More “Trap Bias”

Whenever I read statistics about the “increasing rates of autism”, I heave a big sigh. Those statements invariable contain a whole number of assumptions, many of them flat-out wrong, or at least unexamined. In the epidemiological data, there are diagnostic issues and census issues and statistical issues and of course, the inevitable agenda issues in the reportage of the census results and analyses. I’ve previously discussed a number of these problems, including incidence versus prevalence, and correlation versus causality in the post, “Epidemics of Bad Science vs Epidemics and Bad Science”

What I would like to address today is a related issue with diagnostics and perceived prevalence, meaning, “How do we know who has autism or AD/HD or a learning disability, and how many such people are out there?”

In entomology (and in other zoological branches) we have a concept known as “trap bias”. There are a number of ways of taking a census of an animal population, including using traps. A “trap bias” means that the kind of trap you use to census a population will limit the responders to your census, and thus create unintended biases in the results.

Now, if a few synapses in your brain just fizzled from that wordy definition, let’s try a simple example. Sometimes fishing nets are used to describe trap bias, where a net with large openings will only catch large fish, leading people to think there are no smaller fishes.

But trap bias goes way beyond that. So instead, let’s say I want to find out what kinds of insects are out there, and a rough idea of how many of them there are. If I put a malaise trap out in a field then I will get a broad range of critters. There will be bunches of individuals of some species, and only a few of others.

A little thinking about this way of census-taking will quickly lead one to realise that an aerial trap is not going to catch aquatic or terrestrial insects. It only catches flying insects. We will also get different compositions between one weekly census and another, especially as the seasons go by, because some insects will mature into flying adults later on in the year, and because some insects migrate. So, you can either decide that you’re just going to measure the populations of flying insects, or you use other kinds of traps as well. After all, it’s up to you to decide what the parameters of your census will include.

But there are other, less-readily discernible biases that creep into such a method. For example, if you get a lot of male butterflies of a particular species in your trap (compared to females of the same species), does this mean that there are really more males than females? Does it mean that there are more males in this particular area than females? Does it mean that there are more males during this particular time period? (In some species, the males die after mating season.)

Or, does it have nothing to do with the actual sex ratio, and everything to do with the kind of trap? Some males are more vagile (move about the landscape) than are females. The girls just get to sit and wait for the guys to find them, casually sipping nectar. If they are a species that mates only once, the mated females may also hide out a bit so the males aren’t constantly harassing them. You will get more males in your trap simply because the guys are out bebopping around more than the gals.

When it comes to making diagnoses and determining the prevalence rates of various issues, we also run into trap biases. For example, one runs into a 4:1 male:female ratio mentioned for both autism and AD/HD. Nowadays it is becoming much more recognised that girls will more often have the inattentive than the hyperactive form of AD/HD. Hyperactive boys are a helluva lot easier to spot than girls who are staring out the window and daydreaming. This higher visability makes it more likely that they would be referred for assessment, even if both are having similar academic problems.

But there are other forms of trap bias that can creep into identification issues.

One is diagnostic parameters. There can be subtle differences in how the same syndrome can present between males and females. Girls with Asperger’s are frequently more socially-oriented and verbal than are boys (they don’t always have more social success as they get older, but the dynamics are different between groups of girls and groups of boys). The girls trend towards different interests than the boys, such as fewer train enthusiasts and more Harry Potter enthusiasts. If you think it’s difficult finding someone who’s sufficiently familiar with all the different ways that Asperger’s presents in boys, it’s even more difficult to find someone who’s familiar with girls.

Likewise, diagnosticians who usually see children will also be less familiar with how adults present, because adults have usually figured out a variety of coping mechanisms — which is not to say that adults are not struggling and failing to achieve things, either!

Another issue arises from social expectations about gender and learning. It’s too easy to dismiss a girl having more difficulty in mathematics because she’s “just a girl”, or say that well, boys just lag behind girls in reading and writing skills, and “he’ll catch up in a few years”. Such problems are too easily written off as non-problems, or else unalterable simply because they are related to gender or maturity. Being dismissive is multiply damning: appropriate responses or necessary interventions are withheld, the psychosocial needs of these struggling students are ignored, and all the while, they are repeatedly admonished that neither they nor their academic problems are worthy of attention.

Yet another problem comes from not diagnostics, but treatment preferences. A curious thing happened over the years.

Back in the 1980s, masses of public attention were focused on the “sudden” rash of numbers of kids being diagnosed with ADD or ADHD. People blamed all sorts of things as the cause, including sugar, food coloring, bad parenting, even the latest new toy,  video games . Over time the ruckus settled down, and the crackpot theories faded away (although the food additive one somehow lingers on, like a grape-flavored stain in the rug). We know that ADHD is not caused by bad parenting or video games, and we also know that it’s not just an issue in childhood. No one is even surprised to realise that like other traits, it can run in families. Of course no one in previous generations had been diagnosed with such, because the same diagnostic label had not been used (“minimal brain damage” anyone?). But the thousands of people diagnosed as adults can certainly point to the piles of documented evidence for the difficulties they had as children, and continue to have.

Nowadays the uproar about “sudden” numbers or “epidemics” of autism are the current crisis, and people blame all sorts of things for causing it (the list is not terribly different) as well as come up with all sorts of bogus treatments and cures.

But because of all of these causation myths and the subsequent treatment / cure myths, it’s also easy for struggling people to not get appropriate diagnoses and subsequent treatment or accommodations. After all, all you need to do is just pay attention, try harder, get a good paddling, quit eating anything with artificial food coloring or wheat or dairy or high-fructose corn syrup or monosodium glutamate or artificial sweetener or trans-fats, quit using mobile phones, move away from power lines, quit playing video games, get detoxified, take more vitamins, or whatever else the cure du jour requires. All these things create an imaginary goal of”becoming normal again”, simply by correcting some deficiency in how one chooses to live their life, and then magically everything will be fine. Just like other dismissive attitudes, the reality of the problem invalidated.

But you don’t have to take my word that there are plenty of people out there with undiagnosed issues, girls as well as boys, adults as well as children. Just listen to them.

But before you can listen to them, you will have to quit standing there counting what flies into your trap. Not everyone who needs to be counted comes to your door.


  1. Club 166 said,

    12 April 2008 at 13:40

    Great post! A nicely worded summary of things that can lead us astray when reading some “scientific” writings.

    Now if you can translate this into words that have no more than two syllables, with concepts that are freely understood by any 4 year old, then maybe we can send a copy to Jenny McCarthy and even she’ll understand!


  2. The Goldfish said,

    11 April 2008 at 8:33

    One factor that is perhaps useful to bring into human epidemiology and that big basket of conditions we call “learning difficulties” is class. Better educated parents (who tend to be middle-class) are more inclined to identify a problem and most particularly, push for diagnosis and treatment even in circumstances where their concerns are initially dismissed. I imagine the class issue may be more defined in the US where you seem to have to pay for any kind of expert opinion out of your own pocket.

    In fact, it is often casually said over here that more middle-class children are diagnosed with dyslexia because aspirational parents can’t bear the fact that their kids are “a bit thick” and so demand a scientific explanation for their offspring’s inability to shine academically.

    The reason I think this important is that there is no such thing as a middle-class disease. So whereas some conditions do manifest more in boys than girls or vice versa and some conditions do increase or decrease over the course of time, any kind class bias demonstrates the imperfection of the statistics (that or a truly novel explanation for such conditions). It is, of course, a rather difficult one to measure…

%d bloggers like this: