The relationship between geography and IQ

[ronburgundy]

You would be correct.
IQ is correlated similarly with latitude and GDP at about .75.
[/URL]
That doesn't really undermine the point though. Latitude is correlated to genetic lineage, ancestral history of migrations and exposure to various selection pressures.
The genetic contributor to IQ predicts that it should be correlated to latitude, because latitude impacts some of the evolutionary pressures that would have shaped the genetics that contribute to IQ.

It does mean that latitude could cause other environmental factors that then impact GDP in ways independent of any influence of IQ.
However, analyses that use both IQ and latitude as simultaneous predictors of GDP, show that IQ is the strongest single predictor, which is more consistent with a direct role of IQ rather than a spurious relation due to shared influence from latitude related environment.

I'd be careful if I were you; Every single person who believes that correlation implies causation is going to die.

GDP may be a strong predictor of IQ; after all, education ain't free, and poverty tends to lead to less time in school - children in poverty have to work, rather than study.

I don't assume correlation = causation, but I am not so scientifically illiterate that I think that such a vacuous mantra is a valid intellectual reply to data that shows more than just simple correlations.
I specifically mentioned alternative causal possibilities. Some of which can be and have been tested by more complex multivariate analyses. George S cited latitude as a confound, but in fact latitude is part of any theory about evolutionary shaped IQ. It should be correlated, but different underlying causal structures produce different patterns of covariance, which is why competing causal models can be tested using multivariate analyses. The theory that latitude shaped IQ which in turn is the more proximal influence of GDP predicts that when all 3 variables are in the same analyses, IQ will have more unique variance than latitude that overlaps with GDP. The theory that latitude variables more directly impact GDP and IQ is only indirectly related via correlation with latitude predicts the opposite.
The results I linked show support for the former and contradict the latter, thus favoring a causal model with a more direct IQ-GDP association.

No, that one analysis does not deal with all possible causal model and I never implied it did. In fact, my other post cited an article discussing the need for more sophisticated analytic strategies and longitudinal data in order to tease causality apart, especially the reverse causality you mention.

 

[ronburgundy]

Let me preface by saying that I agree that it is highly likely that much of the between region variance in that map is due to education.

That said, there is no need to isolate "intelligence" from "education". Education is a perfectly plausible causal impact on intelligence, even if intelligence reflects a set fundamental cognitive skills that generally impact learning and problem solving, and is remains stable across the lifespan after a stage of early development.
The non-straw man conception of general intelligence that is studied and accepted as scientifically useful by most cognitive scientists allows for influence from education, and other environmental factor from nutrition to lead exposure, especially in early childhood.

Also, why would you expect a "much smaller variance" if the test were valid? It should vary as much as the thing it is measuring varies, given its level of sensitivity to small variations. Nothing about the amount of observed variance speaks to its validity.
The skew of the variance is a separate issue and does potentially speak to its validity, but maybe not. Not all variables are normally distributed. In fact, evolution often creates non-normally distributed traits due to different selection pressures on different subsets. Also, the map doesn't even show a non-normal distribution at the aggregate level, only that different levels of the distribution are non-randomly dispersed by location, which no reasonable theory of intelligence would predict it should be, given that it would impact how, where, and who migrates, and differential survival of different intelligence levels depending upon geographic location and migratory trajectory to wind up there.

The result that would most undermine the validity of IQ would be if, against all odds, something tied to reasoning and general problem solving wound up randomly distributed across all geographic areas.

The actual specific pattern of how IQ is distributed may well be more in line with effects of education, but neither the variance, its skew, nor it being non-randomly distributed are problematic for its validity, and the latter supports it. Also, if education winds up being the causal factor for the geographic pattern rather than any genetic variance that isn't even a problem for the validity of IQ or the concept of general intelligence. It would only mean that the kind of major educational differences that exist between cultures with centuries of pervasive differences are sufficient to impact whatever the test measures. Only a strawman theory of 100% genetic determinism would be invalidated by that. Education impacts very fundamental cognitive skills that in turn impact lifelong learning across many contexts, some of which are also impacted by genetic factors (which vary more within than between geographic groups). IQ is only claimed to measure such general cognitive skills, so if education is among the factors that influence them, then there is no problem. Also, education can plausibly have massive impact early in life that cannot be undone by later experiences. Thus, even the notion that these skill get set early then remain stable constraints is left untouched by the potential impact of education on cross cultural aggregate differences.

