My problem with statistics is that most people are lazy -- and few ever ask the additional questions like "how did they get that number?" or "is there something else influencing that" or "is there something else to that story?" When used correctly, statistics can shed a lot of light. However, when someone looks at statistics to try to prove their opinion, statistics can be very scary.
As the old saying goes, use statistics as a sober person would use a lampost, for ilumination, not as a drunkard who uses it as a crutch.
Thus is the case with the "pit bull issue." How many newspapers have reported the CDC report (from 1998 that's 10 years old at this point) as a source stating that 32% of reported deaths caused by dogs from 1979-1997 (or most common is the 51% caused by Rottweilers and pit bulls and their mixes). Of those, how many have actually read the entire CDC report that then notes that they find the report to be inconclusive because their stats are subject to breed mis-identification, because they come from newspaper reports that are often biased, that it is unclear how to count statistics by crossbred dogs and even if all of those issues are resolved, the numbers are meaningless without knowing the total numbers of dogs by breed to put them into context? It happens occassionally, but seldom. In fact, the CDC quit counting the statistics because they saw that people were using them inappropriately.
Then enter the likes of Merritt Clifton. Clifton runs a newspaper called Animal People and keeps track of dog bite statistics and is a self-proclaimed expert on this subject. I hadn't really ever heard of Clifton, or his newspaper, until he was quoted with his "statistics" in the Cleveland Plain Dealer a few weeks back. I contacted Clifton about his "stats" because I was curious.
Clifton tracks bite stats by breed of all "attacks doing bodily harm" -- including fatalities, maiming and other injuries requiring significant hospitalization. According to Clifton's numbers, Pit bulls, rottweilers and Wolf Hybrids account for 77% of all such attacks. (I'll be happy to send my correspondence with Clifton if anyone would like a copy, just email me).
Of course, Clifton gathers his numbers from press reports - which suffer the same problem that the CDC found 10 years ago in having media bias of over-reporting certain "breeds" of dog attacks, and under-reporting others. His numbers also suffer from the same breed mis-identification. But Clifton's numbers also come with an arbitrary "serious bodily harm" which is very arbitrary based on media reports. How can you determine if one child's face bite is more physically altering than another's based on a media report?
The more concerning part of Clifton's numbers is that he ONLY tracks them by breed...so there can be no other correllation than can be gathered. When I asked him about tethered dogs he responded : "it is part of the norms of pet dog existence, and if pit bulls on chains tear people up a lot more often than Labs or Beagles on chains, that is a fact of relevance".
While I don't dispute this, if there are 100 Labs, and 100 pit bulls, and 10 of the labs are kept on chains, and 60 of the pit bulls are kept on chains, if every dog that was chained bit someone, and you only tracked attacks by breed, you would never, ever, get to the most relevant point of the statistic. You would only say that "pit bulls attack more" vs "all chained dogs attack". I doubt there is anyone in the country that would try to argue that the majority of Labs live harder lives on the end of a chain than the majority of pit bulls. So what is the relevant point?
I think most animal control offices make this mistake -- and only track dog bite stats based on breed. If tracking by breed is the only way you keep data -- it is absolutely impossible to come up with any other correlation in dog attacks other than breed (which there are many other correlations that are more relevant).
Then, enter the worst of "statistics" comes from this article in the Vancouver Sun (thanks to Social Mange for finding this one earlier last week). The bite statistics found at the bottom of page one come from Clifton's work (with none of the hard questions asked). Then, on page 2, the author states that reported 37 US states have partial bans. First off, only ONE US state has any law mandating and BSL -- and that state, Ohio, had its own Supreme Court overrule that as not legal, so the law remains on the books, but unenforced. Then, 14 US States have actually declared Breed Specific Legislation ILLEGAL in their states. So I guess, that if you look at the 36 remaining states that don't make BSL illegal, then 36 states actually allow for BSL...so I guess that's what she's considering as a state-wide partial ban. Again, misleading, and innacturate.
Statistics are important -- but they must be viewed for illumination...an not twisted and turned to support a particular viewpoint. Otherwise, they're not only meaningless, but worse than meaningless, because they create laws that try to solve problems, but without identifying the actual problem. It's important to identify the actual problems...only then can they hope to be resolved.