Why has autism gone up so much? Has autism gone up so much?

Of all the questions that readers of Freakonomics ask me to explore, understanding the explosion in autism is at or near the top of the list. I haven’t read the original study, but this news report offers an interesting hypothesis about the rise in autism: older fathers. Don’t tell Dubner I’m citing Wikipedia, but there is a nice entry there devoted to the frequency of autism.

Because the term autism covers such a wide spectrum of diagnoses, even pinning down the simplest questions like “how much has autism risen?” or “what fraction of the rise is due to changing standards of diagnosis?” are hard to answer.

I’ve been asking myself lately whether an economist could bring anything useful to the study of these questions? So far I’m not sure what the answer is to that question.



You're cautioning against "drawing sweeping generalizations from your particular case." I am sure you realize you are drawing a sweeping generalization from some vague "observation".


Prosa is a troll (though, I'm not sure this is intentional).


Er, no, I'm not trolling. I am convinced that men with "geeky" pursuits are significantly less likely to be married or otherwise involved than are men with more traditional interests. I've known or known of too many such men to think otherwise.
It would be great if I could prove this statistically, but I highly doubt that would be possible.


Plus: Levitt probably has sufficient resources and a secure academic post so he'd be hard to buy off.

Plus: Levitt doesn't seem to have an axe to grind here (brother developed vaccines, owns lots of Pharma stock, employed by public interest group, etc.)

Plus: Levitt has substantial popular credibility. It's a bully pulpit, as Teddy Roosevelt said about the presidency.

Question: Does Levitt know enough about epidemiology? Or, can he hire some collaborators who do? Or, is economic analysis similar enough (in the sense that both use a lot of aggregate data and try to tease out causality)? The answer is probably "yes" to enough of these.

So, why do it? One reason is that fear of vaccines may be preventing people from vaccinating. As Becker/Posner note in a recent posting on their blog, relatively inexpensive public health measures have high payback. On the other hand, autism is a terrible condition and if evidence is being ignored and the vaccines need improvement, that's bad too. If it's older fathers, then it would be good to know that -- just as it's good to know the risk of Down's syndrome among older mothers.

But that's not enough. Every researcher has to evaluate the odds that they will have something interesting / useful to say on a topic. And, this is ground that's already been plowed by others, which isn't always a plus. And, it has to be something that will hold your interest for a fairly long period of time.




Contrary to what you believe geeky people do not die alone. First of all geeky people can have relationships with other geeky people. Second of all, if money cannot get you women then explain all the super rich with hot women. And in addition to this what exactly do you mean by geeky. If you mean smart people, then you are just about one of the least geeky people on the planet.


Despite Prosa's protest, there actually could have been selective pressures that favoured higher intelligence in the whole population of North America. This could have happened since the years of the industrial revolution or earlier. Individuals who were better of economically may of survived at a much higher rate than ppoor people who would suffer from high rates of infant mortality and other ailments occuring before reproductive years. Those able to achieve a better economic status would have a greater chance of surviving and reproducing. Not to say that people with money must be smart. But more often than not people who are smart can increase there living standards in life. This of course is just a weak hypothesis. I have heard of a hypothesis concerning selective pressures on Jewish peoples, that actualy may have selected for smarter individuals and also, as has been commin in smiliar micro-evolutionary changes, have resulted in the high level of serius diseases such as tay sachs (of course some blame the incidences of disease on the small inclusive marriages that created a narrow gene pool for Jewish peoples and no I'm not just attempting to promulgate the racist stareotype of the Jewish banker).

My point is there could correlations between certain diseases and the level of IQ of individuals. Why, maybe IQ outweighed other selective pressures. Maybe certain genes responsible for certain mental faculties cause disease when paired with two alleles (such as disease like sickle cell and cystic fibrosos), or maybe it is the result of vaccines or some other factor like diagnosis, or a combination of several of these factors.

Some statistical information would be useful, but not concrete, like the actual identification of a gene responsible for autism, which would shurely reveal more than noticing that the incidence of autism coincides with people being vaccinated.



Dear Sirs: Autism is on the rise for many reasons. Older fathers feels valid. Furthermore, the combination of two goal-oriented individuals procreating, for the purposes of giving birth, after a long, honorable journey of career success before the risk of chromosomal effects may be a kind of risky trade-off. We are not just talking about financial success; for example, the astronaut/doctor/fighterpilot/head coach, married to the RCMP officer/FBI agent/brilliant lawyer. Place anecdote here. When or if Autism is the outcome, the weight on a personal level is tremendous and hard to ignore as a social cost. In a nutshell, could determined perseverance as a trait be a marker for potential Autism in offspring? I for one find your blog topic very interesting.


chrisbryan -
By "geeky," I mean an extremely introverted man with very poor relationship skills - you know, the sort that finds it almost impossible to talk to women or maintain any sort of normal human relationships - who takes refuge in sci-fi and fantasy, things of that sort. There's little if any connection with income or intelligence, for that matter.
In my experience, relatively few women display this collection of traits. Women can be very shy, but are seldom if ever introverted to a near-pathological level.



