Fact-Checking Requires Checking Facts

Seattle, WA - Last week we blogged about Fact Cooker Mark Robison of the Reno Gazette-Journal. One of the studies he criticized in the article was Dog attack deaths and maimings, U.S. and Canada, a breed-specific log compiled since 1982 by Animal People editor Merritt Clifton, often described as The Clifton Report. Clifton offers a thoughtful response to Robison's article, (What's so scary about pit bulls?, Reno Gazette-Journal, July 17, 2011), below:

The first requirement of fact-checking is to check the facts -- and cross-check them, to be certain that the facts one is checking are actually the facts about the issue under discussion.
Mark Robison, the "Fact Checker" blogger for the Reno Gazette-Journal, recently flubbed the job in both directions.
Here is a link to Robison's blog, his headline, and the claim he purported to investigate, all of which any reader may cross-check, to be sure we are literally on the same page of the discussion.

Fact checker: What's so scary about pit bulls?
The claim
Northern Nevadans should be especially fearful of pit bulls.

Robison began his "fact checking" in response to comments from one S.C. Freed of Reno. Freed wrote:

"The risks posed by pit bulls are a matter of common knowledge and frequent discussion. Pit bulls are a particular danger. ... At a minimum, every person adopting a pit bull should be furnished with the following: 'WARNING: This dog is one of a breed known to the state of Nevada to pose a danger to small children, the elderly and other dogs.'"

Freed's comments, and further comments that Freed made in a separate message, were all essentially statements of informed opinion, rather than specific factual claims, but to fact-check the information behind an opinionated statement is always a good idea.
Began Robison:

One prominent source used to criticize pit bulls is a Centers for Disease Control and Prevention study on dog bites from 2000 that looked at 238 dog-related fatalities over 24 years. It found that 32 percent involved pit bulls or pit mixes.

The CDC report actually covered the years 1979-1998, a 20-year time frame which is now 13 years behind us. But it is a reasonable starting point.
What is not reasonable is to represent the conclusions from 13 years ago as appropriate in the present context, in view that fatal and disfiguring dog attacks have exponentially increased in the interim, primarily involving the breed category that the CDC found to be most often involved.
Noted Robison:

But the CDC adds a disclaimer warning the study does not - and there currently is no accurate way to - identify which breeds are more likely to bite or kill.

There are many epidemiological problems which 13 years ago were hard to predict, but are now quite well understood. Other examples include the transmission of SARS and the H5N1 avian influenzas. The nature of epidemiology is that the knowledge base is constantly expanding.
Thus, while the data sets from 13 years ago, 130 years ago, or far longer ago may still be of value, the conclusions drawn from them should always be reappraised in light of newer findings.

Another common source about pit bull dangers is Merritt Clifton, editor of Animal People newspaper.
His latest report claims that there have been 173 fatalities from dog attacks over the past 28 years and 47 percent involved pit bulls or pit mixes.

That's me. But Robison made no effort -- none -- to contact me to find out what my latest data shows. Instead, he cited a version which appears to have been at least two years old, apparently obtained second-hand. The current version would list 204 fatalities and 1,076 disfigurements by pit bulls and pit bull mixes.
Robison's next statement takes my data set completely out of context:

But Clifton's research covers less than 2 percent of dog bites requiring hospitalization

First of all, the subject of my data compilation over the past 30 years is not "dog bites," nor "dog bites requiring hospitalization."
As the headline clearly states, the subject is "Dog attack deaths & maimings."
As the preface explains further, "Over the duration of the data collection, the severity of the logged attacks appears to be at approximately the 1-bite-in-10,000 level."
In other words, what I am studying is by definition the worst of the worst, which is most likely to be reported in depth & detail.
"Hospitalization" by contrast may include temporary observation for accelerated heart rate, post-exposure rabies vaccination, treatment of non-life-threatening post-bite infections, & treatment of all sorts of other injuries & conditions which fall well short of the top level of dog-inflicted harm.
"Hospitalization" is to the level of severity I am looking at more-or-less as "fender-bender" is to "head-on collision."
Robison then adds:

and relies only on reports in the media

As purportedly a member of the working media, Robison should know that media reports include police reports, animal control reports, witness accounts, victim accounts in many instances, and hospital reports. They are, in short, multi-sourced, unlike reports from any single source.
Having known several other Reno Gazette-Journal reporters over the years, who were conscientious about multi-sourcing their reportage, I wonder what the rest of the newsroom made of Robison's comment.
Robison further asserts that media accounts:

have been shown to emphasize pit bull attacks over those by other breeds.

