Childhood intelligence in relation to major causes of death in 68 year follow-up | Hacker News Reader

Childhood intelligence in relation to major causes of death in 68 year follow-up

bmj.com

114 pointsgwern70 comments

For a community that likes to point out all the time that correlation is not causation, a lot of the posts so far seem to be making assumptions about causative direction.

But why should we assume that just because they're correlated in this study, high IQ causes longer life expectancy?

Maybe the same early childhood circumstances (such as better nutrition) that led to measurably higher IQ also led to individuals being healthier / more robust later in life.

Maybe exposure to certain toxins that decrease IQ (say, lead) also tends to make certain cancers more likely.

Well adoption studies show that environment doesn't matter much to IQ. Kids raised in the poorest families didn't have higher IQs than kids raised in the richest families (http://sci-hub.la/10.1007/s10519-007-9142-7). Adopted siblings have little correlation in IQ (https://www.ncbi.nlm.nih.gov/pubmed/21807642). And intervention programs to increase IQ in disadvantaged populations always fail (https://www.sciencedirect.com/science/article/pii/S016028961...).

Lung cancer is one of the biggest correlation found here. And smoking is pretty genetic (https://www.ncbi.nlm.nih.gov/pubmed/15971021). Other things might be affected by diet and BMI, but that's also highly genetic.

This study also controlled for socioeconomic status.

> Well adoption studies show that environment doesn't matter much to IQ.

Really? I remember a few adoption study where they found the opposite: They measured the IQs of genetically identical twins that were adopted into different families, and found that family environment mattered a hell of a lot.

I can't find the study that I remember, but here are some adjacent studies that have done similar things with similar results:

http://sciencenordic.com/adopted-children-have-higher-iqs-th...

https://www.ncbi.nlm.nih.gov/pubmed/6872626

http://www.pnas.org/content/112/15/4612

That's interesting. I don't think half siblings should be considered a representative random sample though. It may simply be that babies adopted away were more likely to be from low IQ fathers.

The ncbi link seems to support my hypothesis, maybe I'm missing something.

Yes, there are publications with support both ways. IQ is basically meaningless and it pains me to see it constantly brought up here.

People who need to believe in some Randian meritocracy require cheap and easy metrics to convince themselves and sort people. If you can’t feel confident in your ability to discriminate, then you might have to question the foundations of your ideology, which is painful for many.

How could there possibly not be a thing like general intelligence?

Please don’t put words in my mouth. I’m skeptical of our ability to accurately define intelligence, never mind accurately measure it (especially with a one-size-fits-all test), but I don’t doubt its existence.

Although some people do make me wonder...

I just love the armchair experts who come out of the woodwork whenever this topic comes up. I'm sure you can accurately psychoanalyze the hidden motives. Of a subject you haven't even read the wikipedia page on.

Your middle link there refers to the Minnesota Transracial Adoption Study [1]. That brief seems to assume the reader is familiar with the studies as otherwise it would be very easy to misunderstand what they're saying - which is not what you're saying.

Their study was a large undertaking studying the differences in black/interracial/white children who were all adopted by wealthy white families in Minnesota. It found that at age 7 the difference in IQ between a adopted white child and adopted black child was 20.1 points. At age 17 this changed to 17.8 points. The difference for a half black child and a white child was 6.1 at age 7, and 8.3 at age 17.

The goal of the study was to show the opposite of what it ended up showing. Consequently many individuals, including the authors of the study themselves, have tried to argue that this is not evidence of a genetic link to IQ. Arguments against it have run the gamut from environmental factors from before the children were adopted (which was at less than 1 years old in all cases) to skin color causing environmental effects. These are certainly contributing factors, however they would undermine any and all results of the study - which is indicative of an after-the-fact effort to undermine their own results. In either case, that result certainly showed nothing like "family environment mattered a hell of alot."