Okay well obviously someone that "geeky" is not likely to get many women,especially with that collection of personality traits. Are there many people who fit that category (aside from star treck convention attendees)? By that definition of geeky its impossible to disagree that they wouldnt have difficulties finding women.

Not to sound too offensive to anyone here, but judging by some elements of prosa's description of a "geek" I think this "geek" may be autistic "introverted", "very poor relationship skills", "maintain any sort of normal human relationship" (as those with autism have trouble or are unable to see people as people but are more inclined to see them as objects and thus have trouble maintaining or even having a "normal" relationship with another person...please someone correct me if I'm way of base on this assumption of people with autism)

That's also an interesting assertion prosa...are men more introverted than women, I guess since women have better language skills than men that is likely true (at least according to evolutionary psychologists).

And I agree Trekkies who still live with there mothers are not very likely to be smarter than the average person and are not likely to have a very high income.



"Not to sound too offensive to anyone here, but judging by some elements of prosa's description of a “geek” I think this “geek” may be autistic “introverted”"

I never thought of that, but it sounds logical. Maybe some of people now being diagnosed as higher functioning autistics might've been dismissed as "nerds" or "geeks" in the recent past.

"That's also an interesting assertion prosa…are men more introverted than women, I guess since women have better language skills than men that is likely true (at least according to evolutionary psychologists)."

Strictly based on people I've known I'd have to say that males are more likely to be severely introverted than females. But that, of course, is just anecdotal.
When it comes to forming romantic relationships, however, introversion is a more serious problem for men. Traditional sex roles may be changing, but it's still usually the case that men are expected to approach women, than vice versa.



What could an economist bring to the table? The first and perhaps most insurmoutable barrier to economic analysis is the lack of reliable data upon which to draw conclusions. I speak from the perspective of a parent of a young man who has had numerous diagnoses in his 28 years of life, some of those diagnoses including autism. Not only has the scientific view of what constitutes autism changed in those 28 years, my son's own behavior has changed dramatically. He has gone from behavior that unquestionably falls into the autistic "spectrum," to his present-day functioning, which, due to his own neurological development over time, would probably not bear that diagnosis. So how would you "count" him today? If you can't reliably label people as "autistic," how do you determine the occurrence of autism?


snirtle: Congratulations to you and your son.


I don't think data is a problem. With hospitals digitizing medical data, it's only a matter of time before someone loses another laptop with 26 million health records on it and it gets leaked onto the Internet... ok probably not, but look at the AOL search results fiasco... I bet economists are working on that gold mine as we speak.


"I don't think data is a problem."

It is a problem if the data set is fairly limited, ie, number of years covered.

Levitt might be better off using a data set from a country other than the US such as a European country with nationalized health care and truly centralized records.

Presumably one would wish to look at neuropsychological diagnoses prior to the exponential growth of autism diagnoses to see if some sort of pattern emerges (decreases in some sort of other diagnoses, etc).

As for mercury (which an economist might be able to provide some useful conclusions) Levitt could always look at heavily contaminated areas around the world, ie, parts of China (and developed nations in the recent past) and look at health stats on the children.

Of course there's the issue of underreporting; those who may have had autism in the past may also not have had the resources to engage the services of a physician and hence obtain a diagnosis that was recorded. Alternatively a physician could simply not be qualified to offer a 'correct' diagnosis due to skills or, more likely, lack of medical knowledge regarding autism at the time.

They also tend to keep records for a very, very long time (UK does this very well for all sorts of records).

The vaccine/autism link has been shown not to exist. That does not mean that another study could come along and replace the current view of the scientific community. That's somewhat doubtful. To say that it is '[his] belief' that there is no link is not quite fair. He is restating the scientific community's belief that there is no autism/vaccine link(from research, not anecdotal evidence, etc)

Relying on TIME magazine as a resource on autisim is probably not such a great idea. The same publication hyped the supposed autism/vaccine link years ago...it's not exactly objective.

That inflamation exists in the 4th ventricle, assuming that TIME is correct, does not mean that a vaccine caused it. That's a huge leap and it's pure speculation. Just as likely (in terms of speculation) would be that it's a genetic defect causing inflamation.

Thimersol has been phased out in many vaccines. Autism persists.

There's also the problem of parents who choose not to have their children vaccinated because of the scare stories. I know such a family (relatives) and their son still was diagnosed with autism (he is being mainstreamed in high school this year).