By whom, when, where?
Good fact-checking requires sourcing claims. That is the whole point of fact-checking.
Here we have an unsubstantiated statement which happens to be a falsehood. The most comprehensive study of pit bull coverage ever done by anyone happens to have been mine, presented to the Humane Society of the U.S. Expo in Dallas in May 2007.
Looking at all coverage at NewsLibrary, 1982-2006, proportionally weighted to compensate for the increasing numbers of newspapers in the NewsLibrary collection each year, I found only one year, 1987, in which the amount of pit bull coverage spiked out of proportion to pit bull involvement in total dog attack fatalities and disfigurements.
Continued Robison, changing the subject:

There are bite sources in Nevada that may shed light.

Looking for Nevada sources is reasonable. Looking for additional sources is reasonable. Looking at bite statistics is reasonable -- but the topics of both the CDC report and my report were specifically worst-case attacks, resulting in fatalities and disfigurements.
Bites, per se, were not the subject.
Shifting the discussion from head-on collisions to fender-benders, without even acknowledging the shift, is disingenuous.
Wrote Robison:

In 2005, the Nevada Department of Agriculture and the Bureau of Health Planning and Statistics released "A special report on dog and cat bite injuries and costs in Nevada, 1999-2003."
The report says breed is not important when analyzing bite statistics and "singling out one or two specific breeds for control may result in a false sense of security, and often ignores the true scope of the problem."

Looking at bite statistics instead of fatalities and disfigurements not only ignores the true scope of the problem, but ignores the problem itself: not whether pit bulls bite, but rather, whether they do disproportionate damage when they do bite.
Robison proceeded to cite the opinion of a local animal control officer, which was especially interesting as regards the discrepancy between what the officer opined and what his numbers showed:

Bobby Smith, field supervisor for Washoe County Regional Animal Services, agrees that breed is not helpful in evaluating the danger posed by a dog ...
Smith said he and his staff did not have time to hand-tally bites by breed. But, he said, Fact Checker was welcome to do so ...
Here are the numbers by dog breed:
Pit bull type dogs ... accounted for 13.4 percent of bite reports.

Robison's next egregious blunder may be blamed on another local source who had not done the research to speak with the authority that he should have had:

Washoe animal control manager Mitch Schneider said regarding dogs taken in by his agency, "If our impounds are reflective of our community - and there's some evidence to support that - then those numbers may be as good a gauge as we can get."

What evidence does Schneider have in mind?
Having reported on the animal care & control field for decades, and having studied the literature of the field going back as far as any exists, I am aware that presuming that shelter intake in ANY manner reflects the dog population of the community was recognized as a false presumption as far back as the first demographic studies of the shelter dog population, done more than 70 years ago.
Among the early discoveries were that "indoor" breeds, also known as "lap dogs," were greatly under-represented among shelter populations, because they were less likely to be found running at large; hunting breeds, especially beagles, were over-represented, precisely because their usual use was likely to result in some getting lost to run at large.
Pit bull terriers were first identified as over-represented in shelter dog populations by studies done in the mid-1980s. At that time they were under 1% of the U.S. dog population, but about 2% of shelter dog admissions.
I did my first survey of shelter dogs by breed in 1993. What I was specifically looking for was any statistical confirmation of the anecdotally reported boom in shelter admissions of Dalmatians that followed the re-release of the 101 Dalmatians film & the release of sequels.
What I found was that while Dalmatians rose to 4% of dog admissions to shelters during the time frame in question, pit bulls rose to 5% during the same years, and 10% of the dogs killed in shelters, while still falling short of 2% of the total U.S. dog population based on classified ads listing dogs for sale or adoption.
Ten years later, in 2003, pit bulls had leveled off at between 3% and 4% of the total U.S. dog population (they are currently 3.3%), but were up to 23% of shelter dog admissions and 50% of shelter dog killing.
Pit bulls are currently 29% of shelter dog admissions, nationally, and 60% of shelter dog killing, even though pit bulls have also been adopted out in greater numbers over the past five years than any other breed.
The Reno/Washoe County data shows that pit bulls actually appear to be less common there than across the U.S. as a whole:

