Yesterday’s post about the proposed DSM definition of “tobacco use disorder” got me thinking about how many misunderstandings, and even attempts at disinformation, can be thwarted by just asking “what does that mean?”. If you learned that you or someone was declared to have a psychological problem with tobacco according to the DSM, a common way of defining mental illness, it would probably sound impressive and concerning. But if you just asked, “what does it mean to be defined as having that problem”, the answer would make pretty clear that the definition is so broad as to mean absolutely nothing – nothing impressive, nothing concerning, and nothing more than just you/he uses tobacco. Presumably the dumbest parts of the definition will be eliminated, as I discussed yesterday, but I suspect that what remains will basically define “tobacco use disorder” to mean “tobacco use”.
Similarly, three days ago I wrote about the calculation that suggests that the decision about whether to give statins to healthy people to prevent a possible future heart attack or stroke, which is currently based on a somewhat complicated formula in the UK, could be replaced with a rule of “give statins to everyone over 55” with very little loss in information and perhaps an improvement in cost-effectiveness. The health press misinterpreted this as meaning that the UK should adopt the “everyone over 55” rule, or more conservatively, that the study’s author was saying that. The question the press should have put to the study’s author, or that anyone hearing the press reporters should have demanded an answer to (or at least refused to buy the claim given the lack of an answer) is what “should” means. It turned out to start with the assumption that best means something as conservative as the current policy (in terms of limiting the number getting statins) and that the tests being done to choose whether to give statins would not be done otherwise (and thus represent extra costs). It also strangely assumed that using other free information, such as varying the age cutoff based on sex, was not possible, even though it would provide a gain with no cost, and that somehow the choice should either be the present system or age alone.
Finally, consider my post from five days ago in which I discussed the recommendations governing screening mammography. The “research” I was writing about was a call to screen more aggressively than the current guidelines call for, but reading between the lines, the answer to the question “how do you define what we should do” seemed to just be the circular “screen more”. By contrast I pointed out how the guidelines consider a combination of costs and benefits, recognizing that less screening means fewer life-threatening cancers caught early, but also fewer of various other costs. You would not want to read my essay as a reason for believing that the current guidelines are correct because, though I asserted that the guidelines make sense and pointed out a few of the costs, I never specified exactly what level of costs are considered too high per successful cancer diagnosis. What I pointed out was sufficient to debunk the claims I was addressing, but could not make the case for any particular policy (other than, perhaps, not screening at such a young age that the radiation is more likely to cause cancer than detect it). So, asking “what is your basis for recommending the current policy” would reveal (a) that the claims I was criticizing were based on nothing, but also (b) that I did not present an adequate case for a particular policy either, which would (I would like to think) point out what I was really arguing was not so much the superiority of a particular policy but the flaws in the attacks on it.
In sum, asking about unclear definitions or measures on the path to a conclusion can reveal the meaninglessness of the conclusion, the weakness of the reasoning, or perhaps that what you thought was the conclusion really was not.