Have a listen to Ann Coulter talking to Bill O’Reilly on Fox News last week (the show aired on St Patrick’s Day, hence O’Reilly’s green tie). Coulter is a social conservative columnist and lawyer, well known in the US for her right-wing diatribes. Here she is talking about recent events in Japan, regarding which she takes the somewhat unusual position of claiming that, actually, a meltdown at Fukushima would not be that problematic at all. This is because, contrary to common opinion, exposure to the levels of radiation emitted during nuclear accidents is actually good for you. Literally, she is saying that people in Japan will benefit from radiation exposure. It says something about the rashness of her claims that even a conservative climate-change equivocator like Bill O’Reilly appears to be a little sceptical about them.
One interesting aspect of the clip is that Coulter explicitly attempts to back up her claims by citing relevant mainstream science. It is always difficult to counter research-based arguments without access to the primary sources, and clearly Bill O’Reilly is in little position to question her on the specifics. Given that Coulter’s claims have found a wide-ranging online audience in recent days, it is important that each of her research-based arguments be scrutinised for accuracy. After all, conspiracy theorists and climate-change deniers are always at their most incorrigible when they are able to cite reference after reference to published research studies. And anyway, given that Coulter’s studies sound pretty persuasive, might she actually be right?
Coulter first mentions a study “from Canada” which she says found that the breast cancer rate among women who had received excessive radiation from chest x-rays “was lower than [in] the general population.” Unfortunately for Coulter (and for the women concerned) this was not at all the case. While Coulter identifies a 2001 story in the New York Times Science Section as her source for this study, the original research was published in the New England Journal of Medicine in 1989. It showed that, far from reducing cancer, radiation actually increased the risk of cancer for all groups of patients. In other words, Coulter has it backwards. Her confusion may emanate from a particular nuance of the research data: it was found that cancer risk in these women decreased sharply with age. So, when the NY Times stated that some women “had fewer cases of breast cancer than would be expected”, this referred to comparisons within the sample and not to comparisons between the sample and the general population. All the women in the study had increased rates of cancer: it’s just that the older women had a lesser increase than might expected based on the rates seen in the younger women.
Now, while the NY Times might take some blame for failing to clarify such complexities, we should expect a professional media commentator like Coulter to clarify the issues for herself by consulting the original research report (okay, maybe we shouldn’t quite expect that!). And in fairness to the NY Times, their article contained several strongly-worded caveats. In describing the Canada research it quoted one expert as pointing out that the study was “flawed by statistical pitfalls”, while in summarising the article overall, it quoted another expert as concluding that “There are probably more studies on the harmful effects of radiation than for any other toxic or noxious agents in the environment.” Somehow, Coulter didn’t notice these powerful contextual statements.
The second study Coulter mentions was based on data gathered in Taiwan. Coulter refers to apartment blocks constructed “in 1993” which contained high levels of the radioactive isotope cobalt-60. After 16 years, she says, “only 5 cases” of cancer emerged among their 10,000 occupants, while an equivalently sized group in the general Taiwanese population “should have gotten about 170 cases”. The research she’s describing was published in the Journal of American Physicians and Surgeons. However, that report was based on preliminary analyses which were later found to be flawed. Notwithstanding that, Coulter’s description contained a number of significant errors and omissions. Firstly, the apartments in question were built in 1983, not 1993. While this is a small enough point, it can be noted that Coulter’s error has the effect of making the data appear less out-of-date than they really are (for example, the 16-year period she alludes to ended in 1999 rather than in 2009). Secondly, Coulter inaccurately implies that the researchers looked at all cases of cancer that arose; in reality, the researchers examined only deaths from cancer. Across a 16-year period following first exposure to radiation, cancer death rates among a group of 10,000 are likely to be far too low to support reliable statistical conclusions. It would certainly have been more meaningful to track each and every case of cancer onset during the study period, whether or not death was the outcome; however, the researchers did not do this and Coulter is wrong to imply that they did.
Most crucially, the biggest problem with the Taiwan study was that its findings were confounded by age differences. When the first analyses were conducted, the researchers did not have data on the ages of apartment residents. Thus, in describing their statistics, they explicitly noted that their conclusions are contingent on “assuming the exposed population has the same age distribution as the population of Taiwan”, an assumption they identify as “a critical factor.” However, subsequent studies of this case have shown that, in fact, the age demographic of apartment residents was much lower than that of the general Taiwanese population. On its own this would be expected to result in lower cancer rates. Accordingly, a more complete revised analysis was subsequently published in the International Journal of Radiation Biology. These fuller analyses entirely contradict Coulter’s statements regarding the preliminary study. When age differences between apartment residents and the general Taiwanese population were finally controlled for, the data showed a significant dose-response effect whereby radiation exposure was associated with increased rates of cancer morbidity among apartment residents compared to in the general population. Furthermore, we can note that Coulter’s emphasis on cancer deaths (mortality) instead of cancer occurrences (morbidity) also proves to be very misleading. Rather than Coulter’s “five cases” of cancer among 10,000 apartment residents, in fact there were 141 cases of cancer morbidity among the 7,000 or so persons who were exposed to radiation.
The third and final study cited by Coulter relates to research by Bernard Cohen of the University of Pittsburgh. After describing Cohen as the “coolest cat in her column”, Coulter states that his research covered “90% of the counties in the US”, looking at levels both of radon, a naturally occurring radioactive gas, and of lung cancer. She says that radon and lung cancer “were inversely related so dramatically that it couldn’t be explained by cigarette smoking”. However, this is not the case; Cohen’s research did not have access to precise county-specific data on cigarette smoking. When this information was subsequently added to the dataset, researchers found that there was a negative association between smoking and radon levels that explained away the apparent association between radon and cancer. In other words, it turned out that counties with high radon levels also had low smoking levels, and vice versa. This in turn is related to urban-rural differences: cigarette smoking is more common in cities than in rural areas and, because of the effects of construction, radon levels are lower. Despite the fact that these more comprehensive analyses appeared in major scientific and medical journals, once again Coulter somehow missed them.
So let’s just summarise all of this. In describing her position, Ann Coulter either selectively or innacurately quotes a range of disparate sources. She egregiously misreports the findings of a 22-year-old Canadian study by implying that it found decreases in cancer risk resulting from radiation exposure, when actually it showed the opposite. She describes a preliminary partial analysis of data from Taiwanese apartment block residents, while failing to acknowledge the more complete and subsequently published analysis of the same residents which identified an entirely different conclusion. And likewise she describes an old correlational study linking low radon to cancer that was long ago exposed as confounded by a far more plausible link between smoking and cancer.
Presumably it is merely a coincidence that this string of errors ends up supporting a position that: (a) casts doubt on environmentalist concerns about nuclear energy; (b) justifies greater freedom from regulation for energy companies; (c) protects corporations that expose citizens to radiation from legal vulnerability; and (d) relieves Coulter’s right-wing audience of the awkward burden of having to feel concerned about the Japanese.
No doubt Fox News will be broadcasting corrections shortly.
Brian Hughes is an academic psychologist and university professor in Galway, Ireland, specialising in stress, health, and the application of psychology to social issues. He writes widely on the psychology of empiricism and of empirically disputable claims, especially as they pertain to science, health, medicine, and politics.