Merrit Clifton -- when the numbers just don't add up

Merritt Clifton -- when the numbers just don't add up

Reporters looking to write articles about 'pit bulls' often find themselves looking to report "both sides" of the story.  Essentially, every expert organization in the country agrees that there is nothing inherently dangerous about 'pit bull type" dogs and that breed specific laws aren't the solution -- which leaves only a couple of independent sources to refer to when trying to find the 'pit bulls are dangerous" side of the equation.

One of those sources is Merritt Clifton. I've written about Clifton before, and his tendency to misuse data to support his personal agenda (or, possibly, because having the lone dissenting point of view is possibly the only way he has any relevancy at all to the conversation). Others have written about the inaccuracies of his data too --- and I'll mention those later.

While an initial look at Clifton's methodology shows some glaring flaws, the deeper you look at the numbers, the more glaring those flaws become.

In December 2009 the Clifton Report featured 2,694 "attacks doing bodily harm" and 1,493 "Maimings) in the 27 year stretch from 1982 to 2009.   There were several "interesting" things about these numbers:

1)  Clifton issued a 2006 report that suggested there were 2,209 "attacks doing bodily harm" and 1,323 "maimings".  A little math then shows that from 2006-2009, there were 485 'attacks doing bodily harm" and 170 "maimings" -- however, during the same timeframe, showed that the total numbers attributed to 'pit bulls, rottweilers and presa canarios" went up by 509 and 215 respectively - or more than the grand total of all dogs combined. This is, of course, mathematically impossible.

2) While HCUP estimates show that there have been an average of about 7800 hospitalizations (requiring an overnight stay) in the past 16 years, Clifton's study only included less than 100 per year -- or about 1% of the total hospitalizations -- and because he relies primarily on media reports for his information (and not, say, hospital reports), his numbers are statistically not representative.  Now, Clifton will say that his study isn't meant to be all inclusive, but only cover the "worst of the worst", he is basing which incidents to include off of media report information. It seems like it would be impossible to decide which incidents to include, and which ones to not include, based on media reports on the injuries, and impossible to assume that every, or even most, major attacks are covered by the media.

But let's assume for a minute that Clifton's report is somehow, miraculously, still valid.  We still have some issues to clear up on this 2011 report.

So, I got a copy of the 2011 report directly from Clifton - -this is an updated version of the report from one published a couple of months ago that had a glaring mathematical 'gremlin' in it that even I felt guilty about beating him up over the numbers. So I pointed out that a couple of numbers were clearly wrong, and he went back and redid his math to come up with more 'accurate' numbers.  Which are still, well, odd.

Below are the numbers -- the first number is based on the December 2009 report linked to above, the second number is for the August 2011 report I received from Clifton himself. The third number is my calculation of the difference between the 2 numbers that would reflect the actual number of incidents in each category for the 19 months of 2010 and through the first week of August, 2011.

                                                                2009            2011        19 month total

Attacks causing bodily harm     2694              3383             689

Deaths                                                345                  451            106

Maimings                                         1493               1917            424

So does anything stand out to you about these numbers? 

Does it seem odd to anyone else that a full 20% of the "attacks causing bodily harm" over the past 29 years have happened in the past 19 months?  To put this in perspective, for the first 27 years of this report, Clifton reported an average of just under 100 of these "worst case attacks" each year -- but in the past 19 months it averages out to be about 450 per year -- a 450% increase.

Clifton is now reporting that 22% of all "maimings" over the past 29 years have occurred in just the past 19 months, and 24% of all fatalities.  There is just absolutely no data, anywhere, that reflects anything near this type of increase. The HCUP numbers listed above have shown a gradual 89% increase over the past 15 years -- but nothing close to the 450% that Clifton is showing, and nothing that would show a sudden increase like these number project.

There could be any number of reasons for this -- here are a couple that come to mind:

1) There is no valid statistical way to gather the information Clifton seems to be trying to gather.

2) There is much better access to information via the internet now than there has been historically -- which means that all of Clifton's historical data is majorily incomplete (which we've all known for years).

There are several other questionable numbers that don't add up in the 2011 report. He's showing an increase in 'attacks causing bodily harm" among "pit bulls, rottweilers, presa canarios, Dogo Argentinos, Fila Brasiliero and Cane Corsos" had an increase of 446 incidents over the past 19 months.  He's added more breeds to this category, but if you back out the 16 incidents of the added breeds, you have 430 incidents that have increased in this category.  However, the total number for "pit bulls, rottweilers and wolf Hybrids" went up by 614 -- which would lead to one to believe that there were around 170 incidents involving Wolf Hybrids over the past 19 months, but Clifton only has 81 total by these types of dogs over the past 29 years.

