Tuesday, July 19, 2005

Swimming With Stats

Here’s what MSN Encarta has to say about shark attacks:
In the wake of recent news stories about shark attacks, it's difficult not to hear that telltale Jaws theme song in your head as you head to the beach. But according to the International Shark Attack File, in 2004, for example, there were only 61 unprovoked shark attacks on humans, 30 of which occurred in the United States. With a U.S. population of about 300 million, this means the odds of being attacked by a shark in the United States are roughly 1 in 10 million.
I’ll buy that shark attacks are probably less likely than media reports would lead you to believe. But 1 in 10 million? That’s a dumb-ass statistic if ever there was one. Unless you think farmers plowing their fields in Kansas should be worried about land sharks, the relevant population for calculating the risk of shark attack is people who actually spend time in the ocean. I presume that is substantially smaller than the population of the whole United States.

(Thanks to my dad, who emailed me the link along with a single line of explanation: “perfect example of bogus statistics.” I concur.)


Gil said...


I seem to recall a Johnny Carson monologue routine when Jaws was new and scaring people away from the beaches. He had a list of "Top 10 Tips to Avoid Shark Attack".

#1. Don't leave Kansas City

I think another was: "Use the buddy system. If a shark attacks, give him your buddy"

David_Z said...

that's a terribly misleading statistic. Akin to saying that X people die of gunshot wounds in the US each year, therefore the probability that you will die of a GSW is X/3Million.

What you will find, is that the majority of people who get eaten by a shark, (shot by a gun, arrested for selling crack, etc.,) are actively and often engaging in behavior that puts them at a significantly higher risk than if they had not engaged in that behavior.

Anonymous said...

Whatever the current odds of a shark attack, rest assured that the odds are decreasing rapidly as the shark populations are being decimated by fishermen worldwide, mostly to make sharkfin soup for Asians' snooty tastes. The method of "harvesting" the fins is grotesque. The sharks (no species of shark is immune) are lured with bait, caught, then their fins are brutally cut off and the shark is thrown back into the water to drown to death. The modern methods of catching sharks (as well as tuna, etc.) are designed to wipe out as many sharks as possible in a give area to maximize profits (ain't that grand!). Sharks are a ancient and primative animal that has survived for eons. We have a lot to learn about them and for some unknown reason apparently they don't get cancer. I've heard that some people with cancer ingest shark cartilage capsules. Obviously, this is not good for shark populations either. Unfortunately, today no animal, exotic or barnyard, is safe with so many humans ravaging the planet.

It is interesting to me that economics can be applied to perhaps understand what is happening to sharks without even having to go out in the oceans to count them. First, the price of shark's fin soup is usually the most expensive item on the menu and often has to be quoted. Certainly the high price reflects the rarity of the animal. Secondly, it must be terribly hard to breed sharks in captivity, at least in sufficent numbers to make any kind of profit. Yes, one way to reduce the odds of getting attacked by a shark, bear, cougar, etc. to nil is by wiping out the species entirely. I don't feel good about this, do you? Shouldn't sharkfin soup be outlawed in the U.S. and economic sanctions placed on foreign countries that allow the wanton (or should I say wonton!) slaughter of sharks for soup?

Hoss said...

Instead of offering shallow platitudes regarding how rare shark attacks are, the scientists and news disseminators should tell people where most shark attacks occur on the beach and to avoid those areas. Shark attacks most often occur where there is bait. Bait lives near piers and near sandbars.

Blar said...

I concur as well. Interestingly, the same bogus use of statistics appears in the bestseller Freakonomics (p. 149-150, for all you book owners out there), in a section based on Levitt's 2001 op-ed, which is similarly bogus (here, for all you non book owners out there). Levitt observes that about 550 children under age 10 drown in pools each year in America and 175 die from guns, and that the country has about 6 million pools and 200 million guns, and concludes that a child is "roughly 100 times more likely to die in a swimming accident [at a house with a pool] than in gunplay [at a house with a gun]" (p. 150).

Steve should not be using the number of guns in America. There aren't even 200 million households in America, let alone 200 million households with guns. The relevant stat is the number of households with guns, or, really, the number of households that have both guns and children under 10. A household with 8 guns and no children does not provide evidence to support the claim that it's safe for children to be around guns. His statistics about the number of pools, and the causes of deaths, should also be more closely tailored to the story he tells about whether parents should be more cautious about pools or guns (though he makes more of an effort with these stats, and they seem to be closer to accurate).

Like the shark story, the moral of the story is still probably correct. Shark attacks are rare and a swimming pool is more likely than a gun to kill a kid. The numbers, though, are based on rough estimates with serious problems.