During this year’s TAM, I had the distinct pleasure of accompanying Steve Novella and Michael Shermer to debate an antivaccinationist at FreedomFest, a conservative/libertarian confab that was going on in Las Vegas at the same time as TAM. That antivaccinationist turned out to be Dr. Julian Whitaker, a man who champions Dr. Stanislaw Burzynski and is one of Suzanne Somer’s doctors. There’s no polite way to put this, Steve Novella wiped the floor with Dr. Whitaker, mercilessly pummeling him with facts, analysis, and logic to the point where even the audience appeared to be grumbling. Even though I am someone who has demonstrated over the years a proclivity for applying heapin’ helpin’s of not-so-Respectful Insolence to antivaccinationists like Dr. Whitaker, even I was starting to feel a little bit sorry for him by the end of the debate, although not sorry enough to lay off of him after the debate was over. As you will recall, I went up to the table, along with a statistician I didn’t know, and started asking some very pointed questions about a graph he used to demonstrate the “autism epidemic” and correlate it with increasing numbers of vaccines. That graph is reproduced below, my having taken a picture of it from the newsletter passed out before the debate:

As I said in my last post, seldom have I seen such statistical, mathematical, and scientific ignorance on display in one graph. After all, Dr. Whitaker extrapolated from a dataset starting apparently at 2000, producing in essence exponential graphs in which autism incidence for boys will reach 100% by 2032 and 100% for girls by 2041. Yes, I kid you not. That’s exactly what the graphs claim. So the statistician and I asked Dr. Whitaker how he generated the graph, what datasets he used, what statistical model he used, his justification for using what appears to be an exponential curve fit; what software he used; and in general how he can justify a projection that shows 100% of the population being diagnosed with autism spectrum disorders within 30 years. Dr. Whitaker’s answers were evasive and indicated that he clearly didn’t know what he was talking about. To his credit, he did say that he would send me the information I requested, and I did give him my e-mail address. He also told me that a man named Shawn Siegel had made the graph for him. Siegel describes himself as having done “research,” but his research and reality are related to each other only by the occasional coincidence. One might expect someone like Siegel to produce something so utterly ignorant. He has no scientific training. However, someone like Dr. Whitaker should really know better. But he doesn’t. He not only believes such an utterly incompetent graph, but publishes it in his newsletter and uses it during his talks to fool the scientifically illiterate.

In any case, I haven’t heard back from Dr. Whitaker yet (not that I really expected to), but my readers inform me that he has answered, after a fashion, in the comments of a blog post he had written earlier. I guess the after having been asked again and again how he had done the graph in the comments of the post and on his Facebook page he finally felt compelled to answer. Unfortunately, his answer equals the scientific insight of the original graph, and I don’t mean that in a good way. You’ll see what I mean in a minute:

I understand that the projected rate of 100 percent seems unreasonable. However, the graph is based on estimated autism rates of 1 in 5,000 in 1982 (a rough estimate based on the fact autism rates were not routinely recorded back then) and 1 in 91 in 2009 http://pediatrics.aappublicati…, for an average yearly increase of about 14 percent. (Current statistics reveal that 1 in 88 children are now affected. http://www.cdc.gov/ncbddd/auti…

The graph reflects the assumption that the current rate of autism diagnoses continues in this upward trend unabated. According to these numbers, autism spectrum disorder will saturate the pediatric population by 2040. Anyone who does the math will come to the same conclusion.

No, he will not. Anyone who tries to do the math correctly will conclude that whoever made this graph is an idiot.

I’m half-tempted to let the reader figure out what’s wrong with this for himself or herself. Let’s see. Did Dr. Whitaker really just extrapolate between two points, a 1 in 5000 incidence in 1982 to a 1 in 91 prevalence in 2009. Leaving aside the fact that this apparent increase in autism prevalence is likely mostly due to broadening of the diagnostic criteria, increases screening, and diagnostic substitution, I can’t help but note that Dr. Whitaker lays down a lot of wrong here. The study he cites found an autism prevalence of 1 in 110 in 2007, not 2009. So let’s take a look at what he appears to have done. Now, for the life of me, I can’t figure out how he got a 14% per year increase. Such an increase from 1 in 5000 would equal (0.0002 x 1.14^{27}), which is 0.00688, or roughly 1 in 145, not 1 in 91. If we use the year 2007, which is the year whose data was reported in the referenced paper, the results are even further away. If you use 2012 (i.e., 30 years), then you can get close (0.102 or 1 in 98), but that would be wrong. So right off the bat, something’s weird.

