Wednesday 1 July 2015

Back to school: Elijah Armstrong writes

 

You may remember a previous post on whether being at school boosts intelligence.

http://drjamesthompson.blogspot.co.uk/2015/01/school.html

Re-reading it now I recollect that there are strong papers suggesting a boosting effect on intelligence, yet I had remained dubious about the strength of the effects, and I would still like to see a long data set to confirm the imputed effects. If staying longer at school really boosts intelligence then there should be a discernable and sustained jump in ability from previous levels. My scepticism may be simply because I had followed the previous orthodoxy that schools boost knowledge and skills, but not children’s underlying intellectual horse-power. Intelligence is the engine, education the gears.

You may also recall another one of my quips, cast as a tweet: Academic debates should be punctuated by long moments of silence, broken only by pages turning as the combatants do the necessary reading.

As testimony to the inherent wisdom of this pronouncement, it was a delight to receive a further comment on “School” a mere 162 days later. We do not rush things here. Elijah Armstrong has a “Psychological Comments” VIP card and Executive Lounge entrance key, so it is a pleasure to post up his comments immediately.

Elijah writes:

Sorry for the very late reply, but as the issue is quite important I think a response is warranted.

Ritchie et al. find, in their regression analysis, that school exerts a strong effect on crystallized intelligence but not on fluid intelligence or chronometric g –– as one would expect. If their findings were really the result of measurement error, why don't they find spurious effects across the entire intelligence domain?
Further, this analysis dovetails with other studies using similar designs––studies of uneducated, isolated communities; of regression discontinuity designs; of the effect of closed schools, during desegregation, on the IQs of communities; etc.


Yes, these studies have flaws. Maybe the measurement error that contaminates the Ritchie et al. study was actually confined to the verbal/crystallized domain, or maybe it's a product of the fact that they only used verbal tests to assess childhood IQs. Maybe isolated communities really have low IQs because they're inbred or malnourished, not because they're uneducated. Maybe the schooling effects found using regression discontinuity designs don't persist into adulthood. But these are ad hoc explanations: in essence, degenerative science. The total evidence that schooling influences IQ is quite strong.


JayMan: you implicitly admit that schooling influences IQ, above, but claim that its effects are likely g-hollow. This is an interesting question. However, it should be pointed out that even non-g increases in IQ may be valid. We know this because TBI, foetal alcohol syndrome, prenatal cocaine exposure, and malnutrition all lower IQ –– and are often causes of genuinely low "intelligence" –- but the method of correlated vectors suggests that none of them are on g. Thomas Coyle has also shown that SAT and ACT residuals (controlled for g) are good predictors of performance in allied college majors.


Furthermore, there are specific abilities or sources of variance for every cognitive task that has thus far been studied, including "real-world" ones like job performance: there are no really pure measures of g; so it is at least theoretically possible to increase or decrease a huge variety of specific abilities but leave biological g untouched, and thereby increase "real intelligence". I think this is what the Flynn effect has done, and very likely it's what educational effects do as well.
But this is an open question.

Three good studies on the issue would be:

a) a study of the effects of schooling on a massive array of cognitive abilities, including every Stratum II in Carroll's taxonomy, plus on a large set of information domains (including non-academic ones like sport, fashion, agriculture, etc.), plus on psychophysical variables like brain size and neural conduction velocity;

b) a regression-based mediation study to see if educational IQ gains affect occupational status, mortality, crime, financial distress, income and net worth, all that good stuff (this one suggested to me by Stuart Ritchie);

c) thorough interviews and impressionistic assessments of uneducated people. (I was in Dominica early this year, where most people are quite unlettered and the average Raven's IQ is ~70, and they didn't seem that dumb at all. But I only rarely saw them doing complex or abstract tasks.)

Elijah

5 comments:

  1. "Yes, these studies have flaws. Maybe the measurement error that contaminates the Ritchie et al. study was actually confined to the verbal/crystallized domain, or maybe it's a product of the fact that they only used verbal tests to assess childhood IQs. Maybe isolated communities really have low IQs because they're inbred or malnourished, not because they're uneducated. Maybe the schooling effects found using regression discontinuity designs don't persist into adulthood. But these are ad hoc explanations: in essence, degenerative science. The total evidence that schooling influences IQ is quite strong."

