There are two meaningful controls, which combined hint at a causation. The IQ test from age 13, before many people begin pot use, is a control over who makes the decision to start smoking weed, and the non-smoker and smoker groups provide a control over whether pot made people stupider.
So, what factors could cause the study to show the correlation but not be a direct causation? It would have to be a confounding factor that both makes people want to smoke and something that causes them to get stupider over time that isn't the marijuana. That's possible, perhaps some sort of addictive personality disorder. But I think the simplest explanation is a causation, there is other evidence that marijuana affects the brains of adolescents.
No, I meant real, scientific controls. A study with real experimental and control groups would have groups that the researchers told that they would smoke, or not smoke, pot. That would make the study prospective, not retrospective as in this study.
If a study must monitor behavior that the subjects choose for themselves, the study loses all value, because the result can never be more than an uncontrolled correlation between A (IQ) and B (the choice to smoke pot) and any number of possible correlated factors (C) like socioeconomic, genetic, and other uncontrolled issues.
This is one of the common threads that run through most of these studies, which is why I linked the article -- it's typical of studies that I've been reading for about 35 years, and all of which suffer from the same systematic flaws.
Obviously a real, disciplined study of this topic would assign behaviors in a double-blind study. But such a study would clearly be unethical, which is why it has never been conducted.
> So, what factors could cause the study to show the correlation but not be a direct causation?
The fact that the subjects make their own decisions, the fact that there is no control group as that term is understood.
> It would have to be a confounding factor that both makes people want to smoke and something that causes them to get stupider over time that isn't the marijuana.
Yes, and until we think of a way to study this topic scientifically and without violating ethical guidelines, we will never know. It could be socioeconomic, genetic, peer pressure (smart people are geeks, don't you know that?) or some other factor.
> But I think the simplest explanation is a causation ...
That isn't an explanation unless we study it scientifically, this study doesn't do that.
This is what I find troubling about psychology studies, studies that involve monitoring behaviors but with no meaningful controls -- people don't seem to understand that it isn't science.
> ... there is other evidence that marijuana affects the brains of adolescents.
Yes - and all such studies have the property this study has -- no scientific controls.
This is in no way meant to condone drug use. All I am saying is public policy is being shaped by something other than science.
>If a study must monitor behavior that the subjects choose for themselves, the study loses all value
You'll almost never find medical studies over long periods of time like this, and you'll never find a truly causal study on complex human behavior; there literally is no medical research that can properly do this. This is an epidemiological study, and these kinds of studies DO have value and certainly are scientific. This is the field that "proved" cigarettes are bad for us and that red meat increases the chances of heart disease.
I agree with you that a causal result would be great, but I think we both agree it's not possible for something like this. My point is that they make a very convincing case that shouldn't be dismissed so easily - correlations should be considered for what they are, not thrown out without thought. Your principled view on this works for some sciences but is too conservative for the kinds of things epidemiology studies.
>> If a study must monitor behavior that the subjects choose for themselves, the study loses all value
> You'll almost never find medical studies over long periods of time like this, and you'll never find a truly causal study on complex human behavior; there literally is no medical research that can properly do this.
Yes, true. This is why decent medical results come from animal studies, and extrapolating such studies to humans is fraught with difficulties.
> This is the field that "proved" cigarettes are bad for us and that red meat increases the chances of heart disease.
Yes -- but using animal models, not human ones! Do you want to argue that researchers forced people to smoke cigarettes, or not, to control the outcome? That never happened.
> I agree with you that a causal result would be great, but I think we both agree it's not possible for something like this.
"Not possible" cannot be used as an justification to erode scientific standards.
> My point is that they make a very convincing case that shouldn't be dismissed so easily
They do nothing of the kind. The flaws are obvious and fatal to any useful conclusion. The explanation that a person chooses to smoke pot, and chooses to avoid intellectually stimulating activities, is much more likely than the conclusion that pot in and of itself reduces IQ, especially absent a proposed, testable mechanism.
Obviously if I designed a study that (a) forced a group to smoke pot, and (b) forced the same individuals to think harder within an intellectually enriching environment, I would get a result that contradicted this one. But it would be unethical to conduct such a study.