Gregg Easterbrook noted yesterday in Slate.com (Sept. 5) that the rise in autism correlates with the increase in television viewing among very young children (even greater now with the popularity of "Baby Einstein" et al.). I know correlation does not mean causality, but his argument is provocative and worth reviewing.


Speaking as a high-functioning person on the Autistic spectrum....

I graduated High School in 1991, which is around the time that they added Asperger's Syndrome to the list of things to look for. When I was in school, I was labeled "emotionally conflicted" and shoved in with the juvenile delinquents. That was the label that they had to work with. Today, they might have better sense. (Not bitter. How could they have known?)

What I would really like to see is some analysis of a DECREASE in diagnosis of EC, ADHD, and the other labels that they slapped on us for lack of a better term.

Anyway, I think that the increase is as much an increase in reporting than anything else. There are probably other factors, and I'd bet that there's a fair amount of false reporting, just as there were misdiagnoses of "emotional conflict" when I was in school.


One of the tough issues with doing empirical research on autism is the very poor quality of data on the "Y variable" (i.e. the prevalence of autism). I took a look at this as a health policy person in D.C. a few years ago. I recall there being little in the way of reliable national prevalence numbers, and the in-depth studies (which often focused on particular local populations) reported *widely* varying prevalence levels. You run into similar problems with a lot of diseases where diagnosis is primarily "clinical" (human-based) rather than "biological" (compare the variation in prevalence estimates for Alzheimer's disease). Also confounding empirical research is the fact that autistic individuals often have a lot of comorbidity (OCD tends to come with the territory, for example). Then again, my impressions are a few years out of date; perhaps methods of measurement have improved or diagnostic criteria have become more standardized since then.



Is there a correlation between autism diagnosis and state funding for autism treatment?!

Much of the autism diagnosis is based on a questionere filled out by the parents. When treatment is funded by the state, both the parents and the diagnosers have economic motivations for an autism diagnosis. Note that this is different from wanting the children to have autism symptoms. Whatever the symptoms are, the parents want them treated by the state, and the diagnoser wants more children to treat to justify her department budget.

Is there a correlation between autism diagnosis and state funding for autism treatment?!


Remember the show "Qunicy" with Jack Klugman (he played a coroner). There was one show when he reassures parents by saying "he autistic, its not like hes mentally retarded." The implication is that autism is more acceptable than "MR" and hold out some promise for treatment. As someone who worked with MR folks, I certainly encountered the belief that autism was somehow a preferable diagnosis.

To pick up on CPoint's post (#34): given the expanding defintion of autism, how much of the growth in the numbers of folks tagged as autistic results from that diagnosis being subsituted for another label? One study I've seen suggests that there's an inverse relationship between a diagnosis of MR and autism.

Probably the only thing safe to say is that autism is overdetermined. Maybe a master of regression analysis like Steve can sort out the major variables.


I sympathize with CPoint (#34). I feel I'm lucky that Asperger's wasn't on the medical radar when I was a child -- I didn't talk until nearly 2, for example. [further boring symptoms available on request.]

I was just labeled as "stupid", partly because my dad had dropped out of school at 16 (while his older siblings got master's degrees).

This was my luck. "Stupid" is a much easier thing to overcome than a mis-diagnosis from a health professional.

For a benign example of labeling issues, consider a manufacturer's customer list that divides customers into schools, food service, grocery, convenience, drug, etc. stores. The purpose is to devise effective programs for these different customers AND to track the effectiveness in increasing sales.

One problem: if you do a special school program, the number of "schools" goes up. If you do a special food service program, the number of "food service" customers goes up. A customer (and the sales person serving that customer) will try to reclassify school cafeterias one way or the other in order to get the special program and hit their cost or sales targets.

Second problem: the marketing person in charge of the special programs doesn't care, because an increase in customers in his classification -- and a corresponding increase in sales -- lets them get their own bonus.

Similar -- and less humorous -- examples exist in medical diagnosis.



this may be a case of the Law of Unintended Consequences. Its still too soon to say what its a consequence of, however. suggestions here that its a consequence of more educational testing and diagnosis is one possibility as is vaccination. Another may be ultra sound scans during pregnancy, as the latest research on mice shows tiny but significant changes in brain development caused by ultra sound scanning. We already know that babies 'feel' the scan as some wriggle in response to it.
What is most important is gathering facts by large scale health screening of apparently healthy populations, then retro analysing them for trends. this is expensive and governments resist it - why look for trouble ?


My husband is a geek--not an autism-spectrum geek (though I know at least one, who is married and has two kids, the older of whom is autistic)--but a WoW-playing, Star Wars-fan geek. We're planning to start a family within the next year. I don't expect that our children will have a higher-than-normal risk of autism. I think that "geek" needs a better definition for this discussion.