The question is: what is an IQ test actually measuring?

I doubt that's clear, but I'll just go with the words I find and go from there. From wikipedia:

An intelligence quotient (IQ) is a score derived from one of several standardized tests designed to assess human intelligence.

So what the hell is 'intelligence'?

Let's go with dictionary.com

capacity for learning, reasoning, understanding, and similar forms of mental activity; aptitude in grasping truths, relationships, facts, meanings, etc.

If you're going to build a test that measures someone's 'capacity for learning and reasoning' and you don't isolate all other variables, including education, income level, etc, etc, then you're not really measuring someone's 'capacity for learning', you're actually measuring their capacity for learning multiplied by environmental factors. I assume that's what's happening because I wouldn't expect as high as a 30% variance based on genetics alone, and I certainly wouldn't expect an economically poor geographic area to be consistently low-scoring.

In other words, someone who scores a 95 in Sudan might actually be a 120 in Canada given better economic conditions. If you don't account for those factors which directly affect how a person interacts with the IQ test itself, then you're not measuring intelligence as defined as 'capacity to learn and reason'.

OTOH, you probably ARE measuring people's 'capacity to learn' based on genetics AND environmental factors, so the map gives us a pretty good idea that education and infrastructure actually makes people smarter.

And in IQ's defense, it is still a useful tool when you localize your sample to a smaller variable set. For instance, if you were to test only people in Ontario, one person's 90 to another person's 120 might mean a little more.

To be clear, nothing you say above supports your prior claim that the amount of variance in IQ or the skew in its distribution is a problem for the validity of the measure or the construct.

Its hard to tell, but you seem to be still presuming that the definition of "capacity to learn and reason" assume an innate genetically determined capacity. That isn't true.
Whatever the current state of a physical brain is impacts how efficiently it will change (i.e., "learn") in response to any given future amount of information or "education", and how well it can perform various tasks. That current brain state is a product of interactions between genetics (possibly impacting things like neural plasticity, axon transmission speeds, the size and interconnectedness of various structures, etc..), environmental factors that impact brain development including those that impact brain biology in rather permanent ways directly (e.g., in utero and early childhood nutrition) or indirectly (e.g., early childhood cognitive experiences during times of high brain plasticity), and those that impact brain states in more transitory ways (e.g., exposure to specific information that impacts current performance on tasks specific to that information, like taking a high school chemistry class and thus doing better on a chemistry test).

Valid General intelligence tests have variance that is impacted by all these, except for latter transitory and information/task specific factors in bold. Because they are directly tied to exposure to particular information and tasks, they do not reflect a capacity to learn but rather what you did learn due to whether you were exposured to that info. The reason that the Raven's Matrices have become the gold standard of IQ tests is that it presents a task that is not trained in formal education, does not depend on language, and yet involves cognitive processes that are so basic, they likely play a role in learning from countless types of information in school and the real world.
It requires that you be able to extract regularities and variance in complex stimuli, be able to hold distinct arrays of information in memory at the same time, and then reason a prediction of which pattern out of several choices covaries in a similar multi-dimensional way.

People perform these basic operations all the time in daily life and they underlie much of what is needed to learn in school. Yet, the particular task and puzzles in the Ravens are not taught or practiced. Thus, how well people do is not about how much exposure they had to anything specific to that task, but how well they can execute those basic cognitive operations that the task requires which is common to reasoning and learning for complex information more generally.

