Math is hard. Let's go right-wing!

Michelle Malkin is very excited about a flaw she thinks she's found in the standard study of the effects of being uninsured on mortality.

Malkin notes that studies frequently find that a moderate fraction of those self-reporting as uninsured - 10%, say - turn out to be insured after all. She thinks this means... actually I've read through her post four times now, and I'm not sure what she thinks it means. She's just thrilled to have found a flaw.

Suppose, if you will, that being uninsured does indeed have some negative impact on health. We can probably trust that this is true; if not, the insurance companies are even more nefarious than we think.

Then if some people in the group labelled "uninsured" are in fact insured, this will decrease the effect of not having insurance found by the study. It means being uninsured is worse for you than the study finds, and people are dying from lack of insurance at a higher rate than is routinely reported. As the author of the study put it - Malkin quotes him in the post, but doesn't appear to have parsed it - "If present, such misclassification might dilute the true effect of uninsurance in our sample."

(To head off a possible objection, the figure for the number of uninsured Americans, and thus the extrapolation to the number of needless deaths we get annually, is not dependent on self-reporting.)

Similarly, if not having health insurance was just a temporary condition for the people in the study at the time they were surveyed - if they've all gotten health insurance since - that means that the difference in death rates is due not to people going all this time without insurance, but people missing out on insurance only briefly! We thought it was this bad for you to go your life uninsured - but actually it's that bad for you to go just a year or two without insurance.

Sorting out statistics like this can be difficult, so let's make an analogy with cookies. Let's say you make two batches of cookies. You follow the exact same recipe, only with one batch you use yummy chocolate chips... or maybe those big chocolate chunks, let's say semi-sweet, from Ghirardelli... sorry, I got distracted there.

Anyway, with the second batch, you replace the chocolate chips with lab rat poop. Then you feed people your cookies and see their reactions. Turns out people think the chocolate chip cookies are five times as good as the rat poop cookies.

Then, like in those old Folger's commercials, it turns out that your rat poop supplies were actually only half rat poop and were half chocolate chips. Does that mean rat poop is better than you thought? No. It means that rat poop is more disgusting than you thought - twice as disgusting, in fact.

For the rest, Malkin seems to have the routine right-wing problem of not understanding that people can be convinced by study that they should believe in something like health-care reform, rather than the normal conservative mode of people who already believe something doing bogus studies to try and prove their point. I guess that's kind of how it goes.

Moral of the story: if you are in no way a scientist, and you are also an idiot, you probably shouldn't try to debunk peer-reviewed papers.