It may surprise you to learn that I’m relatively new to the blogosphere, both as a writer and as a reader. For example, when I started this blog in the middle of April, I had no idea there were so many science blogs. I was also naive enough to think that that any scientific question about a causal association between vaccines and autism had been definitively settled in favor of the null hypothesis. (When I tell my epidemiologist friends that I’m spending time on this, they act as though I’ve joined some weird cult whose sole purpose is to convince people to stop using leeches for bloodletting.)
One of the consequences of this novelty for me is that I spend far too much time surfing the blogosphere. On the other hand, it’s fun and I’ve learned a lot, especially about the way people think. For example, in my last post I talked about my observation that some anti-vaccination commenters (on other blogs) don’t seem to understand the scientific method. I certainly got a lot of comments agreeing with this statement and expanding on it.
I’ve also come across another kind of anti-science argument that’s quite different. Paradoxically, this type of argument seems to use scientific and mathematical ideas to attack someone else’s scientific reasoning, but the attack isn’t truly scientific at all. Example: On the blog One Dad’s Opinion, there’s a nice summary of an interview with Dr. Saad Omer, a physician and vaccine epidemiologist at the Johns Hopkins School of Public Health. The interview in its entirety can be found in the online magazine Science Progress.
In the summary of the interview with Dr. Omer, there’s no math. Nevertheless, one commenter goes on the attack:
“Doing the math according to Dr. Omer and assuming 3% Unvaccinated for measles (9,120,000) and the number of current cases (64)
Rate (64/9,120,000) * 33 = .02%
Odds of being struck by lightning: 1/5000 = .02%
If you don’t worry about being struck by lightning, you shouldn’t be worried about getting measles either.”
(This commenter calls him or herself “Anonymous,” but for the sake of this post I now rename him or her MathAttack**, in memory of an old MS-DOS game my grown-up kids played when they were in grade school.)
In response, commenter HCN points out that the problem with above the “math” is that the unvaccinated usually hang around in the same groups. HCN gives examples from San Diego, Washington State, Switzerland, and Salzburg, Austria — all outbreaks involving clusters of unvaccinated children — and concludes (correctly, in my opininion), “Your math only works if the numbers were evenly spread out. They are not.” MathAttack is right back on the warpath: “Is the number of individuals that are too young evenly spread? Is the number of individuals that are immune deficient and cannot be immunized evenly spread? Is the number of elderly too old to have ever received a vaccine evenly spread? Is the remaining pool of unimmunized a minority of these other populations? Are you determining distribution based on a city block, a state or the country? My numbers are fine. If anything they are conservative, making lighting MORE dangerous.” Epi Wonk (yours truly) also makes a comment, the main point being that, MathAttack’s “math is ridiculously simplistic” and “the spread of measles in extremely complex…” MathAttack responds to me, “I will gladly update my calculations from estimates to actual when someone can provide the actuals. Exactly how many people do not have immunity to measles for every reason? Explain spontaneous outbreaks if there is no foreign exposure? Explain why foreign exposure defeats herd immunity? Explain how reducing risk of exposure, other than vaccination, is included in the calculation? Justify your calculations to demonstrate all the necessary variables, both positive and negative, that arrive at the current number of case and reasoning for the locations noted on the CDC web site.”
Then MathAttack goes back on the warpath against Dr. Omer: “I haven’t been approaching Dr.Omer’s statements from the right perspective. Let’s reverse engineer what he has said: “For example, in a national-level study it was found that kids who are exempt from vaccination requirements had thirty-three time — not percent, its times higher risk of acquiring measles with those who are vaccinated.” When was the study done and what was the risk of acquiring measles if vaccinated at that time? Has the risk of acquiring measles changed since that date? If so, why? Science Progress also started the interview with ambiguity by asking about exempt from vaccination without defining if exempt is ONLY by choice or all individuals that are not vaccinated. Two very different numbers.”
