Correlation does not imply causation 20 Dec 2009 One of the oldest fallacies is the “fallacy of affirming the consequent“. Basically it is the logical form of the old statistical fallacy of taking correlation to imply causation. Just because two things are true doesn’t mean there is some kind of causal relation, or even logical relation, between them. Here’s a case in point: A new study has found that autism is found in roughly 1% of all American children, up from 0.66% in 2002. The kicker lies in the details: “autism spectrum disorder includes classical autims, Asperger’s syndrome, and “Pervasive Developmental Disorder – Not Otherwise Specified”. Given that the latter two diagnoses are relatively recently introduced, and that it takes a while for diagnosticians to become familiar with new categories, it is unsurprising that things have “changed”, but it doesn’t seem to me to imply an increase in autistic spectrum disorders, so much as an increase in diagnoses. Apart from any other considerations, lumping autism, Asperger’s and a trashcan categorial like PDD together is a Very Bad Way to classify disorders of the mind. Like so much of the DSM… General Science Systematics trashcan categorial
General Science Spin versus framing: the tragedy of PR 31 Aug 200818 Sep 2017 For my sins, I was once a public relations guy, for an educational institution, and I held positions roughly in that domain (e.g., as public communications manager for a medical research institute, although I managed the means not the message) for the bulk of my professional life until I finally… Read More
Administrative My present work 10 Oct 2010 Life can be … interesting, for Chinese values thereof. No, I don’t mean the Nobel Prize, although good choice. I mean that I’m presently undergoing some kind of curse. I think it’s called “work” . Work!? So what I’m doing doesn’t translate to meaningful blog posts. Not always (but eventually,… Read More
Australian stuff Why anti science? 23 Nov 2013 Over the past few decades there has been an increasingly negative attitude by governments, pundits, religiosi and faux philosophers against science. We have seen an increase in denialism about climate change (one of the most well supported scientific models of the day), vaccination, evolution, medical research in general, and the… Read More
I don’t think any conclusions can be drawn either way. PDD-NOS has been used to diagnose kids with autism for decades. Autism is one of those disorders that doctors do not like to diagnose, in part because of concerns over long term stigma and negativity about long term outcomes. Unfortunately, I’m not terribly sure it’s a bad way to classify disorders of the mind. (Autism Spectrum) Rather it seems like dicing up disorders which require identical treatment and present with similar behaviors (albeit in different severities) is unhelpful. We’re not going to know if there’s an actual increase in autism, however, until we can settle on a definition. While now Autism, Aspergers, and PDD-NOS is considered “autism spectrum” are the total numbers (all three in 2002 compared to autism spectrum diagnoses in 2009) different?
OK; I give up: I think I agree but what in Helva is the ‘DSM…’? You can be almost as maddening as physicists who insist on using acronyms because they are too lazy to spell out what they mean!
The DSM is the Diagnostic and Statistical Manual of Mental Disorders. It gets reformulated every few years. John is right that it’s flawed, but at the same time clinicians need something to use for diagnosis, and when they don’t know the underlying cause(s) (as is the case with the various disorders subsumed under “autism spectrum”) pretty much the only option is to sort according to similarity of symptoms. With any disease, if you’re diagnosing by symptoms alone, there’s always the chance that you might lump together several similar symptoms and then conclude (incorrectly) that they must all have the same underlying cause. I think that’s why John doesn’t like the DSM method, and he’s right about that. But you gotta start somewhere. At least when we have some kind of systematic way to organize symptoms we have a framework within which to start looking for underlying neurological causes to see whether our categories reflect causes or not. The anti-vax movement, claiming that vaccines cause autism, is a perfect example of the kind of flawed reasoning John is criticizing. They treat ASD as if it is one thing with one underlying cause, then note that the onset of the various disorders labeled “autism” begins around the same time that vaccines occur, and then declare they’ve proven causation. It’s nonsense.
Classifying diseases by symptom is like classifying animals by color. You can make it really complicated. You can become an expert in it. But in the end, it’s still bullshit. Even classifying by taxonomy is bullshit, just less so than color. If the classification has no underlying rational relevance, it’s less than useless. It’s actively counter-productive.
In the absence of a known cause or a reproducible assay, diseases are doomed to suffer inflation because there is a huge reservoir of aches and incapacities in the population at large and a correspondingly enormous demand to explain ’em. If and when a mechanism underlying some portion of the identified cases is discovered, the bubble bursts as may be happening now with chronic fatigue syndrome if it turns out that the recently discovered retrovirus really does cause its classic symptoms. Then some of the erstwhile sufferers find themselves victims of just another goddamn garden-variety disease while the rest go back to being classified under the rubrics of neurosis, hypochondria, depression, or some other learned name for ignorance. This sort of thing seems inevitable, at least until we understand what ails us a lot better than we do now.
Thanks WES; that explanation was extremely helpful. Now I can actually say that, having raised a child diagnosed with ‘Asperger’s Syndrome’, I do agree with John.
I wish I had my G.P’s e-mail he should be reading this a.s.a.p. From personal experiance this is not confined to D.S.M or mental conditions when underlying causes are not the result of one condition. The diffrence in the standards used to diagnose by G.P,s and consultants in the N.H.S can be rather eye opening I have found to my surprise. I can understand and pick up the way my consultant worked and arrived at conclusions. My G.P’s working methods were somewhat unfamiliar and something of a mystery until reading John and Wes’s comments. I was speculating he was using some form of folk taxonomy I was as yet unfamiliar with.
You can’t diagnose a disease by the symptoms. For example, there are thousands of diseases that present with “flu-like symptoms.” It could be the flu. You won’t be able to tell until you go looking for the source of the disease. And considering how very little we understand about the basic functioning of the brain, psychology remains nothing more than another nineteenth century pseudo-science. It’s no different than chiropractic or phrenology. It’s just more pernicious.
Well, fine—correlation, not causation. But really, if you keep getting a shock from your hair dryer when you use it with wet hands— are you going to say that because the shocks are only correlated with wet hands, therefore you needn’t dry your hands before use? No, for all practical purposes–correlation is treated as causation. All that science searches for is close correlation: when x happens, y happens. If you stop x phenomena in a cancer cell and it dies–are you going to say that we can’t prove causation–therefore the treatment is a failure? No matter how close the correlation, one may always say that such correlation does not prove causation.——and you can always surmise yet another even more closely correlated step yet undiscovered—that is , something else could be causing the phenomenon— which again would not show causation —-So What? So What?
I think it’s generally accepted in the field that the vast majority of the increased incidence of autism spectrum disorders is primarily attributable to increased awareness and secondarily attributable to diagnostic substitution. Both reflect increased diagnoses rather than increased prevalence.