder to do as time passes.

The Case That This Is Really A Manifestation Of IQ

An alternative possibility is that IQ, which has been shown to have a strong hereditary component, is simply low in both poor children and poor parents, relative to their middle and upper class peers. There are a number of reasons to think that this might be the case:

1. The earlier you get deep, almost insurmountable difference in cognitive function, the harder it is to attribute that difference to environment. If environment has an immense impact in the first four years, it is hard to think that it has so pitiful an impact in the rest of a person's life.

2. This is particularly true given that somone who learns a new language in a total immersion environment as a young teenager can usually become an almost accent free teenager with a near adult fluency in that new language. The language learning ability, per se, does decline as one gets older and older second language learners almost never attain unaccented fluency in the new language, but the cognitive barrier to learning a new second language closes about a decade after one turns four.

Factors like vocabulary and ideomatic usage are large data sets that have to be learned, but the set of vocabulary words that distinguish the dim from the bright is on the order of a few thousand words, and the number of irregular usage situations necessary to really distinguish one as educated or not probably number under a thousand.

In order words, for the "talking to baby produces lifetime achievement" hypothesis to hold water, one has to infer a more indirect relationship -- that talking to baby stimulates a broad spectrum of time limited brain development activity -- rather than simply inferring that talking to baby provides more opportunities to learn words and master other elements of language.

3. It is hard to believe that the correlation between social class and parenting style is a strong as it would have to be to produce the achievement gaps observed. Certainly, it is easy to imagine that professional parents talk more to their babies on average than parents who are in poverty. But, is that difference really close to universal? Are their really no bright upper class kids who had low levels of child-adult verbal interplay as babies and toddlers?

4. If talking to babies is so important and anyone can do it, why hasn't our culture in millenia after millenia attached more importance to this, out of intuition? Why does a little bit less talking to a baby seems socially like a fairly minor parenting point instead of a huge parenting imperitive?

The Case That Talking To Babies Matters A Lot

Of course, there are also reasons to take these findings, particularly in light of Mendelsohn's study that shows that early intervention does matter a lot, seriously. The literature on the long term benefits of high quality pre-schools is one. The fact that a lot of brain development does take place at a very young age is another, as is the literature supporting a breast feeding-IQ link. The literature that shows that IQ looks much more hereditary in middle class and upper middle class families covered by twin studies than it does in poor families, where environmental factors seem to play a larger role, is another. The apparently evidence that the Flynn effect (i.e. long term secular trends in entire populations towards higher IQ in recent times) is mostly due to improvements environment for those who were at the low end of the IQ scale in the earlier periods is another.

[UPDATE added 1/12/2011]

Razib at Gene Expression does a brilliant job of summarizing the cutting edge literature on the link between social class and the heritability of of intelligence. He lays out the data from the 2003 study that shows that genetics matter much more in high socio-economic status families, while shared environment is the predominant factor for those at the bottom of the socio-economic scale. The tendency of improved socio-economic status to make genetics a larger share of IQ variation is surprisingly linear from about the 25th percentile of socio-economic status (below which the effect is less intensely connected to socio-economic status percentile) through about the 85th percentile (after which the returns to additional socio-economic status dampen modestly).

He also presents striking data from a 2010 study demonstrating that while cognitive ability has little to do with socio-economic status at 10 months at any socio-economic level, that by the time a toddler is two years old socio-economic impacts on cognitive ability are already pronounced with heredity being much more relevant (50% of the variation v. almost none in low SES toddler and an intermediate amount for those in the middle) in better off toddlers.

Put another way, genetics seems to be primarily relevant to peak potential IQ, but peak potential only begins to be relevant once one's environmental conditions are favorable enough to have an opportunity to achieve that potential.

Great environments prevent you from being brought down by your circumstances, but they also aren't enough by themselves, as he succinctly sums it up, "Practice may make perfect, but perfection may be a goal to which you aspire only if you have initial talent or inclination." Heredity becomes more important as one gets older and has more control over one's environment. Those who received boosts from exceptional environments beyond their comfort range slack off when given the chance, those who were held back by poor environments tend to thrive once they are free to make their own choices.

His comparison to stature, is apt. He notes that "environment explains more of the variance in height in developing nations, while it explains almost none of the height in developed nations." Better nutrition in rich countries allows people to reach their full potential in height. But, in poor countries poor nutrition prevents people from reaching peak potential. His observation that there "are probably diminishing marginal returns on “nurturing” environments for a child when it comes to their intellectual development," which flows from the fact that genetics accounts for close to 85% of variance in IQ by the time one hits the high end of socio-economic status, is also apt.

Still, he also notes, as I have, that social class differences in IQ (that often flow from IQ), which are exacerbated by assortive marriage (or assortive mating anyway), potentially confounds the model of good environment as something that facilitates peak potential. In other words, perhaps environment is less capable of suppressing the IQ of generally bright children than is in the case of generally less bright children whose intellectual development advances only with good parenting.

Along the same lines, perhaps some of the variation in hereditary influence is a product the of the IQ measuring instrument which is designed to arbitrarily reproduce a normal distribution when it could be that this is a poor approximation of the structure and distribution of intelligence for those at the extremes of the range that IQ tests are designed to measure. The high level of inaccuracy of IQ tests at predicting academic performance in autistic children is suggestive of this possibility, as are the strongly diminishing correlation of IQ with success in the real world amongst very high IQ individuals. Arbitrary units for quantifying IQ that inaccurately measure to relative magnitude of IQ differences in the middle range relative to some hypothetical absolute scale would also add systemic noise to the genetics-environment data in the middle of the range.

By analogy, it you try to do temperature calculations using temperatures measured in degrees Celsius, which arbitrarily sets zero at the freezing point of water and one hundred at water's boiling point, for high temperature phenomena like metallurgy, the correct equations of thermodynamics are going to work reasonably well. But, a failure to use a Kelvin scale (which puts zero at the laws of physics defined absolute zero of approximately -273.15 degrees Celsius) for room temperature and low temperature phenomena, using Celsius temperatures in the correct thermodynamic equations is going to produce widely inaccurate results. Purely proportional relationships to temperature, for example, will look very unproportional, will make a correct mechanism for the relationships difficult to conceptualize, and will produce ugly equations in situations where temperature is just one of several variables.

If we used some non-arbitrary zero and peak number to measure intelligence, or used an intelligence measurement unit related to some real world quantity other than percentile in performance (e.g. seconds of processing time to solve some problem that people in the IQ range that the test is measuring can solve in reasonable amount of time), some of the observed IQ properties might disappear as artifacts of the measurement method.

[End Update.]

Maybe IQ is mostly genetic if you have a decent environment, but that insufficient verbal interactions with adults relative to some critical threshold is as important as insufficient folic acid in pregnant women.

Also, even if IQ is a significant cause of the gaps at age four [update: or even two!], if talking to babies more could narrow that gap, even modestly, this provides a concrete way to make everyone in society better off on average by doing something that almost no one believes to be harmful to either parents or children. The downside risk of following this parenting advice is nil. But, the guilt potential for parents who ignore this advice, if it later studies confirm that it is very important, is huge.

And, if this really is important and will soon become part of the catechism of important tips for young parents, it certainly wouldn't be the first time that a parenting choice previously believed to be of trivial importance, like putting babies to bed on their backs, actually has a very significant positive impact.

No comments: