IQ → Academic Achievement

Share on FacebookShare on Google+Email this to someoneTweet about this on Twitter

Count ‘em: one, two, three new studies on the relationship between IQ and academic achievement in the latest issue of Intelligence (volume 35, issue 1)

Before the studies’ precis, a little background on why such studies are necessary. More than anything, such studies are needed because folks such as S. Ceci and R. Sternberg (very prominent and oft-cited) advocate that (traditional) IQ tests are just measuring little more than school related achievement. So, IQ and academic achievement are only related because, for reasons X, Y, and Z (pick your own environmental variables), some folks get more out of school, and it just so happens that the same folks do well on IQ tests due largely (if not entirely) because school achievement and IQ tests are measuring the same thing. Consequently, g is an irrelevant artifact of those damned psychometricians.

An alternative hypothesis (explicated nicely in Jensen1,2), however, is that due largely to genetic factors (which influence both individual differences and environmental influences), people enter school with wide variability in cognitive ability and “readiness to learn.” This initial variability then heavily influences (although not completely determines) the amount a given student will pick up as he/she matriculates. As a student gains more information, his/her initial ability and the new information acquired then interact so he/she is able to expand his/her knowledge further, and so on and so forth. Therefore, while one needs access to “information,” the child’s general cognitive ability is the engine driving his/her educational achievement.

Don’t miss the point here. These are two separate, testable, hypotheses. (A) IQ and academic achievement are synonymous. That is, people are smart (or not so smart) almost solely because they had (or did not have) a good education. (B) IQ is independent of academic achievement, although the former significantly influences the latter. That is, you can come from a good school, but not be so bright, and do poorly on achievement tests; likewise, you can come from a school that is not so good, (but meets some very minimum standard), but be bright, and do very well on academic achievement tests.

Now, the studies…….

1) Treena Eileen Rohde and Lee Anne Thompson: Predicting academic achievement with cognitive ability

This study is likely the weakest only because they used a group of college students from an elite university. Not that there is anything wrong with this, but when you see the samples in the studies below, it is a noticeable concern.

Their major contribution was that in predicting (standardized) academic achievement, speed of information processing and spatial ability can explain small, but significant, amounts of variance unexplained by general vocabulary (Mill Hill) and perceptual organization (Raven’s Matrices), although the latter two tests, hands down, did the best in predicting academic achievement across various indicators.

In their own words:

General cognitive ability measures (Raven’s, Vocabulary) and specific cognitive abilities (working memory, processing speed, spatial ability) collectively accounted for between 16% [GPA] and 54% [SAT] of the variance in academic achievement.

2) Marley W. Watkins, Pui-Wa Lei and Gary L. Canivez: Psychometric intelligence and achievement: A cross-lagged panel analysis

This study had 3 advantages over the former: (1) it is longitudinal, (2) the data is from a much wider scope of IQs, and (3) the data comes from all over the US. The drawback, and major caveat, is that the data is all from special education (broadly defined) testing, so the applicability to the entire population is in question. Still, the mean Full Scale IQ score from the WISC-III (the IQ instrument used) is 90 with a SD of 15 (in the general population it is 100 and 15), and the subtest scores hover around 8 with SDs that hover around 3 (in the general population it is 10 and 3).

Because they have longitudinal data on both standardized IQ and standardized achievement tests, they can specifically test the IQ—>Achievement hypothesis (see preamble). What do they find?

This notion of intelligence estimating a student’s ability to succeed in school assumes the temporal precedence of intelligence to achievement. . . Regardless, the present study supports the view that intelligence, as measured by the VC [Verbal Comprehension] and PO [Perceptual Organization] dimensions of the WISC-III, influences or is related to future achievement whereas reading and math achievement do not appear to influence or are not related to future psychometric intelligence.

Stated more bluntly:

. . . the present study provides evidence that psychometric intelligence is predictive of future achievement whereas achievement is not predictive of future psychometric intelligence. This temporal precedence is consistent with the theoretical position of Jensen (2000)[1] that intelligence bears a causal relationship to achievement and not the other way around.

3) Ian J. Deary, Steve Strand, Pauline Smith and Cres Fernandes: Intelligence and educational achievement

Before getting on to the study, a brief word about Dr. Deary. He is the current badass of differential psychology. Because of his background (degrees in medicine and psychology), he is able to investigate psychometric, chronometric, genetic, and neurological aspects (often concurrently) of both intelligence and personality (look at the range on his vita). As if that were not enough, he has challenged the whole field of differential psychology by obtaining multiple population level, longitudinal data sets. So instead of trying to infer from a sample a few hundred to the target population, he is gathering population level samples of thousands of individuals. Case in point:

Deary’s study looked at how cognitive ability measured at age 11 predicted academic achievement at age 16. Unsurprisingly, the IQ-Achievement correlations for the Sciences are around .6 (math highest, chemistry lowest), with similar coefficients form Arts/Humanities and Social Studies. Surprisingly, for practical fields (e.g., P.E., Art) the coefficients are a little lower, but not that much, averaging around .5. Here is a pic of the correlation table: (the n is in parentheses; it obviously changes as not every student took every class)

Deary took the analysis a step further however and did a little latent variable modeling. As the IQ test had three components/subtests (verbal, nonverbal, quantitative), he correlated a latent g factor with a latent academic factor using the following subtests: English, English Literature, Math, Science, Geography, French (n=12519). The correlation between the latent factors was .81. That is: 66% of the variance in latent (general) academic achievement can be explained by latent cognitive ability—measured 5 years previously. While he hypothesizes that such things as “school ethos” and “parental support” are good areas to search for the other 34%, based on Rohode’s work, it is likely going to be found in residual, first order factors (see Carroll or McGrew).