I cannot stress this enough -- this is junk science, and public policy is being steered by junk science, by people pretending to be scientists, pretending to be conducting objective science.
> Your principled view on this works for some sciences but is too conservative for the kinds of things epidemiology studies.
"It's too difficult" can never be used as an excuse in defense of sloppy science. Especially in the presence of a public that cannot distinguish between sterling and junk science.
All that remains is for the public to become aware how sloppy science is steering public policy, and I need to add that studies that come to the opposite conclusion (and they exist) are rarely published. The reason? "It's sloppy science."
Again, bottom line, none of this is meant to condone or encourage drug use -- there's no science to support that behavior either.
I realize that epidemiology can be sloppy science, but it's the only way we can even hope to tackle immensely important questions like this. It shouldn't be cast aside because it fails to hold up to impossible standards. According to your reasoning the vast majority of convincing evidence that cigarettes are bad for you is junk science (sorry to reuse that example but it's a very effective one).
Epidemiology is not an answer, it's part of an exploration of a question. This study is one of possibly many future studies that will attempt to etch away at the null hypothesis "there is no relationship between intelligence and smoking marijuana before adulthood". One day ten other studies like it may provide a more convincing story, or may disprove the idea altogether. Until then, this is probably the best research on the question yet. It is what will help get scientists further funding on the topic. And that's why it's important.
> I realize that epidemiology can be sloppy science, but it's the only way we can even hope to tackle immensely important questions like this.
First, that's false -- straight-up false. We can violate ethical guidelines and get terrific, reliable science. All we have to do is abandon our principles, and victimize innocent subjects. Example the Alabama Syphilis study:
The above is a despicable example of subject treatment, but it falsifies your claim.
Meanwhile, without abandoning our principles, we can use animal models, as is done with many kinds of study -- heart disease, cancer and many others.
> This study is one of possibly many future studies that will attempt to etch away at the null hypothesis "there is no relationship between intelligence and smoking marijuana before adulthood".
The study under discussion is a sham masquerading as science, it has no scientific value, the design is so poor that no reliable conclusion can be drawn from it. The level of control for confounding factors is nonexistent. Apart from that central design flaw, both the subjects and the experimenters know who the pot smokers were. It's a joke pretending to be science, and abandons every recognized experimental discipline.
The study certainly doesn't "etch away" at the null hypothesis, because there are too many pedestrian explanations for the outcome that aren't even in the thinking of the experimenters.
> One day ten other studies like it may provide a more convincing story, or may disprove the idea altogether.
Disproof is also impossible with such a sloppy study design. I can't believe what you're saying.
> Until then, this is probably the best research on the question yet.
Or the worst, and both claims can be true at once. Your remark reminds me of a Woody Allen joke. His girlfriend says, "But sex without love is an empty experience!" Allen replies, "Yes, but as empty experiences go, it's one of the best!" As meaningless science goes, this study is one of the best.
As I said earlier, I've been reading such studies for about 35 years, and (a) they all suffer from fatal design flaws, and (b) they always support the prejudices of the granting agencies and/or those of the experimenters. Also, important, studies that come to the opposite conclusion, or that don't come to a clear conclusion, tend not to be published.
> It is what will help get scientists further funding on the topic. And that's why it's important.
More bad science is important? I ask you to think about what you're saying. This study is trash masquerading as science. Anyone can design and carry out an equally sloppy study (and psychologists do this all the time) that comes to the opposite conclusion, through the simple expedient of ignoring the many confounding factors inherent in retrospective studies that don't have a control group or the kind of experimental discipline that can lead to reliable scientific conclusions.
> why isn't the longevity of sustained use a possible and, in my opinion, a more likely causal explanation?
Because the experimental subjects aren't being told to smoke pot, they are being asked whether they do. This means there is no control in the study -- it's not science, it's stamp collecting.
