This is Science Fictions with Stuart Ritchie, a subscriber-only newsletter from i. If you’d like to get this direct to your inbox, every single week, you can sign up here.

Fatal dog attacks are suddenly much more common in the UK. Between 2001 and 2021, the annual number of fatalities never went above 6 – but in 2022, the number was 10. There have been five so far in 2023.

Many people have blamed a new breed of dog known as the American Bully XL, which is linked to at least nine of these recent deaths – including all of the deaths in 2023. It’s a huge, highly-muscled variant of the pit bull terrier, and it can go above an astonishing 60kg in weight.

It’s not just deaths: dog injuries are on the rise – and there have been some sad stories of American Bully XL attacks on other dogs, too. That’s perhaps no surprise since these dogs were specifically bred to be aggressive, assertive, fighting dogs.

Because the Bully XL is a new breed, it gets around the UK’s Dangerous Dogs Act, which bans four specific kinds of dog (the pit bull terrier, the Japanese Tosa, the Dogo Argentino, and the Fila Brasileiro – and no, I hadn’t heard of the last three either).

In other words, the XL Bully falls into a loophole in the law. Should we ban the bully?

Not so fast. Several of the most well-known animal charities in the UK, including the RSPCA, the Dogs Trust, and the Kennel Club, have formed the “Dog Control Coalition”, which lobbies against the whole idea of so-called “breed-specific legislation”.

A central belief of the coalition is that a dog’s breed doesn’t tell you anything about its behaviour. For example, this week a coalition spokesperson told ITV News that: “Research shows us that no breed of dog is more likely to be aggressive than another”. Similarly, the RSPCA’s head of companion animals said, in a tweet about the XL Bully, that “breed is not a reliable predictor of risk”.

To me, this is a really counterintuitive claim. Everyone knows that different breeds of dogs behave differently – don’t they? Or is this some kind of scientific myth that’s now been debunked?

Before we come back to the specific question of American Bully XLs, let’s look at the evidence for differences in dog breed behaviour in general.

 The evidence on breed and behaviour 

Perhaps the most prominent recent study on this question – reported on by many different media outlets at the time – was a 2022 study from the journal Science. Experts collected data on 18,000 dogs and, well, it says it in black and white in an editorial statement on the paper: “dog breed is generally a poor predictor of individual behaviour”.

Case closed, surely. Except, not really. Although there were 18,000 dogs in the study overall (it ran analyses on dog genetics), there were actually only around 2,000-5,000 dogs in the part of the study that looked at behaviours, depending on the analysis. That might still seem like quite a lot, but there were a lot of breeds: it means that there were only about 30-40 dogs per breed, which is hardly a reliable sample for finding average differences in behaviour.

Even then, they did actually find an appreciable difference in “biddability” (how independent dogs were versus how much they responded to humans’ directions), where some breeds clearly showed higher levels of the trait.

More importantly, a much bigger 2022 study, with 46,000 dogs, did indeed find substantial behavioural differences by breed, and in many ways they’re what you’d expect just from cultural stereotypes: terriers are more aggressive and prone to chase little animals; scent-hounds are extremely trainable, and so on.

A 2020 study from Finland looked at data from nearly 14,000 dogs, and found clear breed differences in many behaviours that were related to anxiety (though interestingly, the Staffordshire bull terriers in the study – the closest thing to a pit bull – tended to have a lower score for the measure of aggression than other breeds like German Shepherds).

There are, of course, problems with this kind of research. The main one is that the dog behaviours are reported by their owners: it’s a bit like the well-known problem of self-reporting in personality testing. If you’re filling in a personality test for a job, you’re hardly going to report that you’re actually quite lazy and closed-minded.

It’s the same for dogs. Not many people want to believe that their beloved dog is actually threatening or aggressive, so they might report lower levels of that trait, and that might be especially true for dogs where the stereotype is one of aggression. Even those who do want their dog to be aggressive (those who view it as a status symbol), might still under-report aggressiveness if they’re filling in a survey.

But that leads us to the biggest problem with the research: the participation in these studies is voluntary. The kinds of people who submit their dog’s data to a study like this are people who are generally interested in science and helping out with university research – does that describe the kind of person who breeds dogs, or buys dogs that have been bred, to be violent and intimidating? I’d wager that the number of American Bully XL dogs or anything similar in these kinds of surveys is next to zero, for that reason.