It is easy to see why people assume that this is something innate, because we tend to think of experiences as only giving us specific types of information. But experiences change the brain, and some experiences (especially those that directly impact brain biology or occur early in life) can shape the brain in ways that have life long causal impact on executing these kind of basic mental operations.

In addition, the impact of genes is not likely to be direct, but rather create a biological context that moderates how experiences shape learning. For example, any genes that impact brain plasticity is not directly impacting learning, but rather is impacting how the brain changes in response to experiences. Thus, a kid with a highly "plastic" brain kept in dark box isn't able to take advantage of that genetic advantage and can emerge with lower general intelligence than a kid with average plasticity but optimal experiences. But the key is that those interacting factors play a role in early brain development that then becomes "set" in a way that it cannot be undone and has lifelong impact on reasoning and learning across many domains. With twice the practice the lower IQ person may reach the same level of understanding, but it is unlikely that they can change the fact that it takes them twice the practice.

The biggest geography differences in that map are probably much more about the environmental factors that impact brain development prior to formal education, than they are about what kids are taught in school.

 

[ronburgundy]

And the purpose of theories of racial intelligence was to rationalize slavery and mistreatment of non-whites. It unscientific purpose has nothing to do with what is true and what the measure reflects. Nothing about how g is measured makes it a measure of a cause rather than an effect. In fact, g is nothing but a statistical outcome on various tests, thus it is only a direct measure of effects and not causes. It is a measure of the common effect of some yet to be clarified cognitive skills on the outcome performance of various cognitive tasks. The question is what are those underlying skills and what impacts their development?, and how does the relative importance of those factors change depending upon whether your talking about variance within groups that share similar environments versus between groups, who often by their very definition as groups differ between groups in their environments more systematically than people within each group does?.
You repeatedly draw conclusions for the last question, claiming, that its genetics and that impact does not differ for various types of group comparisons, even though none of your data ever even speaks to the question.

The article you cited contains this paragraph, briefly laying out the case:

Does higher intelligence beget better educational outcomes?

In longitudinal studies that measure psychometric intelligence first and educational attainments later (thus assessing that causal chain), there is a moderate to strong correlation between the two, as assessed by years spent in full-time education, the highest qualification obtained by a person or the scores obtained on educational assessments.⁵ [Jencks C . Who Gets Ahead?: The Determinants of Economic Success in America. New York, NY: Basic Books; 1979.] For example, in a study of approximately 70 000 children in the UK, the general factor from the Cognitive Abilities Test (CAT) battery taken at age 11 years correlated about 0.8 with the general factor of grades on the General Certificate of Secondary Education (GCSE) examinations taken at age 16 years.⁶ [Deary IJ, Strand S, Smith P, Fernandes C . Intelligence and educational achievement. Intelligence 2007;35:13-21.] The general factor of the CAT test had very similar loadings from the three domains of verbal, non-verbal (abstract) and quantitative reasoning. Older studies have reported correlations ranging from 0.60 to 0.96.^7–9 [Bouchard TJ . Twins reared together: what they tell us about human diversity. In: Fox SF, editor. Individuality and Determinism. New York: Plenum Publishing Corp; 1984. p. 147-84.; Kemp L . Environmental and other characteristics determining attainments in primary schools. Brit J Edu Psychol 1955;25:67-77.; Wiseman S . Environmental and innate factors and educational attainment. In: Meade J, Parks AS, editors. Genetic and Environmental Factors in Human Ability. London: Oliver and Boyd; 1966. p. 64-79.] The conclusion from such studies might be that intelligence has stronger causal effects on educational results than vice versa.​

Note the very important and explicit qualifier in what you quoted. It said, "might" because that conclusion only seems to be supported by a superficial analysis of only the limited subset of studies reviewed in that paragraph. The very next paragraph lays out the case for causal impact of education on IQ.

"The conclusion from such studies might be that education influences the development of intelligence. However, this requires the caveat that the so-called ‘intelligence tests’ should be scrutinized to examine the extent to which they contain materials that appear in the taught curriculum.

