A news story in The Telegraph proclaimed the following headline:
Sharks nine times more likely to kill men than women, study says.
Australian scientists baffled by finding that men are targeted in 84 per cent of all unprovoked shark attacks, and make up 89 per cent of all shark bite fatalities.
This story got me wondering whether this finding could be flawed for the following reason:
Suppose that men are nine times more likely to be swimming in shark-infested waters than are women (for any number of reasons including men are more adventurous or have more access to SCUBA gear or any other related reasons). Furthermore, suppose that sharks have no preference in attacking men or women and attack either gender with equal probability.
It just appears sharks attack men more often because sharks happen to find men nine times more often than women because men are nine times more likely to be swimming in their hunting grounds.
If this is true, then it would appear sharks attack men nine times as often as women. But that is not the truth. That is only the appearance.
The truth is that sharks find men in their waters nine times as often as they find women and so it appears they attack men nine times more often than women.
If this is the truth, I would very much like to know what the scientific term is (from the field of Statistics) for this kind of flaw in a Statistical finding. Does anyone know?
The finding is that “men are targeted in 84 per cent of all unprovoked shark attacks, and make up 89 per cent of all shark bite fatalities”. There’s nothing flawed about it. Nor is it flawed to say that “sharks are nine times more likely to attack men than women”.
It’s flawed to read those factual statements to mean that sharks like/prefer to attack men over women. But nothing in the study says that. That sort of reading is purely the innate tendency for humans to read causation into correlation.
Of course, the lead author of the report in question is directly quoted as raising just this possibility right near the beginning of the article you link to. The scientists in question are fully aware of it. The only real “problem” here seems to be that the newspaper article - or its headline, anyway - has been phrased slightly misleadingly in order to sound more dramatic.
It’s not an appearance: it is reality. Sharks really do attack men 9x as often as the attack women. If someone says “I have another shark attack victim here,” you could assume it was a man, and you’d be right nine times out of ten.
Although it may be true, you need to be careful about what claims you can make based on it. From only that simple statement, you can not infer “therefore, if a man and woman are swimming together in the ocean when shark arrives, the shark is nine times more likely to attack the man than to attack the woman,” because the cause for the disproportionate frequency of attacks on men is not known (it could be the shark’s preference, or it could be the overrepresentation of men among ocean swimmers, or something else like behavior, e.g. maybe men are more likely to antagonize a shark and goad it into attacking).
A similar argument is sometimes made for motorcyclists regarding white helmets: this study reported that wearing a white helmet was associated with a reduced risk of crash-related injuries as compared to wearing a black helmet. That really is a true statement; the correlation does exist, no argument about it. But the cause is still unknown. It could be - as is sometimes claimed - that wearing a white helmet really does get you noticed so that drivers don’t crash into you as often. Or it could be that the kind of rider who chooses a white helmet has a safety-oriented mindset and is more likely to exhibit riding behaviors that keep him from getting into a crash (e.g. riding at a safe speed and following distance, getting training, wearing other safety gear, paying attention to traffic, etc.).
No…I don’t think that’s the term. I think this thread is more along the lines of “misleading sample reporting.”
“Correlation is not causation” is more like, “The more dogs there are in Italy, the fewer cats there are in France,” implying that the increase of dogs causes a decrease of cats.
I once heard this explanation for Great White shark attacks on (predominately male) surfers.
Sea Lions are the natural prey of Great Whites. A surfer lying on his board and using his hands and feet to propel it look, from the bottom, using shark eyes (their primary sense is smell) kinda like a sea lion. So they take an exploratory bite - and come up with foam, fiberglass and a bit of some strange meat.
Most attacks of which I have heard involve a single bite - actually eating a human is exceedingly rare, at least among the Great Whites. I did hear of a Hammerhead being cut open and a woman’s head popped out.
The headline of the article implies that simply being male means you have a 9X greater likelihood of being attacked by a shark. The text of the article contains speculations regarding the true cause of the gender mismatch, but the headline itself is misleading.
Well, maybe not flawed exactly, but I’d argue that it’s at least misleading. When you say “Sharks are…” it sounds like you’re saying something about sharks. You’re implying that the “nine times more likely” thing is a property of the sharks, rather than of the attacks or of the victims.
Perhaps the sharks smell the testosterone? The sweaty testicles and armpits reek with it? And it attracts them. Just a WAG. But it might be a explanation.