The misunderstanding here probably comes from one questionably phrased sentence in that abstract, "Black and interracial children scored as well on IQ tests as adoptees in other studies." That is more clearly read as "Black and interracial children scored as well on IQ tests as black and interracial adoptees in other studies." In other words, they're comparing differences between 'same group' individuals and trying to see what determines the variance there. And they found that at a young age this is frequently related to environmental factors, whereas at older age genetics becomes dominant even among 'same 'group individuals. In the abstract's rather kinder phrasing, "Our interpretation of these results is that younger children are more influenced by differences among their family environments than older adolescents, who are freer to seek their own niches.".

I seem to be unable to edit my post. The link was to: https://en.wikipedia.org/wiki/Minnesota_Transracial_Adoption...

The study itself is paywalled. And given the decades of controversy it sparked, the Wiki entry is quite expansive.

Paywalls are just s state of mind.

SciHub p, non-paywalled link: https://dacemirror.sci-hub.la/journal-article/c16af30ee6a2c0...

your first link is broken

Science may be blocked in your country.

I think you should update your reference baseline. Start with looking for papers who cite those papers, see what they think and what they found.

For one, with recent advances in genomics, we now realise that conclusions from twin studies in general aren't aging so well. And all the above don't really do genetics, they used seemingly close-knitted cohorts.

In themselves I find all the papers you pointed dubious at best, and more likely garbage. The experimental setups in all of these were rather weak to start with, and did not allow to discriminate between causation and correlation, and the authors all achieved/generalized conclusions that were not supported by their own data.

Our societies fascination and obstination with deriving things from IQ is, well, fascinating... No of these start with asking a question to understand a tractable and useful problem. They're all a my "d!ck is bigger than your d!ck" kind of ranking problem.

This are quite unspecific criticisms. IQ is very well researched and widely supported, as are twin studies.

> Participants 33 536 men and 32 229 women who were participants in the Scottish Mental Survey of 1947 (SMS1947) and who could be linked to cause of death data up to December 2015.

> In this prospective cohort study, all individuals born in Scotland in 1936 and registered at school in Scotland in 1947 were targeted for tracing and subsequent data linkage to death certificates

Hard to get a more comprehensive population study than that.

Interesting.

I wonder whether the Flynn effect (increase in IQ in the general population over time [0]) can therefore explain at least some of the increase in longevity in the 20th century.

[0]: https://en.wikipedia.org/wiki/Flynn_effect

Not directly, but it is quite possible the Flynn effect and the increase in longevity have a common cause.

Shame the Flynn effect ended a few decades ago.

Do you have a source on the Flynn effect ending?

Wiki has a whole section on it [0].

1. https://en.wikipedia.org/wiki/Flynn_effect

Thanks, never saw that before.

This kind of linear thinking is too simplistic. Health will boost intelligence and vice versa in a non-linear fashion, if I am any judge, so an extrinsic impuls doesn't preclude intrinsic causation.

Did they control for economic class? If all the smarter people are richer, it would follow that they lived healthier lives with better access to medical care.

Yes. From the linked page:

> Sensitivity analyses on a representative subsample of the cohort observed only small attenuation of the estimated effect of intelligence (by 10-26%) after adjustment for potential confounders, including three indicators of childhood socioeconomic status. In a replication sample from Scotland, in a similar birth year cohort and follow-up period, smoking and adult socioeconomic status partially attenuated (by 16-58%) the association of intelligence with outcome rates.

If you control for positively correlated variables, the results are more dramatic right? Higher intelligence leads to lower death And higher intelligence to higher income then lower death. So actuarial effects multiply.

My first question too:

> In the absence of individual level data on socioeconomic status in the full SMS1947, we used school attended as a proxy measure to adjust for potential confounding by background socioeconomic status. A Scottish cohort study has shown that primary school has moderate correlation with paternal social class.

All participants lived in a society with socialized healthcare built around a “free at the point of service” model. That’s not to say that socioeconomic status doesn’t play a part (as other commenters have noted) but the difference in access to healthcare due to wealth is relatively small in the UK in general

There are still wide health inequalities in the UK.

Poor people eat worse food; don't get as much exercise; and smoke more.