Take home message: While general cognitive ability and academic achievement are not isomorphic, the former is necessary for the latter, while the converse is not necessarily true. Spearman suggested this more than a century ago, and, to quote the last sentence in Deary’s work,

These data establish the validity of g for this important life outcome.

1. Jensen, A. R. (2000, August). The g factor and the design of education. Paper presented at the annual meeting of the American Psychological Association, Washington, DC.

2. Jensen, A. R. (1989). The relationship between learning and intelligence. Learning and Individual Differences, 1, 37-62.



  1. Gor blimey, Guv, who’d a doubted it? More seriously, I wonder if there is a “threshold age” at which g dramatically improves as a predictor of performance-five-years-hence? Age 8? Age 16? Age 22? Age 25? Presumably the old literature would give a pretty good idea.

  2. And I wonder how much of that “threshold age effect” is due to different rates to maturity among the races. 
    AKA will the threshold age be different or more predictive at different ages for monoracial and multiracial societies?

  3. Good point, OK. It’ll differ between the sexes too, I suspect, unless it happens to be a pre-puberty age.

  4. Once again Ian Deary has got excellent population data and combined it with other data sets to produce clear results. His work is really taking on the concept of intelligence in terms of real life utility, and when his health and longevity studies are eventually completed we shall have a strong case for g being a fitness indicator of the highest order.

  5. Let’s see: Kids sit side by side in the same classrooms getting the same instruction from the same teacher(s), but some have higher achievement than others. Since everyone must be born absolutely equal, the difference can’t be related to innate ability (intelligence, IQ). That would be unfair. Prize logic.

  6. Alex, 
    Fascinating material. Yet basically psychometricians have been trying to hammer home the message that education is determined by educability and not vice versa ever since … the era of Cyril Burt and the London School. 
    The question is: will the message EVER reach the mass media and the educated public in general? 
    How many goddam meta-analyses and 50-year mass longitudinal studies and monozygotic twin studies will be needed before the egalitarians concede defeat? 
    The staying power of ideology never fails to amaze. 
    Cathal Copeland

  7. Could someone post Dreary’s study to the documents section of the forum? It sounds like an interesting read.

  8. Let’s see. If we allow that g is causative in relation to educability, then we should probably also allow that g has some causal relationship to success in life (see, for example, Gottfredson) … which in turn means that maybe the lack of the same levels of achievement by certain groups is not (all) due to pervasive discrimination … 
    That might be a tough sell.

  9. Kevin: It is posted. 
    Gardner writes: 
    The same means of early detection can so easily be put to malevolent purposes. First of all, we won’t just determine deficits that can be addressed, but also ones that cannot be addressed. Second, we run the risk of stigmatizing children from birth?”Oh, you are destined to be illiterate,” or “you’ll never be likeable, because of your social deficits.” Moreover, we will be likely soon to turn not just to deficits, but to efforts to produce the perfect child?to enhance perfectly adequate capacities through genetic, neural, or pharmacological interventions. Not only does this seem to go against human nature and fate as we have known it; it will also privilege further those who are already privileged. 
    Gag. He is setting up this straw argument of people who want perfect children and will stop at nothing to get them. Oooo. Maybe they will turn out to be the next Batman villain, because in real life I have never seen these people. I have met parents who want the best for their, but to make up a group of uber diabolical parents who want perfect children is ridiculous as there is no such thing as a perfect genotype. Success of one’s attributes is very dependent on the environment. What is successful in the Congo is not in, say, Cambridge , MA. Moreover, even in Cambridge, there are a whole host of traits that are successful, so how would one choose? I wish people would stop making not-so-subtle references to Nazism every time genetics/intelligence, etc. are mentioned. Stalin was all for “equality of the masses,” and we know how well that experiment worked. 
    I agree. It amazes me that Burt did much of his work in Education, and it was well received. Now, his ideas are looked upon with horror in Colleges of Education. I think a good chunk can be changed by changing the structure of the academic faculty. Not that everyone has to toe the London School line, but having an arena of open dialog and an environment where contradictory hypothesis are allowed to be voiced and (gasp!) tested would be a great step forward.

  10. Alex, yeah, I did not post the Gardner stuff seriously, and I expect that most efforts to emeliorate learning difficulties will be genetic in nature.

  11. [...] IQs were not. Preschoolers with high vocabularies are just reflecting their superior education. But here’s a nice overview of three recent studies that specifically test whether education drives cognitive ability or the other way around. All [...]