Statistics, brains, and science


Categories: Personal Autism

So, it has come to my attention that I am sometimes a little weird. No, wait. A lot weird. And this isn’t just a matter of being autistic; even comparing my behavior to that of other autistics I know, I am somewhat further out there in some ways. So, as a result of this, I’ve spent a cheerful hour or so filling out the Minnesota Multiphasic Personality Inventory (MMPI), this being the “restructured” version of the MMPI-2.

The MMPI is sort of magic. See, instead of trying to figure out what answers mean, they went with a radically different approach: Drop the theoretical basis, go with raw numbers. The test consists of a large series of true/false questions. Lots of people take the test. Many of them have known traits. Correlations are discovered between answers and traits. It does not matter why these correlations exist. Since there are a ton of questions, you can get a pretty good estimate of the likelihood that someone has a given trait without any actual understanding of the mechanism.

Which is, it turns out, useful.

One common complaint about the MMPI is that people may misunderstand the questions. The beauty of the mechanism is that, as long as other people with similar traits tend to misunderstand the questions the same way, it still works. But I found a few questions interesting, and felt like commenting on them, and on my responses to them.

9. I often feel guilty because I pretend to feel more sorry about something than I really do.

Answer: False. I sometimes pretend to feel more sorry about something than I really do, because if I don’t, people feel like I don’t care about them. Maybe I don’t. I don’t know; I am inclined to feel that “preferring that they not feel bad” is a kind of caring, but it’s not the same as the one where you automatically feel sorry about things, apparently. But why would I feel guilty about that? I think the assumption is that some people who do this feel like there is some way in which their not-feeling-sorry is a bad thing. I don’t get it. I am this; I am not something else. I can’t feel good or bad about that. I don’t really have a lot of experience of “guilt” in any recognizable form.

199. Peculiar odors come to me at times.

Answer: False. Actually, it’s obviously true, but I think the intent of the question is to imply that these odors are not adequately explained by things like “my spouse has a cat who has never quite come to terms with the notion of fully digesting food”. In short, I think they are looking for olfactory hallucinations, not people whose cats are perenially mildly unwell.

134. I am not easily angered.
155. I get mad easily then get over it soon.

Answers: True, and True. You might think these contradictory. The distinction I’m making is that “easily angered” implies (to my reading) that I am the passive object of some other actor’s angering behavior. So, say, taking offense at something someone says is “being angered”. But getting mad is different. And it is fairly hard for other people to anger me, but quite easy for me to get mad about things and then get over it.

279. Most men are unfaithful to their wives now and then.

Answer: False. I sort of suspect a lot of the questions that use qualifiers like “most” or “hardly ever” or whatever intend statements about representativeness, but I tend to interpret them as statistical claims. So I have been told that roughly 60% of people have at least one affair. But! “now and then” implies, to me, more than one event. And I am pretty sure that at least a fair number of people have one affair, but not two. So that would put us under the 50% mark. Thus, not “most”.

In general, there’s a certain amount of ambiguity, and I suspect that the MMPI may produce less useful results when taken by autistics, because some questions may be persistently misunderstood in ways that differ from the normal ways you’d expect for whatever other madnesses they may have.

I have filled out the entire form except the “gender” item. I’m aware that I have very little measurable gender identity. I suspect that I am more inclined to identify as female than as male, except that in the lack of any significant dysphoria, it’s a lot simpler to just present as male. But for purposes of norming answers to tests? Suddenly that seems like it might be worth getting “right”, except of course that I have no idea what the right answer is.

Comments [archived]

From: Dave Leppik
Date: 2013-07-23 10:59:42 -0500

Pretending to feel sorry can be a way to manipulate people, if you overdo it. It is a form of pretending to care about people whom you don’t care about. Watch any movie where the bad guy kills someone and then comforts the dead person’s loved one as a way to gain the trust of that person. That’s about as extreme an example as you can find—and it sometimes happens in real life.

As for the “most men are unfaithful” question, that would be a way of rationalizing an affair. Even if most men were unfaithful, it’s kept secret, so experiential evidence always suggests that most men are faithful— unless you are a marriage counselor or in some other extreme circumstance. And even there, you may take the evidence to prove that you are in an unusual circumstance (or that your own marriage is under siege) thereby bolstering your faith that, outside your circles, most men are faithful. That is to say, belief about whether or not most men are faithful is used to rationalize behavior one way or another.

Oh, and as someone who writes surveys from time to time, it’s often necessary to word a question in a way that is technically inaccurate or ambiguous but most people will understand— writing it in a more accurate or unambiguous way would sound too awkward, wordy, or legalistic. Which would cause people to get confused or over-think their answers, leading to a greater number of wrong answers. Language is a stochastic communication channel, with both sides doing error correcting; the Platonic ideal of accuracy is an ideal at best, an illusion at worst.