Wednesday, June 22, 2011

Baron-Cohen IT autism prevelance study: some comments

Another study by the Simon Baron-Cohen research group has been getting some media attention today. I read this media report this morning. Curiosity piqued, I tried to find the primary source for this, but unfortunately it had been behind a pay wall. In the comments section of my previous post, a reader asked me for my take on the study and stated that he/she was able to download it for free. I gave them my take from what I had read in the media reports before reading the actual study. I was wondering if it were still behind the paywall. Skeptical, I tried again, and sure enough, I was able to download the entire study and read it. My comment was short and I feel the study deserves a post all its own.

Baron-Cohen in the past has made the bold statements and strong hypothesis that autism genes are responsible for so-called systemetizers and this is why autism has stayed in the general population in spite of the fact that autistics seldom have children. He seems to have gone further and intimated that my suffering may be necessary to society. I have taken him to task for this elsewhere.

Martine Roelfesema and Simon Baron-Cohen and a few other authors have recently published a study in JADD, suggesting that autism may be much more prevalent in areas of the Netherlands where a concentration of IT companies such as Philips and IBM are based. They surveyed three different metropolitan areas of Holland and found prevalence rates 3-4 times as high in the area with high tech companies than in the other two cities. They suggested that genes responsible for systematization may be responsible for this.

Superficially, this study is interesting and makes good fodder for journalists looking for an intriguing story about autism and its alleged gifts. However, on closer analysis and reading, it would appear the study has some limitations and possible flaws.

While the CDC in the American prevalence studies have found great differences in prevalence in various parts of the country, it would appear that Arizona which has one of the highest prevalence rates does not have industries that would employ so-called systemetizers. I don't know whether this is true in New Jersey or not.

There are also great variations in prevalences in various parts of California according to past CDDS data. Though the Silicone Valley has higher rates than rural parts of California it is substantially lower than in Los Angeles, which would refute Baron-Cohen and company's notion about there being a casual relationship between technology abilities in parents and autistic offspring.

Variations in prevalence in Holland would be far less likely due to differences in income, population density, socioeconomic status as the country is far more homogeneous in its population than the U.S.

Recruitment in the study was voluntary. Only a bit more than half the schools who were asked to survey consented. The response rate was substantially higher for the high prevalence group than the other two areas. Weighting factors that I concede I don't understand were used to supposedly eliminate this bias, but I'm still skeptical. The response rates from special ed schools was much higher than mainstream schools. The age range of kids surveyed was 4-16. In one of the two control communities, the prevalence rate was about the same in the older kids than in the younger kids. In the other two communities it was much higher as later diagnosis was included. This could have influenced the results of the study. The diagnoses were not verified, though this may have been infeasible.

Most germane of all, the Baron-Cohen group did not control for persons moving into the high prevalence area because of the possibility of better services, etc. so this might invalidate the findings. There may be other issues in the study that I will think of and I might go in and edit this post later if I can think of them.

This stuff also interests me, because, as regular readers of my blog know, my father was a pioneer in the computer industry, and I wondered if his genes somehow caused my autism. However, there seems to be a more pertinent history from my mother's family than from my father's, as I had a very depressed maternal grandmother and her brother (my great uncle) was likely to have been an undiagnosed autistic in the very early 1900s. Though neither my sister nor any of my first cousins on my mother's side have autism, there seems to be a high rate of dyslexia and ADHD amongst them, which would be far higher than the occurrence of chance.

Baron-Cohen has also stated (though I can't recall exactly where) that it is possible that males who were good systematizers might have been unable to find a mate at one time and with the advent of the IT revolution and silicone valley this might no longer be a problem and two engineers and programmers would be likely to marry and produce an autistic offspring. This does not jive with the personal experience of my father and the men whom he worked with going back to the early 1950s, before a big computer revolution and increase in autism prevalence. My dad and I'm pretty sure all of these systemetizers he worked with were able to marry and produce children. I'm pretty sure none of their wives were engineers or worked in the IT profession (this includes my mom of course) and with the exception of one individual from England who had a daughter who was either severely retarded or autistic, I don't think any of the other engineers and computer people dad worked with had children that were on the autistic spectrum.

The currently published study is only phase I. They state that phase II of the study which will in part address some of the issues I have discussed in this post are forthcoming. I will await it.


SM69 said...

I have posted a fuller analysis of this study on my blog which can be found here:

The main conclusions are here:

1-First of all, why have the authors not confirmed that parents are indeed in IT technology, if they want to link IT parents to ASD? It would be an obvious information to capture in their evaluation, and one that is crucial to their main conclusion, I would have thought.

2-Secondly, there are only superficial explanations as to why there is a such high discrepancy in the schools response rates across these three cities. Is there more special need schools in the responding ones from Eindhoven? No mention of this. What is the age of the kids covered by the survey in the three towns?

We must stress that the main potential explanation to a difference in rates is that the populations have been captured differently- and this has to be fully addressed in order to proceed towards any conclusion.

3-Thirdly, are dyspraxia and ADHD really a suitable validation to show that the populations have been captured in similar representative manners? – In my opinion, no they aren’t and the reason is that, unlike what the authors claim, Dyspraxia (at least in the UK, and I doubt very much it is different elsewhere) is essentially NOT diagnosed even when the kids have clear motor planning issues. As for ADHD, we also have a huge number of children in schools, who present with hyperactivity that are simply seen as being difficult and challenging without receiving any proper diagnosis. I would estimate that the figures given are well under the actual values for Dyspraxia and ADHD. In other words there are not accurate estimations of numbers and therefore cannot be used as reference points.

SM69 said...

4- Fourthly, what else is happening on Eindhoven? A quick look in Wikipedia gives us a good account of the actual high industrial development.

“Philips’ presence is probably the largest single contributing factor to the major growth of Eindhoven in the 20th century. It attracted and spun off many hi-tech companies, making Eindhoven a major technology and industrial hub. In 2005, a full third of the total amount of money spent on research in the Netherlands was spent in or around Eindhoven. A quarter of the jobs in the region are in technology and ICT, with companies such as FEI Company (once Philips Electron Optics), NXP Semiconductors (formerly Philips Semiconductors), ASML, Toolex, Simac, CIBER, Neways, Atos Origin and the aforementioned Philips and DAF.

Eindhoven has long been a centre of cooperation between research institutes and industry. This tradition started with Philips (the NatLab was a physical expression of this) and has since expanded to large cooperative networks.”
Do these industries relate in any way with the higher Autism rate? This possibility has not even been explored in the discussion section.

5- Lastly, and I will stop it here, let’s look at what is happening in Utrecht, the town that was found to have the lowest rate of ASC. Again, a look at wikipedia gives a good account of the city development and this is what is found, as you will there there is hardly any industry at all, but interesting to see that the university is the largest in the country, one would have thought it would have foster a large concentration of systemazing brains!

In fact the town looks so peaceful, I could not resist to include a photo of the Dom Tower above.

“Utrecht University, the largest university of the Netherlands, as well as several other institutes for higher education.

The economy of Utrecht depends for a large part on the several large institutions located in the city. Production industry has a relatively small influence in Utrecht. Rabobank, a large bank, has its headquarters in Utrecht.

Utrecht is the centre of the Dutch railroad network and the location of the head office of the Nederlandse Spoorwegen (Dutch Railways). NS’s former head office ‘De Inktpot’ in Utrecht is the largest brick building in the Netherlands (the “UFO” featured on its facade stems from an art program in 2000). The building is currently used by ProRail.
A large indoor shopping center called Hoog Catharijne (nl) is located between the central railway station and the city center. The corridors have been considered public places like streets, and the main route from station to city center is therefore open all night. Over the next 20 years (counting from 2004), parts of Hoog Catharijne will disappear as a consequence of the renovation of the station area. Parts of the city’s network of canals, which were filled to create the shopping center and central station area, will be recreated.
At the west side of the central railway station can be found the Jaarbeurs, one of the largest convention centers of the Netherlands.
One of Europe’s biggest used car markets is located in the Voordorp district. It is open every Tuesday except on official holidays. With thousands of second-hand vehicles on sale the market is a special point of interest for customers from Eastern European countries who even organize special one-way bus tours for shopping there.”

Can anyone tell me how anyone can use the data presented in this paper to substantiate a link between Parental IT Hub and Autism?

jonathan said...

The ages were 4-16, they did specify that in the study. I don't think they were claiming to have found a causative link but only a correlation so far, but supposedly were going to do a phase 2 study to see if there was a link and control for other factors such as people moving to Eindhoven for better services and such.

SM69 said...

I was referring to age range and average age in each of the 3 cities, not the whole total group- when you have that sort of difference in capturing the population:

Response in the Eindhoven region was higher (75.5%) than in the Haarlem (49.8%) and Utrecht regions (45.7%).”

you have to demonstrate a lot more that the populations are equivalent. Note that there is no actual number given, only estimates, and the correction to account for the school not responding are elusive. Why is there not more information about the kids forming these 3 groups?

Interesting that they suggest that the lower rates in Utrecht could relate non-Western foreigners living in

Somali backgrounds have 3-4 fold increase in autism rate, and I can tell you they are not in IT business. As a matter of fact they are not in any business at all.

Anonymous said...

"in spite of the fact that autistics seldom have children."

What about autistics who get women and girls pregnant by raping them and getting away with it, whether in forced arranged marriages or in war zones?

Unfortunately, mutual consent and the social skills it takes to reach mutual consent aren't required for reproduction. L(

Anonymous said...

"suggesting that autism may be much more prevalent in areas of the Netherlands where a concentration of IT companies such as Philips and IBM are based. "


"...According to Ensmenger, a second type of test, the personality profile, was even more slanted to male applicants. Based on a series of preference questions, these tests sought to indentify job applicants who were the ideal programming 'type.' According to test developers, successful programmers had most of the same personality traits as other white-collar professionals. The important distinction, however, was that programmers displayed 'disinterest in people' and that they disliked 'activities involving close personal interaction.' It is these personality profiles, says Ensmenger, that originated our modern stereotype of the anti-social computer geek...

"...Today, we continue to assume that the programmers are largely anti-social and that anti-socialness is a male trait. As long as these assumptions persist, says Ensmenger, the programming workforce will continue to be male-dominated. Although the stereotype of the anti-social programmer was created in the 1960s, it is now self-perpetuating. Employers seek to hire new recruits who fit the existing mold..."