So, it is possible that intelligence causes differences in educational outcomes, or that education causes intelligence differences, or a bit of both. Indeed, it is probably more complex than this. Readers can find further detailed consideration of possible non-linear effects of schooling on mental test scores, and the parts played by measurement error in intelligence and education measurement in a rather technical paper by Hansen et al."​

The rest of the article goes onto to point out the analytic flaws and assumptions underlying epidemiological work claiming to show evidence of either causal direction, and how more and different types of data and analyses are requires to disentangle the mess of potential causal models capable of producing the non-experimental, correlational data inherent to all of these studies.

The idea that you are approaching this topic objectively and non-politically is undermined by your cherry-picking of the one paragraph in an 8 page paper that generally refutes your position that one causal direction is "known" while the other is "only speculation".

This evidence should be considered on top of the foundational knowledge that intelligence variations are mostly heritable, a point known from twin studies and a point also acknowledged by that article, implying: if education can change IQ, there is a small upper limit of that effect due to the much larger influence of genetic variations.

IOW, the centerpiece of your argument is the statistical fallacy that underlies almost all your arguments about IQ, namely your assumption that whatever determines the majority of variance at the individual level must be what is responsible for group-level differences. I have explained to you multiple times why this is false, so I won't bother with it again. But besides that, your "mostly heritable" claim is not true. The article cites that the estimate for genetics on childhood IQ is "< 50%", which means it is mostly environmental during the childhood years where education is occurring and where IQ and both amount of level and quality of education would be influencing each other. Regardless, "mostly" is irrelevant, because their is more than enough variance for genes and environment for either to be the primary determinant of group-level differences, which need not be and often are not due to the same proportion of causes responsible with within group or overall variance.

ronburgandy, you asked a question that was seemingly NOT about group differences, and I answered the inquiry assuming you were inquiring about within-group variation, not between-group variation. I am not conflating those two patterns. What did you mean to ask about?

Even in terms of within group variance, your claim is refuted by everything the article, outside of the 1 paragraph you picked. The causal direction of the IQ-education relation is far from established, and there is evidence for both directions, and most of it is limited in allowing strong conclusions.
And that is true even for within group relationships where there is much less environmental variance in educational opportunities. Within a culture, amount of education is heavily impacted by one's demonstrated aptitude. Between culture differences are less so, and are far more determined by cultural differences in resources and ideological value placed on education. This not only opens the door, but tears it off its hinges to allow for far greater impact of differences in educational opportunities on cognitive development of the sort that impacts IQ. That doesn't even factor all the massive environmental impacts on physical brain development outside of formal education.

The comment about between group comparisons refers to your OP and the thread in general, where you use purely within-group analyses of explanatory variance to support claims the between group differences have the same underlying factors causing then in the same ratios. You do that in nearly every thread.

 

[bilby]

I'd be careful if I were you; Every single person who believes that correlation implies causation is going to die.

GDP may be a strong predictor of IQ; after all, education ain't free, and poverty tends to lead to less time in school - children in poverty have to work, rather than study.

I don't assume correlation = causation, but I am not so scientifically illiterate that I think that such a vacuous mantra is a valid intellectual reply to data that shows more than just simple correlations.
I specifically mentioned alternative causal possibilities. Some of which can be and have been tested by more complex multivariate analyses. George S cited latitude as a confound, but in fact latitude is part of any theory about evolutionary shaped IQ. It should be correlated, but different underlying causal structures produce different patterns of covariance, which is why competing causal models can be tested using multivariate analyses. The theory that latitude shaped IQ which in turn is the more proximal influence of GDP predicts that when all 3 variables are in the same analyses, IQ will have more unique variance than latitude that overlaps with GDP. The theory that latitude variables more directly impact GDP and IQ is only indirectly related via correlation with latitude predicts the opposite.
The results I linked show support for the former and contradict the latter, thus favoring a causal model with a more direct IQ-GDP association.