    Uh, no. I'd call these valid criticisms. It's not good science to build a case on something where every substantiating piece is suspect, as is the case here. It's not like racial differences in IQ or the results of behavioral genetics where clear results emerge from multiple corroborating sources. These are all pieces of evidence that are individually specious and indeed, contradicted by other evidence (such as a study of all the working age male twins in Denmark that finds that education has zero additional effect on eventual income).

    "JayMan: you implicitly admit that schooling influences IQ, above, but claim that its effects are likely g-hollow. This is an interesting question. However, it should be pointed out that even non-g increases in IQ may be valid. We know this because TBI, foetal alcohol syndrome, prenatal cocaine exposure, and malnutrition all lower IQ"

    Such studies have all the same problems. Every single one could be (and many have been shown to be) genetically confounded. Wait, on that one...

    "Furthermore, there are specific abilities or sources of variance for every cognitive task that has thus far been studied, including "real-world" ones like job performance: there are no really pure measures of g; so it is at least theoretically possible to increase or decrease a huge variety of specific abilities but leave biological g untouched, and thereby increase 'real intelligence'"

    Does learning how to play Tetris make me smarter? I think the ideas of near transfer and far transfer come into play.

    "I think this is what the Flynn effect has done, and very likely it's what educational effects do as well.
    But this is an open question."


    It sure is.

    "Three good studies on the issue would be:

    a) a study of the effects of schooling on a massive array of cognitive abilities, including every Stratum II in Carroll's taxonomy, plus on a large set of information domains (including non-academic ones like sport, fashion, agriculture, etc.), plus on psychophysical variables like brain size and neural conduction velocity;

    b) a regression-based mediation study to see if educational IQ gains affect occupational status, mortality, crime, financial distress, income and net worth, all that good stuff (this one suggested to me by Stuart Ritchie);

    c) thorough interviews and impressionistic assessments of uneducated people. (I was in Dominica early this year, where most people are quite unlettered and the average Raven's IQ is ~70, and they didn't seem that dumb at all. But I only rarely saw them doing complex or abstract tasks.)"


    Agreed. Except I would add some way to differentiate the exogenous effect of education from endogenous effects. Who fails to get educated, especially in today's world?

    ReplyDelete
  2. I will add:

    "If their findings were really the result of measurement error, why don't they find spurious effects across the entire intelligence domain?"

    The nature of the study teases out specific kinds of error: on whatever abilities which are relevant to educational attainment.

    ReplyDelete
  3. When i was teen, i no had any motivation to study, specially stem stuff like Biology and Geometry, because i always question why study mathematics or physics if i don't use it (specially, the complex material) in my future** Because, probably, my brain was still developing. Today, when i see some school material like Biology ( cells, plants), even i don't like read it, i memorize faster than in my cheathood.

    And because, my self conservation natural strategy is not correlate with math or memorize superficially the material.

    I always questioned and when they did not give me convincing arguments, I always internalized bitterness to do what we did not want to do.

    ReplyDelete
  4. Statements like "school exerts a strong effect on crystallized intelligence but not on fluid intelligence" are difficult for me to interpret in light of research - already 10 years old now - reporting that the structure of human intelligence is verbal, perceptual, and image rotation (VPR), not fluid and crystallized. You can read all about it in

    Johnson, W., & Bouchard, T. J. (2005). "The structure of human intelligence: It is verbal, perceptual, and image rotation (VPR), not fluid and crystallized." Intelligence, 33(4), 393-416.

    Cattell was an interesting fellow, but not of the same caliber as contemporaneous psychologists like Eysenck.

    ReplyDelete
  5. Elijah writes:
    Mark, I've criticized Gf-Gc theory myself. (I'm somewhat biased against the theory, as my Gc is >2 SDs higher than my Gf!) I've come around to the theory somewhat –– I think Hunt's remark that Gf-Gc theory is most relevant for occupational and age-related criteria, but VPR is the better statistical fit, is likely correct. But all of this is beside the point. If you don't accept Gf-Gc theory, you could simply rephrase my statement above as "education affects verbal ability, but not perceptual/rotational intelligence or g".

    ReplyDelete