More and more people are worried about the long-term effects of contact sports on the brain. In football (soccer), studies have found that repeatedly heading the ball can lead to memory problems and an increased risk of serious brain diseases. This has led to rules limiting heading the ball in youth leagues and calls to protect professional players in similar ways. In American football, research shows a high number of former players have a brain condition called chronic traumatic encephalopathy (CTE). This has prompted the National Football League (NFL) to change some rules and introduce better safety equipment. Rugby, a sport known for its hard collisions, is also becoming more aware of head injuries. As a result, new rules require players to rest after a concussion – and there are stricter rules about preventing head contact during games. Some older players are taking legal action because of the brain injuries they suffered. Lawyers are representing over 500 former players from both rugby union and rugby league, claiming that repetitive head impacts during their careers caused long-lasting brain damage. The lawyers argue that the sports’ governing bodies failed to protect these former players from the effects of blows to the head. A recent BBC article said that “almost two-thirds of the claimants in a concussion lawsuit against rugby league authorities” had symptoms of CTE. Two-thirds is a lot, but is it really that surprising? It’s important to remember that the players in this lawsuit are a self-selecting sample. These people have been chosen for inclusion in the class action lawsuit precisely because they have evidence of brain damage. We should expect a high prevalence of conditions like CTE in this sample. So we must be careful not to infer something about all rugby players that is not supported by the data. However, perhaps the BBC article is not so troubling, since the condition for selection – that the players were part of the lawsuit – is clearly stated. More problematic are articles in which the conditions for the selection of the studied sample are not so clearly laid out. Another BBC article, published in 2023, summarised the results of studies investigating the prevalence of CTE in the brains of deceased rugby players. It reported that “68% of the brains had traces of the brain condition CTE”. This might suggest to readers that CTE is very common among all rugby players. In American football, the problem appears to be even more prevalent. In 2017, the BBC ran an article with the headline: Brain disease affects 99% of NFL players in study. The piece led with the sentence: “A study of American football players’ brains has found that 99% of professional NFL athletes tested had a disease associated with head injuries.” This sounds extremely alarming and might lead readers to surmise that nearly all professional NFL players will develop CTE. The study also surveyed the brains of college and high-school students, concluding: “Of the 202 total players, 87% were found to have traces of CTE,” giving the impression that most American football players at all levels might expect to develop CTE. Selection bias CTE research is difficult because the disease can only be diagnosed by examining samples of a patient’s brain tissue after their death. Consequently, for the NFL study, researchers at the Boston University School of Medicine, who conducted the research, drew their sample from the VA Boston Healthcare System’s “brain bank”. The bank, established to better understand the long-term effects of repetitive head trauma, holds hundreds of donated brains potentially damaged through sporting or military activities. And herein lies the problem. Many of the brains held in the bank were donated by families who suspected that their loved ones had CTE. The study hugely overrepresented players who were likely to have CTE in comparison to the general American football-playing population. To their credit, the scientists who conducted this research were at pains to point out their sample was not representative and should not be used to draw population-level conclusions. In particular, the conclusion that many sports fans reading the headlines will have come to – that a huge proportion of American football players will suffer from CTE – is not supported by the study. Somehow, that message got lost between the research article and the media’s reporting of it. The eye-catching statistics about the prevalence of CTE in rugby players, derived from a study at the University of Glasgow, are the result of a similar misrepresentation of the underlying research. In this case, the brains that were analysed came from three brain banks (from Scotland, the US and Australia). All of these repositories take donations of brains from people who were more likely to have suffered from neurological conditions, and so are unlikely to be representative of the underlying population of ruby players. The weight of evidence linking repetitive blows to the head to brain harm (particularly to CTE) is growing stronger. Studies comparing footballers to the general population show the increase in neurological conditions among football players is probably not a statistical fluke. However, if we seek to truly understand the risks of undertaking these contact sports, loved by billions, then we need to look beyond the startling headlines. Selection bias, caused by a disparity in the reasons why brains are donated for study, means it’s not enough just to sample from the brains we have available in order to establish an estimate of the prevalence of such diseases. Instead, we need to understand who is missing from the studied population, and use that information to infer how a potentially biased sample might cause the statistics we read in the headlines to be unrepresentative.