Then, there is the notes section. I got clarification from Clifton that the cases in the Notes section are included in the actual numbers.  So among the incidents that Clifton uses to mention how vicious 'pit bulls are' include:

1) A woman whose horse was startled by a pit bull and the horse kicked the woman

2) a girl who ran from a dog that never bit her and ran in front of a school bus and was hit by a bus (the girl likely ran from the dog because of the same hysteria that Clifton has helped create)

3) An attack incident involving 12 mixed breed dogs and 1 'pit bull' -- this shows up in Clifton's numbers as only a 'pit bull' attack because Clifton does not count mixed breed dogs in his reports.

4) An animal control officer who died from head injuries she suffered when she fell while trying to catch a dog.

5) A police officer who died from blood infection he received as part of a dog bite

6) And several incidents where the victims died from heart attacks

7) A person who was hit by a train when the dog they were walking took off after its owner (who was on a bicycle) and pulled the dog walker in front of an oncoming train.

8) A case where a dog's breed ID was labeled as an Akita, Weimaraner, Pit Bull and various combinations of mixes of these breeds -- but Clifton includes this case as a 'pit bull' even though he specifically notes that cases in which the breed of dog is uncertain are not included in the report.

All (and who knows what else) are bizarrely added to Clifton's report -- along with a host of other non-bite related deaths and injuries. This, I guess is fine, except he then used the data to support that certain types of dogs are more aggressive than other types of dogs based on a collection of data that includes non-bite-related incidents.

Clifton has also has apparently heard other peoples' criticisms that his data does not take into account breed population numbers -- and this time has decided to use, catch this, online classified advertisements for puppy sales, as a method to determine populations of various breeds. While I acknowledge that there is no good way to determine the number of dogs per breed, I am more than willing to acknowledge that online classified ads is a ridiculous method of trying to get that information.

Clifton has for years used his report to drive hysteria about 'pit bull' type dogs, however, every bit of actual research and science disagrees with his report.

Actual scientific studies indicate that there arefew differences in aggression between different breeds of canines.  The Science of DNA testing has also proven that visual identification of dogs (which all breed ID was solely done by until the past couple of years) is an unreliable way to identify a dog's heritage.  Clifton acknowledges in his 'analysis" that breed specific legislation is very difficult to enforce because for 'pit bulls' there is no breed standard -- and yet, doesn't acknowledge that his data is completely skewed by this same complete inability to identify types of dogs.

But Clifton carries on - ignoring science and ignoring the faulty numbers in his own reporting. And yet, because he is nearly the lone voice for the 'pit bulls are mean" crowd, the press eats it up. 

A lot of others have written a lot about Clifton's "report" over the past few years, most focused on his failed information gathering methods and other innaccuracies in the report (particularly in the Analysis section) and here are a few of my favorites:

No Pit Bull Bans -- which has a great part about his analysis about 'pit bulls' and his failed logic on that front.

Lassie Get Help -- has a great report that walks through the failures of the numbers -- but at a one-time glance at the 2006 report.  They just continue to look worse over time. She then takes on the inaccuracies in the analysis section in her part 2 -- including the false idea that German Shepherds are herding breeds.  There is also the strangeness that Clifton apparently thinks Bull Mastiffs and Presa Canarios are the same breed.

One other thing about Clifton's "analysis", Clifton uses insurance issues to justify the need for breed-specific legislation. Clifton seems completely unaware that several national insurance companies have no breed restrictions or premiums, and many of the ones that do have done so because a handful of people like Clifton have created an unfounded hysteria around certain types of dogs. 

Clifton then goes on to compare 'pit bulls' to "pumas" (yes, pumas) even though the dogs are clearly domesticated and Pumas are not....but hey, if you're numbers aren't going to really make sense, then why should the analysis of those numbers?

So this is Merritt Clifton. Possibly the leading voice in the "Breed Specific Laws are a good thing" movement - and these are his numbers. Filled with numbers that don't add up, an under-represented sample size, breed identification issues, reports of aggressive incidents that weren't, data periods in which more pit bull attacks are counted than total dog attacks, 19 months worth of data accounting for 20-25% of the report, random analysis statements and comparisons between dogs and pumas.

THIS is the leading voice of the anti-pit bull movement.  The people who oppose such legislation include professional dog training organizations, professional veterinary organizations, the national animal control association, animal welfare groups and real scientists.

Sometimes there is only one real side to an argument. This is one of those cases.