An obvious explanation for the study outcome is that people who smoke pot might avoid intellectually stimulating activities, which would change their measured IQ. And is that possible -- do our choices affect our IQ? In a word, yes:
12 comments
[ 3.2 ms ] story [ 44.6 ms ] threadThe conclusion drawn from the popular account (i.e. "could make you dumber") is not supported by the study in any scientific way.
They don't regain IQ points when they stop because they never had them. :D
But this is just my uninformed unscientific guess.
There are two meaningful controls, which combined hint at a causation. The IQ test from age 13, before many people begin pot use, is a control over who makes the decision to start smoking weed, and the non-smoker and smoker groups provide a control over whether pot made people stupider.
So, what factors could cause the study to show the correlation but not be a direct causation? It would have to be a confounding factor that both makes people want to smoke and something that causes them to get stupider over time that isn't the marijuana. That's possible, perhaps some sort of addictive personality disorder. But I think the simplest explanation is a causation, there is other evidence that marijuana affects the brains of adolescents.
No, I meant real, scientific controls. A study with real experimental and control groups would have groups that the researchers told that they would smoke, or not smoke, pot. That would make the study prospective, not retrospective as in this study.
If a study must monitor behavior that the subjects choose for themselves, the study loses all value, because the result can never be more than an uncontrolled correlation between A (IQ) and B (the choice to smoke pot) and any number of possible correlated factors (C) like socioeconomic, genetic, and other uncontrolled issues.
This is one of the common threads that run through most of these studies, which is why I linked the article -- it's typical of studies that I've been reading for about 35 years, and all of which suffer from the same systematic flaws.
Obviously a real, disciplined study of this topic would assign behaviors in a double-blind study. But such a study would clearly be unethical, which is why it has never been conducted.
> So, what factors could cause the study to show the correlation but not be a direct causation?
The fact that the subjects make their own decisions, the fact that there is no control group as that term is understood.
> It would have to be a confounding factor that both makes people want to smoke and something that causes them to get stupider over time that isn't the marijuana.
Yes, and until we think of a way to study this topic scientifically and without violating ethical guidelines, we will never know. It could be socioeconomic, genetic, peer pressure (smart people are geeks, don't you know that?) or some other factor.
> But I think the simplest explanation is a causation ...
That isn't an explanation unless we study it scientifically, this study doesn't do that.
This is what I find troubling about psychology studies, studies that involve monitoring behaviors but with no meaningful controls -- people don't seem to understand that it isn't science.
> ... there is other evidence that marijuana affects the brains of adolescents.
Yes - and all such studies have the property this study has -- no scientific controls.
This is in no way meant to condone drug use. All I am saying is public policy is being shaped by something other than science.
You'll almost never find medical studies over long periods of time like this, and you'll never find a truly causal study on complex human behavior; there literally is no medical research that can properly do this. This is an epidemiological study, and these kinds of studies DO have value and certainly are scientific. This is the field that "proved" cigarettes are bad for us and that red meat increases the chances of heart disease.
I agree with you that a causal result would be great, but I think we both agree it's not possible for something like this. My point is that they make a very convincing case that shouldn't be dismissed so easily - correlations should be considered for what they are, not thrown out without thought. Your principled view on this works for some sciences but is too conservative for the kinds of things epidemiology studies.
> You'll almost never find medical studies over long periods of time like this, and you'll never find a truly causal study on complex human behavior; there literally is no medical research that can properly do this.
Yes, true. This is why decent medical results come from animal studies, and extrapolating such studies to humans is fraught with difficulties.
> This is the field that "proved" cigarettes are bad for us and that red meat increases the chances of heart disease.
Yes -- but using animal models, not human ones! Do you want to argue that researchers forced people to smoke cigarettes, or not, to control the outcome? That never happened.
> I agree with you that a causal result would be great, but I think we both agree it's not possible for something like this.
"Not possible" cannot be used as an justification to erode scientific standards.
> My point is that they make a very convincing case that shouldn't be dismissed so easily
They do nothing of the kind. The flaws are obvious and fatal to any useful conclusion. The explanation that a person chooses to smoke pot, and chooses to avoid intellectually stimulating activities, is much more likely than the conclusion that pot in and of itself reduces IQ, especially absent a proposed, testable mechanism.