Impound stats by primary breed show that pit bulls make up about 17.3 percent of dogs, retrievers 15.8, Chihuahuas 9.5, medium herding dogs 6.5, German shepherds 5.8 and small terriers 4.5.
So pits are 17.3 percent of the impounds and 18.6 percent of bites, or the bite rate is 9.3 percent higher than their impound rate.

Robison continued to make a comparison of one notoriously dangerous breed to another:

Chow chows, by comparison, have a bite rate more than twice their impound rate.

And then he quoted his demonstrably careless source again:

"As an officer or a private citizen, give me a pit bull any day over a chow chow, a German shepherd or a Rottweiler," said Schneider

What does Schneider mean by "give me"?
If he is talking about his personal preference in dogs, his opinion has no relevance to the issue of relative risk.
If he is talking about relative risk, all four cited breeds are in the elevated risk category. German shepherds and their mixes are 1.5 times more likely to kill or disfigure a person than the average dog; chows are about twice as likely; pit bulls and Rottweilers are each 11 times more likely.
An apt comparison would actually be to the average dog. Schneider also raised yet another apples-to-oranges issue:

He added, "If you want to look at anecdotal evidence, talk to people who deal with pit bulls daily. We see 5,000 dogs every few years, and you won't find anybody on my staff who thinks pit bulls are aggressive."

What humans term "aggressive" covers a broad spectrum of behaviors, which often mean very different things to dogs. Among them:

• Play-chasing;
• The air-biting & growling of two dogs who are just becoming acquainted;
• Enthusiastically greeting a person by leaping up on the person or putting paws up;
• Trying to steal food;
• Barking, even when the barking is just a friendly hello;
• Fear-biting;
• Defense of food;
• Defense of pups;
• Growling at a perceived territorial intruder;
• Rabid behavior;
• Irritability at being suddenly awakened (mostly seen in older dogs);
• Response to pain, either acute or chronic;
• Predatory behavior;
• Assertion of dominance;
• Aggression of the sort that has been bred into fighting dogs, basically by breeding out many of their social sensibilities and breeding into them hair-trigger reactivity to stimulus, so that they will fight immediately instead of going through a repertoire of behaviors meant to avoid a fight.

Any of these 15 behaviors (among others) may seem very threatening to a person, and can be very dangerous to a person, but each has a different trigger, and a different response will be appropriate.
Only the last of these 15 behaviors is specifically characteristic of pit bulls. The most distinctive part of it is the reactive element. Unless something triggers a pit bull, the pit bull will probably not appear to be "aggressive." It is what happens after the triggering event that makes pit bulls uniquely dangerous.

Hundreds of pit bull type dogs are adopted out each year by Nevada Humane Society yet Smith couldn't think of a fatality involving a dog attack, let alone a pit bull, in the 4 1/2 years since he's been here.

Simply being on the lucky side of the odds does not mean much. As a subset of my fatal & disfiguring dog attack log, I have tracked fatal & disfiguring attacks by shelter dogs & dogs adopted from shelters for nearly 30 years now.
Only two ex-shelter dogs (both wolf hybrids) killed anyone during the 1980s; none did in the 1990s; eight have since 2001, including four in the past two years. Two of the killer dogs were pit bulls. One was a Rottweiler. This interestingly mirrors their proportionality in fatal and disfiguring dog attacks overall.
Robison goes on:

The last prominent Washoe County attack was last year in Spanish Springs when a dog went after a woman's two little dogs - one was killed, the other injured - and her arm was ripped open. The attacking dog? A boxer.