That's an awful lot of long words that don't really say much except that you don't like me suggesting that you might be wrong.

No, that one analysis does not deal with all possible causal model and I never implied it did. In fact, my other post cited an article discussing the need for more sophisticated analytic strategies and longitudinal data in order to tease causality apart, especially the reverse causality you mention.

...and then another lot of big words that admits that I was correct to point out that you might be wrong.

Perhaps you have a mistaken belief that 'a valid intellectual reply' must be verbose and sesquipedalian?

 

[ronburgundy]

I don't assume correlation = causation, but I am not so scientifically illiterate that I think that such a vacuous mantra is a valid intellectual reply to data that shows more than just simple correlations.
I specifically mentioned alternative causal possibilities. Some of which can be and have been tested by more complex multivariate analyses. George S cited latitude as a confound, but in fact latitude is part of any theory about evolutionary shaped IQ. It should be correlated, but different underlying causal structures produce different patterns of covariance, which is why competing causal models can be tested using multivariate analyses. The theory that latitude shaped IQ which in turn is the more proximal influence of GDP predicts that when all 3 variables are in the same analyses, IQ will have more unique variance than latitude that overlaps with GDP. The theory that latitude variables more directly impact GDP and IQ is only indirectly related via correlation with latitude predicts the opposite.
The results I linked show support for the former and contradict the latter, thus favoring a causal model with a more direct IQ-GDP association.

That's an awful lot of long words that don't really say much except that you don't like me suggesting that you might be wrong.

Your comment didn't contradict anything I said, so it cannot possibly suggest that I am wrong. You inability to understand anything beyond vacuous platitudes is not my failing.

No, that one analysis does not deal with all possible causal model and I never implied it did. In fact, my other post cited an article discussing the need for more sophisticated analytic strategies and longitudinal data in order to tease causality apart, especially the reverse causality you mention.

...and then another lot of big words that admits that I was correct to point out that you might be wrong.

Again, I never claimed anything that is contradicted by your comment. It is your inability to understand English combined with scientific illiteracy that leads you to that false inference.

Perhaps you have a mistaken belief that 'a valid intellectual reply' must be verbose and sesquipedalian?

No, I think it should have actual intellectual content and a rational argument, which yours typically do not.

 

[fromderinside]

That's an awful lot of long words that don't really say much except that you don't like me suggesting that you might be wrong.

Your comment didn't contradict anything I said, so it cannot possibly suggest that I am wrong. You inability to understand anything beyond vacuous platitudes is not my failing.

No, that one analysis does not deal with all possible causal model and I never implied it did. In fact, my other post cited an article discussing the need for more sophisticated analytic strategies and longitudinal data in order to tease causality apart, especially the reverse causality you mention.

...and then another lot of big words that admits that I was correct to point out that you might be wrong.

Again, I never claimed anything that is contradicted by your comment. It is your inability to understand English combined with scientific illiteracy that leads you to that false inference.

Perhaps you have a mistaken belief that 'a valid intellectual reply' must be verbose and sesquipedalian?

No, I think it should have actual intellectual content and a rational argument, which yours typically do not.

How rational can an argument be when available resources are nearer the equator and fitness stress is nearer the poles.

 

[Loren Pechtel]

Your comment didn't contradict anything I said, so it cannot possibly suggest that I am wrong. You inability to understand anything beyond vacuous platitudes is not my failing.

No, that one analysis does not deal with all possible causal model and I never implied it did. In fact, my other post cited an article discussing the need for more sophisticated analytic strategies and longitudinal data in order to tease causality apart, especially the reverse causality you mention.

...and then another lot of big words that admits that I was correct to point out that you might be wrong.

Again, I never claimed anything that is contradicted by your comment. It is your inability to understand English combined with scientific illiteracy that leads you to that false inference.

Perhaps you have a mistaken belief that 'a valid intellectual reply' must be verbose and sesquipedalian?