So looking to these kinds of studies to tell us anything about American Bully XL dogs in particular is probably going to be misleading: not only are the self-reported behaviours biased in general, but they simply don’t apply to this specific breed anyway.

But it should be clear that the claim that dog breeds have nothing to do with behaviour is on an extremely shaky foundation, and just serves to confuse and muddle the debate.

The claims made by members of the Dog Control Coalition and others shift between breed not being a “reliable” predictor of risk, and it not being a predictor of risk at all. Everyone would agree that breed isn’t a perfect predictor of behaviour – far from it. And everyone would agree that within a breed there can be huge variations in personality. But I don’t know how you can look at the research – flawed as it is – and claim that it has nothing at all to do with how a dog will act, on average.

The ”on average” point is also important. It’s a straw-man argument to say anyone claims that all members of a particular breed will behave in some set way: the point is that some breeds are more likely than others to be aggressive, on average. Or, as we’re about to see, that some breeds are more likely to cause damage when they act aggressively.

 Statistics with bite 

I know what you might be thinking. Almost every time I bring up this topic, I hear a response that’s some variation of the following: “isn’t it the owner’s fault, really? It’s not the dog – it’s the way the dog has been trained – so it’s ultimately a matter of educating humans rather than banning and killing specific breeds of dog”.

The RSPCA would seem to agree. “Any dog”, states one of the experts quoted in an RSPCA report arguing against breed bans, “can be made to bite with enough provocation”.

The response to this argument is that an owner could train a toy poodle for extreme levels of aggression, but even if it tried really hard the toy poodle wouldn’t be able to cause much of a problem for anyone. The sheer size of the American Bully XL means that if it does become violent, the damage it can do is dramatically different.

The RSPCA responds to this criticism by citing a blog article that states that the idea that pit bulls bite differently is a “myth”. That article is from 2015 and hasn’t been updated, but I was easily able to find research in the scientific literature that contradicts it.

For example, a 2019 study of 240 dog bite patients in the US state of Virginia and a further 26,000 bite reports from across many other studies showed that not only were pit bulls the most likely type of dog to bite, but that when they did, they had among the most severe injury rates of any breed. It was the same in a 2020 analysis of 207 facial dog bites from the Massachusetts General Hospital: pit bulls topped the bill. A review of the literature in 2022 also found that pit bulls were near the top of the list for injuries of children.

 An XL problem? 

This kind of research is tricky to do, for many reasons. For example, often the breed of dog isn’t recorded when someone appears at the hospital with a bite injury. Not only that, but as we’ve seen for the case of the American Bully XL, the very definition of a “pit bull” isn’t particularly clear, making it hard to compare across different datasets and studies; and since XL Bullies are new, they’re not going to feature in many of the studies anyway.

That means that any conclusion you draw from the studies I’ve cited above – or those cited by the Dog Control Coalition members – should be taken with a major pinch of salt.

Not only is this a difficult topic to research, but it’s also hard for many to discuss rationally. It’s a very emotional topic for many people on both sides – those who want to keep and defend their dogs, and those who have had a family member injured and want to see the breed banned.

But it doesn’t help that animal charities are promoting such black-and-white statements that are either based on very poor evidence or are contrary to what a great deal of evidence tells us.

That, of course, is the science: the legislation is a completely different matter.

Unless the last couple of years have been a strange statistical blip (which is always possible), something bad is happening to these numbers in the UK, and it seems likely that American Bully XLs play an outsized role. Misleading statements like “breed doesn’t predict behaviour” won’t help us get to the bottom of it: only a clear-eyed look at all the evidence will.

Other things I’ve written recently

I wrote two global warming pieces this week: one on how global warming probably was a part of the story of the Greek wildfires, and one on the worrying changes in Antarctic ice (yes that’s right -Antarctic. For years it stayed stable while everyone worried about the Arctic – but now it’s showing worrying impacts of climate change, too).

I also wrote this sceptical article on new claims that physics has found (one of) its Holy Grail(s) – a room-temperature superconductor. It would be wonderful news, but don’t hold your breath.

Science link of the week

If you’ll allow me a tiny amount of self-promotion, I’ve started a (free!) science podcast with the previous writer of this newsletter, Tom Chivers. You can hear our first episode, on weight-loss drugs, right here.

This is Science Fictions with Stuart Ritchie, a subscriber-only newsletter from i. If you’d like to get this direct to your inbox, every single week, you can sign up here.