Boxers, according to my 30 years of data, are three times more likely than the average dog to kill or disfigure someone.
Robison's next argument:

Twenty-seven dogs are registered as dangerous in Washoe County; three are pit bulls.

If pit bulls are 3.3% of all dogs, but 11% of the registered dangerous dogs, that alone signifies elevated risk.
Robison's next-to-last claim involves yet another "fact" which was insufficiently cross-referenced:

It's also worth noting that the Netherlands had a ban on pit bulls for 15 years before repealing it in 2008 after concluding the ban did not decrease dog bites.

This is not what actually happened. Dog bites per se were never the reason for the legislation; fatalities and disfigurements were the reason. The argument that the ban did not decrease total dog bites did not gain political traction until Dutch coalition politics in 2008 enabled Dion Graus of the far-right anti-Islamic immigration Party for Freedom to win the repeal in exchange for his party's participation in the multi-party ruling coalition.
Robison's conclusion:

The verdict
No good evidence has been presented by critics of pit bulls about their excessive danger to Northern Nevadans.

But Robison apparently never even asked "critics of pit bulls" for evidence. He didn't ask me. He didn't ask Colleen Lynn. He didn't ask Alan Beck. He didn't ask Dawn James.1
He didn't ask PETA -- an organization which, incidentally, last had anything good to say about me in 1988 & last responded to my questions in 1991.
Robison did, however, present four photos of dogs, asking readers to identify "Which of these dogs is a pit bull?"

Dog photos: The images, clockwise from upper left, show a bulldog, a boxer, a Presa Canario and an American pit bull terrier.

Since all four of these are usually considered to be "bully" breeds & are breeds of elevated risk, the photo ID question was & is irrelevant. People of good sense avoid trouble with any of them.
Robison concluded his column:

If one uses media reports to determine rates of danger -- as pit bull critics do -- then over the past decade in Washoe County, you're more likely to be killed by your own mother than a pit bull.

There are approximately 105 million mothers in the U.S. right now, i.e. 105 million people who have the opportunity to commit maternal filicide.
During the 30 years 1976-2005, there were approximately 164.5 maternal filicides per year.
This suggests that there is about one identified case of maternal filicide per 638,298 mothers in the U.S. per year.
There are between 2.4 million and three million pit bull terriers in the U.S., depending on how closely one defines pit bull. Pit bulls have killed an average of 16 people per year over the past four years.
That works out to about 1 fatality per 150,000 to 187,500 pit bulls.
Which is to say that if pit bulls & mothers had equal opportunity to kill children, pit bulls would be 3-4 times more dangerous. But in actuality, practically all mothers are in the presence of their children almost every day for 20 years.
Only about half of all people who have dogs live in households with children, so even if people with children keep pit bulls at the same rate that that keep other dogs, pit bulls would have only half the total exposure to children as mothers.
Thus pit bulls are at minimum 6-8 times more likely to kill someone than a mother is likely to commit maternal filicide.
Of course there is much more to the matter than this. Filicidal mothers relatively seldom inflict disfiguring injuries to their offspring before killing them, and accordingly are seldom apprehended in the act of killing children after tearing their limbs or face off.
When the frequency of pit bull terriers disfiguring humans is taken into account, pit bulls are 35-50 times more likely to commit mayhem than human mothers.
Of greater significance, human mothers who commit filicide and/or disfigure their children are almost always belatedly recognized as severely mentally ill. Many of these mothers follow their acts of filicide by committing suicide.
Pit bull terriers who kill or disfigure humans are by contrast perfectly normal, doing what they were bred to do.
And their is nowhere on record even one case of a pit bull killing himself/herself in remorse after tearing a person apart.

Merritt Clifton
Editor, ANIMAL PEOPLE

Related articles:
07/20/11: Blogger Targets New Fact Cooker, Mark Robison of the Reno Gazette-Journal
04/26/11: Blogger Dissects Deceptive Online Pit Bull Identification Test, 'Find the Pit Bull'