No, I think it should have actual intellectual content and a rational argument, which yours typically do not.

How rational can an argument be when available resources are nearer the equator and fitness stress is nearer the poles.

Diseases are near the equator.

 

[fromderinside]

Your comment didn't contradict anything I said, so it cannot possibly suggest that I am wrong. You inability to understand anything beyond vacuous platitudes is not my failing.

No, that one analysis does not deal with all possible causal model and I never implied it did. In fact, my other post cited an article discussing the need for more sophisticated analytic strategies and longitudinal data in order to tease causality apart, especially the reverse causality you mention.

...and then another lot of big words that admits that I was correct to point out that you might be wrong.

Again, I never claimed anything that is contradicted by your comment. It is your inability to understand English combined with scientific illiteracy that leads you to that false inference.

Perhaps you have a mistaken belief that 'a valid intellectual reply' must be verbose and sesquipedalian?

No, I think it should have actual intellectual content and a rational argument, which yours typically do not.

How rational can an argument be when available resources are nearer the equator and fitness stress is nearer the poles.

Diseases are near the equator.

Yeah, and starvation is near the poles. The difference being that withdrawing energy from a systems is different from adding energy to a system when it comes to fitness.

 

[Loren Pechtel]

Diseases are near the equator.

Yeah, and starvation is near the poles. The difference being that withdrawing energy from a systems is different from adding energy to a system when it comes to fitness.

Near the equator the population limit is disease. Near the poles the population limit is the food supply.

 

[ronburgundy]

Yeah, and starvation is near the poles. The difference being that withdrawing energy from a systems is different from adding energy to a system when it comes to fitness.

Near the equator the population limit is disease. Near the poles the population limit is the food supply.

For the record, my arguments acknowledge the role of disease and of limited resources (not just food but shelter). This issue is irrelevant to what bibly and I were arguing about, which is whether or not I or the research I linked to made a causal conclusion from a simple bi-variate correlation (we didn't).

BTW, the issue of selection pressures goes beyond those in a particular locations. Migration across regions with vastly differing selection pressures, adds a inter-regional selection pressure of needing high ability to quickly understand the variation is local factors and to solve the problem of changing one's previous learned actions to fit the new situation without having to rely upon trial and error where you'll be dead before you can figure it out. Migration across climates increases reliance upon efficient abilities to abstract and transfer deeper relationships and principles to seemingly disparate contexts.
Thus, the selection pressure comparison is NOT between living in Norway vs. Africa. It is between those that went from Africa to the Middle East to SE Asia to Central Asia to Eastern then Northern Europe compared to those who stayed close to their African origins. Also, it isn't about simple degree of selection pressures. It is about the specific nature of those pressures, if and how one could overcome and deal with them, and what traits would facilitate that. For example, diseases that quickly wipe out people and have no cure may select mostly for particular factors related to the immune system having zero to do with cognition. Whereas, a climate that offers limited food and shelter unless you can figure out how to find it, get to it, or build it via completely novel means might select more for solving novel problems via analogical reasoning.
It isn't how hard it is to survive, but what factors impact the probability of survival that matters most for changes in the distribution of various traits in a sub-population.

Again, I am not claiming that this is the explanation for the IQ variance in the OP map. I'm just clarifying what a valid evolutionary account of part of that variance would generally look like, and why, in principle, it is quite plausible for the skills tapped by IQ tests to vary with geographic region, due to the vast differences in ancestral trajectories between the people who currently occupy those regions.

Current environment impacting post-conception brain development, of which formal education is but a minor contributor, could also be a reason for these differences in the OP map. The problem is that the ideologues on one side ignore the data that shows the inadequacy of some of these environmental accounts (such as that latitude impacts diseases that hinder brain development), while ideologues on the other side don't give environmental accounts a fair shake and wrongly infer that the genetic contributor to within-region IQ variance strongly favors genes as the major factor in between region differences.

 

[fromderinside]

Biology is my gig. Thermodynamics is my instrument.