I then suggested that MathAttack carefully read the paper cited by Dr. Omer, which was published in JAMA and is entitled, “Health Consequences of Religious and Philosophical Exemptions From Immunization Laws: Individual and Societal Risk of Measles.” (which, incidentally, should be read along with a paper in Pediatrics entitled, “The Cost of Containing One Case of Measles: The Economic Impact on the Public Health Infrastructure–Iowa, 2004” and another paper in JAMA entitled “Historical Comparisons of Morbidity and Mortality for Vaccine-Preventable Diseases in the U.S.”) MathAttack seems to have read the article, and is smarter than JAMA’s peer reviewers, editors, and readers (given that there were no letters to the editor critiqing the study): “There were MANY assumptions that were normalized, or used statistical modeling rather than actual data. The math of the study may not add up according to some…”We assume that the vaccine reduces the transmission probability to each child by a given fraction, which is the vaccine efficacy. Estimation of efficacy also may be biased if vaccination is not random or if a vaccinee and a nonvaccinee do not have the same exposure to the infecting agent.” Definitely changes the risk of exemptors if they actively assure they are not exposed to the infecting agent. The distribution of the transmission probabilities over the communities was determined so that the overall numbers of expected cases in exemptors and nonexemptors were close to the observed frequencies. We used age-specific population data from the Bureau of the Census to extrapolate the percentages into estimated numbers. Thus, we were able to estimate age-specific measles incidence and the relative risk of measles for exemptors compared with vaccinated persons.” Would the individuals that do not respond to a census requests most likely be vaccinated? Does the Bureau of the Census accurately report that population over time?
Wow. Now it might seem to the naive reader that MathAttack is making scientific arguments. But you don’t have to read very carefully to realize that, in the context of this mini-debate, this person absolutely cannot tolerate uncertainty. Indeed, I would argue, with Karl Popper and R.A. Fisher and many others, that uncertainty is an essential hallmark of the scientific approach. Moreover, any approach to knowledge that insists on certainty and claims to be scientific is pseudo-science.
This insistence on certainty seems to me to be related to what Daniel Engber, in a recent article in Slate, has called “The Paranoid Style in American Science,” although it would be more correct to call it “the paranoid style in American pseudo-science.” I highly recommend reading the entire three-part article, in order to understand the context for the following quote, which comes from part 3. The author, Daniel Engber, likens today’s “radical skeptics of science” to the “conspiracy-minded, radical right” of the 1950’s and early 1960’s that historian Richard Hofstadter wrote about in his famous 1964 essay, “The Paranoid Style in American Politics.” Engbar writes:
The paranoid style, Hofstadter wrote, “is nothing if not scholarly in its technique.” In his mainstream enemies, the conspiratorial thinker sees “a projection of the self”–he’s just like them but more discerning and more rational. Indeed, for the paranoid skeptics, it’s not that science is wrong but that the scientists aren’t scientific enough. (Emphasis in original.) …If nothing can withstand our critical scrutiny, then everything seems equally probable. (You can’t prove a conspiracy … but you can’t prove anything, can you?) Thus manufactured uncertainty has devalued the real thing: The less sure we are of the world, the more precision we crave. Skepticism sells itself, and the scientific consensus—no matter how considered or probable–starts to seem a little cheap…Exactitude may sound like good science–atomic clocks, sub-micron optical tweezers, and all that good stuff we use to keep satellites in orbit and Web sites streaming. But an obsessive fear of uncertainty is the opposite of science…Organized science engenders trust, and…requires the acceptance of some degree of doubt.
See also the recent article in the Annals of Internal Medicine that discusses certainty and uncertainty in evidence-based public health.
*Special thanks to Kevin Leitch for pointing me to the Dr. Omer on Vaccines post at One Dad’s Opinion and to James Hrynyshyn at Island of Doubt for blogging about the three-part series on radical skepticism and the rise of conspiratorial thinking about science by Daniel Engber, in Slate.
**In all seriousness, this post is not meant as a personal ad hominem attack on Anonymous/MathAttack, who seems highly intelligent. I just chose this as a beautiful example of this type of reasoning and argumentation.