You can predict average life expectancy by post-code, and that's based on socio-economic factors.

> Poor people eat worse food; don't get as much exercise; and smoke more

How is choosing to smoke more an inequality?

I suspect because if your peer group smoke it is much more difficult to choose to not to smoke?

That’s the kind of thing that population health researchers study.

Answers include availability, lack of availability of healthier alternatives, targeted marketing, being missed by anti-smoking campaigns, and peer group influences.

While smoking may be a personal choice, it is not made in a vacuum.

Smoking is “reward now, pay later”

http://journals.plos.org/plosone/article?id=10.1371/journal....

There are pretty big differences in life expectancy here in Scotland - which given that it is heavily location based suggests that "socioeconomic status" plays a huge role.

"Men in the least deprived areas of Scotland may live 12.5 years longer than those in the most deprived areas while women in the least deprived areas could expect to live 8.5 years longer than those in the most deprived."

http://www.bbc.co.uk/news/uk-scotland-scotland-politics-2964...

Edit: Interactive visualization of deprivation in Scotland: http://simd.scot/

False. All participants were born and took their IQ tests a year before the very first elements of the NHS came into existence.

And health (and all other) outcomes are strongly affected in the UK by geography. On average you'll live in Sussex nearly a decade longer than in Cowdenbeath. In Glasgow as late as the mid-1980s the average male life expectancy was 58 years, and then Scotland had the shit kicked out of it, on purpose, for twelve years, by the 1979 Thatcher government. Cuts in healthcare were inflicted disproportionately-severely on places which did not vote for the Conservatives, and nowhere were they more severe than central Scotland and south Wales.

False? The subjects have lived >75% of their lives with national health. And the older you are the more you use health services. While it may be the least good in Scotland, it’s enormously better than being uninsured in the US where people are understandably reluctant to visit the doctor’s office at all.

Nothing you say contradicts the parent’s post. So ‘false’ is incorrect as well as obnoxious.

It would seem reasonable to suppose that much of this is related to the known association between delayed gratification (self control) and intelligence (the marshmallow experiment described as for instance in https://jamesclear.com/delayed-gratification)?

And yet intelligent people are said to use more drugs:

https://www.psychologytoday.com/blog/the-scientific-fundamen...

It's not clear if they controlled for other factors. It may be that they simply have more money to buy them.

Drugs are not that expensive, but not everyone is so curious to try them.

Figure 4⇓ shows the associations between groupings of childhood intelligence score (10ths or quarters) and deaths related to 15 specific cancers. About half of these showed inverse patterns of association with a degree of linearity, including cancers of the oesophagus, colon or rectum, stomach, liver, lung, kidney, bladder, and blood. The strongest association was evident for death related to lung cancer: the risk in the highest performing 10th of childhood intelligence was reduced by two thirds compared with the lowest performing 10th. Cancers showing negligible or irregular associations with childhood mental ability included mouth, pancreas, skin, ovaries, breast (women only), prostate (men only), and brain or central nervous system.

The problem is higher IQ mean means a healthier lifestyle, which is what protects against cancer, not genes that give rise to high IQ itself. That's why the effect is strongest for smoking, because less intelligent people may be more likely to smoke. Same for drinking, which increases odds of stomach and esophagus and liver cancer. Somewhat disappointing becase it confimrs that we have already suspected, whic his a helathier lifestyle means less risk of certain cancers. Not surprisingly, the effect is non-existent for brain cancers.

>The problem is higher IQ mean means a healthier lifestyle, which is what protects against cancer, not genes that give rise to high IQ itself.

Based on what, exactly?

I assume he's talking about a correlation, and probably not strictly limited to a healthier lifestyle, but also to higher income (on average), which provides potentially better access to healthcare, etc.

I think you can see the difference in these effects by the difference in deaths from smoking vs other cancers.

So the reduction in - for instance - prostate cancer may be due to access to healthcare vs reduction in lunch cancer which might be a proxy for “smarter” decisions.