Obviously if I designed a study that (a) forced a group to smoke pot, and (b) forced the same individuals to think harder within an intellectually enriching environment, I would get a result that contradicted this one. But it would be unethical to conduct such a study.
I cannot stress this enough -- this is junk science, and public policy is being steered by junk science, by people pretending to be scientists, pretending to be conducting objective science.
> Your principled view on this works for some sciences but is too conservative for the kinds of things epidemiology studies.
"It's too difficult" can never be used as an excuse in defense of sloppy science. Especially in the presence of a public that cannot distinguish between sterling and junk science.
All that remains is for the public to become aware how sloppy science is steering public policy, and I need to add that studies that come to the opposite conclusion (and they exist) are rarely published. The reason? "It's sloppy science."
Again, bottom line, none of this is meant to condone or encourage drug use -- there's no science to support that behavior either.
Epidemiology is not an answer, it's part of an exploration of a question. This study is one of possibly many future studies that will attempt to etch away at the null hypothesis "there is no relationship between intelligence and smoking marijuana before adulthood". One day ten other studies like it may provide a more convincing story, or may disprove the idea altogether. Until then, this is probably the best research on the question yet. It is what will help get scientists further funding on the topic. And that's why it's important.
First, that's false -- straight-up false. We can violate ethical guidelines and get terrific, reliable science. All we have to do is abandon our principles, and victimize innocent subjects. Example the Alabama Syphilis study:
http://www.dartmouth.edu/~thabif/newfiles/tuskee.html
The above is a despicable example of subject treatment, but it falsifies your claim.
Meanwhile, without abandoning our principles, we can use animal models, as is done with many kinds of study -- heart disease, cancer and many others.
> This study is one of possibly many future studies that will attempt to etch away at the null hypothesis "there is no relationship between intelligence and smoking marijuana before adulthood".
The study under discussion is a sham masquerading as science, it has no scientific value, the design is so poor that no reliable conclusion can be drawn from it. The level of control for confounding factors is nonexistent. Apart from that central design flaw, both the subjects and the experimenters know who the pot smokers were. It's a joke pretending to be science, and abandons every recognized experimental discipline.
The study certainly doesn't "etch away" at the null hypothesis, because there are too many pedestrian explanations for the outcome that aren't even in the thinking of the experimenters.
> One day ten other studies like it may provide a more convincing story, or may disprove the idea altogether.
Disproof is also impossible with such a sloppy study design. I can't believe what you're saying.
> Until then, this is probably the best research on the question yet.
Or the worst, and both claims can be true at once. Your remark reminds me of a Woody Allen joke. His girlfriend says, "But sex without love is an empty experience!" Allen replies, "Yes, but as empty experiences go, it's one of the best!" As meaningless science goes, this study is one of the best.
As I said earlier, I've been reading such studies for about 35 years, and (a) they all suffer from fatal design flaws, and (b) they always support the prejudices of the granting agencies and/or those of the experimenters. Also, important, studies that come to the opposite conclusion, or that don't come to a clear conclusion, tend not to be published.
> It is what will help get scientists further funding on the topic. And that's why it's important.
More bad science is important? I ask you to think about what you're saying. This study is trash masquerading as science. Anyone can design and carry out an equally sloppy study (and psychologists do this all the time) that comes to the opposite conclusion, through the simple expedient of ignoring the many confounding factors inherent in retrospective studies that don't have a control group or the kind of experimental discipline that can lead to reliable scientific conclusions.
This is not science, it is psychological science.
Because the experimental subjects aren't being told to smoke pot, they are being asked whether they do. This means there is no control in the study -- it's not science, it's stamp collecting.
An obvious explanation for the study outcome is that people who smoke pot might avoid intellectually stimulating activities, which would change their measured IQ. And is that possible -- do our choices affect our IQ? In a word, yes:
http://www.nais.org/publications/ismagazinearticle.cfm?ItemN...
The only way to control for this factor would be to control the subjects' drug-use behavior, but that is unethical and has never been done.
The linked study is typical of "psychological science" -- that is, science in name only.