It’s much, much worse than that. What Dr. Whitaker apparently does is to make an estimate of a yearly percent increase from two data points, one of which is pretty uncertain and the other of which is separated by 25 years (except that, apparently, he used 27 years). It then takes that estimate and extrapolates out into the future, assuming that autism prevalence will keep increasing at 14% per year until it hits 100%. Either that, or he fitted an exponential curve to the data in this table and extended an estimate based on data points over an eight year period thirty years into the future. Again, this is wrong on many levels. I did think of back calculating to figure out exactly how Shawn Siegel or Dr. Whitaker (or whoever) did construct this graph, but, really, why bother? (My readers can do this is they wish, but, again, why bother to do the work that it would take?) His base assumptions are so wrong, so utterly brain-dead, that all that is left is to point out that his assumptions are so utterly ridiculous that any extrapolation he does can be immediately dismissed.

If you don’t believe me, then just look at the graph again and consider something that was mentioned in the comments of my original post. Look at the graph. It shows autism prevalence for boys reaching 100% by 2032 and autism prevalence for the entire population reaching 100% by around 2038. Now consider: According to the same graph, autism prevalence for girls won’t reach 100% until 2041. This is impossible on a mathematical basis alone. For the autism prevalence to be 100%, it must be 100% in both boys and girls, but Dr. Whitaker’s graph doesn’t show girls reaching 100% until after it reaches 100% of the population at large. The only way this could happen is if prevalence among boys went higher than 100%. Again, this is impossible.

None of this stops Dr. Whitaker from opining:

Again, the fact that it goes to 100 percent is certainly alarming and hard to believe. Confounding factors include the likelihood that not all children will be vaccinated, the potential for changes in autism diagnostic parameters, and the possibility that rates of increase will decline. (On the other hand, rates may also increase.)

For those who take issue with this projection, check the relevant data I’ve provided and do your own projection.

I did, and I conclude that Dr. Whitaker doesn’t know what he’s talking about. You can’t take two data points, one of which is uncertain, extrapolate an exponential curve between them, and then assume that that rate of increase will just go on indefinitely. Diseases like this might follow a sigmoidal curve, but there would be no way of estimating where the curve is likely to level off without many data points. Even then, one extrapolates more than a little bit beyond existing data at one’s own peril. Doing such an extrapolation based on two data points Of course, the truly ignorant assumption behind these graphs, even more so than the bizarre mathematics, is the idea that, even if autism prevalence were increasing this amazingly rapidly, it must be vaccines that are causing it.

Perusing the original newsletter that I described, I now can’t believe I missed the part where Dr. Whitaker was in essence taunting Steve because apparently he couldn’t make it to FreedomFest last year. According to Steve, he was invited at the last minute and details weren’t properly nailed down before his responsibilities at TAM overtook him. According to Dr. Whitaker:

Have you ever been to a widely publicized debate between two adversaries where only one of them showed up?

Well, I have. I was all set to debate Steven Novella, MD, from Yale University School of Medicine at FreedomFest in Las Vegas in mid-July. We were to argue the pros and cons of the mandated vaccine program that Americans have endured for years. But Dr. Novella simply didn’t show, and I was left to present my point of view: strong opposition to this nonsense, as you’ll read in the first story. I had looked forward to the debate, but at least I was able to tell the hundreds of people in the audience about the horrific damage vaccines are inflicting on our children—and that is hardly debatable.

Except, of course, that it was very debatable. Dr. Whitaker’s day of reckoning had been delayed a year, but it came, and when it came he looked about as foolish as foolish can be. I strongly doubt that there will be a rematch, as I suspect that even Dr. Whitaker must realize how thoroughly humiliated he was.

Finally, Steve makes an excellent point:

…debating cranks and true-believers can be effective skeptical outreach, if you have sufficient mastery of the topic at hand. I would add that mastery includes more than knowledge of the science of the topic itself, but also knowledge of the arguments used by the other side. Just as many solid evolutionary scientists have been demolished in debates against slick creationists (like the infamous Duane Gish), it would be folly to go up against an anti-vaccinationist without a thorough knowledge of their propaganda.

I’ve said it before, and I’ll say it again. I usually don’t agree with debating cranks like Dr. Whitaker. However, not everyone agrees with me. I respect that, and Steve’s commanding performance almost changed my mind. If, however, you’re going to take on an antivaccine crank like Dr. Whitaker (or a creationist crank or an alternative medicine crank, or any other crank), you absolutely need to know the common distortions, misinformation, and tropes used by that flavor of crank. It’s not enough to know the science. You need to know how that science is distorted by cranks. Steve knew that. My guess is that Dr. Whitaker’s originally scheduled opponent probably didn’t. Too bad for Dr. Whitaker that his original opponent had to bow out and Steve was actually available this year.