>The problem is higher IQ mean means a healthier lifestyle, which is what protects against cancer, not genes that give rise to high IQ itself.

This is not the case, see _The association between intelligence and lifespan is mostly genetic_ (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4795559/). The common factor is almost certainly mutational load.

Or you could say that smarter people don't smoke.

You could also say that smoking is more popular among poorer populations, the depressed, those whose parents smoked, etc.

During the timeframe of this study, not smoking would have been less common.

Or maybe higher IQ means lower likelihood of having to work as a coal miner (or in other occupations that expose the worker to similar types of pollution and carcinogens)?

Here's my hot take: genes that cause a given nervous system to be better than typical at making decisions that lead to better outcomes for the organism will cause the organism to be better than typical at cell and DNA repair.

Is there a way to express the figures in a more tangible way? Something like: the smartest 10% lived N months longer than the dumbest 10%.

And with respect to specific cause of death, does a hazard ratio of 0.72 for respiratory disease mean that respiratory disease killed 72 of the smartest 10% each year for 100 of the dumbest 10%?

I can’t find any evidence of whether they controlled for obesity (which is certainly believed to be a factor is several of the diseases that a correlation was found for). But other than that, I’d say that it is a very interesting study.

They can't as they were looking at death records.

Low adolescent IQ is a strong predictor for obesity later in life. [1] So I'm not sure how or why you'd want to control for obesity. You could be looking at a comorbid effect. Low IQ -> obesity -> early death.

[1] - http://circ.ahajournals.org/content/124/Suppl_21/A12601.shor...

Interesting that suicide had a weak correlation. Intelligence obviously has little bearing on depression

Or intelligent people may be more likely to apply cognitive techniques (perhaps they experience depression earlier and/or more commonly which may build resilience) and/or may be more likely to seek professional advice (e.g. because they see the value, move in circles in which it is more acceptable, or make more money and can afford to).

How is that obvious?

I'm not sure how you got to 'obviously'. Could you elaborate?

Does "weak correlation" not mean that there was very little difference either way? I just meant this study seems to show that suicide is unaffected by intelligence. Either I read it wrong or I worded my content badly but I was just trying to draw a conclusion from what I thought the data said

"Suicide" is simply too rare, and too broad a category. Is not using a seat belt suicide? Is a heroin overdose? Is standing on a cliff to estimate what it would be like to jump and loosing ballance at the thought of it?

How many suicides were there in the 60000 studied, perhaps 10? That's not enough to derive a meaningful result, I think.

Edit: My estimate would be roughly correct, sad enough, ignoring my doubts on suicide as a cultural rather than medical category: http://news.bbc.co.uk/2/hi/uk_news/scotland/5297916.stm

There is so much more to this. Inequality, poverty..

Poverty hinders proper brain development and wiring. Life expectance in London is 25 years less for the poor! Economy rules! It is not popular info, but for those interested: https://youtu.be/GvkchZADaaA

I'd love to see the relation between IQ and happiness.

I imagine it would be a negative one!

Possibly, but I'm not so sure. If you're smart, you will make more money, and yes of course money doesn't make you actually happy etc etc, but studies have shown it actually does up to a point (or, at least, money gets rid of things which would otherwise make you unhappy, such as medical care, lack of new things, etc). It would certainly be interesting to see a study on this.

On the other hand:

http://www.independent.co.uk/news/health/intelligence-mental...

>If you're smart, you will make more money //

On average, just being 'smart' isn't sufficient.

I would be surprised if there's not a correlation between intelligence and average salary. Is there a study on this?

There may not be one, but a high IQ can enable you to have a more comfortable misery.

Have a look at this, albeit tongue in cheek, survey by card against humanity.

https://thepulseofthenation.com

One of he questions was ‘would you rather be dumb and happy or smart and sad’ and then the correlate it with voting for trump.

I can't imagine wanting to be dumb. However, if I got really depressed, then things may change. Another factor in this is wealth. Does dumb mean poor and smart rich?

Reply on news.ycombinator.com