“WHAT IS A PLACEBO FOR DENTAL FLOSS?” This article by Michael Fumento reports on news articles that claim, as the headline on this New York Times article said: “Feeling Guilty About Not Flossing? Maybe There’s No Need”. The news articles are based on the fact that, as Fumento notes, “The flossing recommendation was removed from this year’s federal dietary guidelines, which by law must be evidence-based.”
There is, of course, evidence supporting the helpfulness of flossing but the “evidence-based” movement rules out any evidence that is not from randomized studies. “Anecdotal” evidence is dismissed even though, as Fumento points out, there are thousands of dentists who have done thousands of dental examinations. Fumento says: ” Yes, this is anecdotal evidence, but it’s a lot of anecdotes by a lot of professionals.” Fumento puts his finger on the erroneous assumption which is now, by federal law, reducing the application of medical knowledge: “…ultimately all this is driven by flipping on its head the old maxim that “absence of evidence is not evidence of absence.â€
I pointed out here in connection with a similar “evidence-based” rejection of breast self examination that: “It is hard to conceive of a controlled experiment relating to breast self examination so it seems unlikely that breast self examination will ever be part of ‘evidence-based medicine’.” Fumento points out that flossing can never be part of “evidence-based medicine”: “what is a placebo for dental floss?”
Likewise the cupping fad. Somebody claims to have invented a “placebo cup.” Hmm… And you can always cup in the wrong place, the idea being to compare those cupped in the right places. But the placebo effect means anybody cupped ANYWHERE who isn’t a skeptic is bound to benefit. And cupping has a lot more “authority” than a pill because it’s dramatic, slightly painful, and leaves marks. But with flossing we have a biological explanation: food rotting against gums and enamel does SEEM like it would be harmful. Any biological claims for cupping are going to be rather less straight forward.
I do not understand why a placebo is necessary to evidence-based quantitative inference. What is lost when a random sample of people who have been subjected to treatment is compared to one that has had no treatment? The placebo is intended to fool participants into thinking that they have received the treatment under study.
A different issue affecting experimental design: The rejection of anecdotal evidence becomes less important as social media tools enable characterization of very large populations, which can be considered the entire universe of interest to the researcher, rather than a sampling of that universe. The only behavioral assumption needed is that there is some distortion of the results because of inaccurate or false reporting in individual data included in the reports filed by social media participants. A supporting research topic would be how to estimate the probability that this occurs in the social media reports and how this limitation affects the margin of error in the quantitative estimate of the results for the universe.