Analysis of AVMA Pamphlet: The Role of Breed in Dog Bite Risk and Prevention
Analysis of AVMA Pamphlet: The Role of Breed in Dog Bite Risk and Prevention
Analysis of AVMA Pamphlet: The Role of Breed in Dog Bite Risk and Prevention
a. Construct validity – You must provide a clear operational definition of the variable
you are measuring. That operational definition must reflect the reality of the
construct (in this case ‘dog bite’) that you are using. The usual criterion for this is
whether most people would agree that the term you use accurately reflects the
reality you are measuring. It is unlikely that the majority of people would agree
that a single dog bite, no matter how wounding, is the same as a sustained,
catastrophic mauling or fatal attack in which a dog delivers up to a hundred deeply
wounding bites.
b. Content validity – Your test must actually measure the real, clearly defined
variable in question.
c. Criterion validity – The criterion on which cases are rated must be clear and valid.
If the AVMA wants to group puncture wounds together with disfiguring,
dismembering, life-changing attacks, they must explicitly state this.
d. Concurrent validity – Two or more groups (eg, pit bull types vs companion dog
types) should differ on the measure in the expected way. When they don’t, you
need to examine the measure and the criteria you are using.
e. Discriminant validity – The measure should discriminate between the construct
being measured (in BSL, this is catastrophic and fatal maulings) vs. other
unrelated constructs (simple dog bites).
!
Section One: “Breeds implicated in serious bite injuries”. The first paragraph repeats some
of the same fallacies and contains some new ones:
!
“BREEDS IMPLICATED IN SERIOUS BITE INJURIES
In a range of studies, the breeds found to be highly represented in biting incidents were
German Shepherd Dog, 1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,50 mixed breed,
1,4,6,8,10,11,12, 19,17, 20,50 pit bull type,5,9,13,16, 21,20,22,23,24,25,26,27 Rottweiler,
15, 18,22,24, 25, 28 Jack Russell Terrier,21,25,26 and others (Chow Chow,7,23 Spaniel,
14,26 Collie,3,29 Saint Bernard,20 and Labrador Retriever2 ).”
!
Flaws: Failure to operationally define ‘serious bite incidents’. Shift in this first paragraph from
‘serious’ to all reported bite incidents. Failure to define ‘highly involved’. Dated and biased
sources: Two thirds of the sources cited use data from the previous century, when pit bulls were
extremely rare. Of the more recent studies cited here, more than half are written by pro-pit bull,
anti-BSL activists.
!
The second paragraph adds new flaws:
!
“If you consider only the much smaller number of cases that resulted in very severe injuries
or fatalities,21,23 pit bull-type dogs are more frequently identified. However this may relate
to the popularity of the breed in the victim’s community, reporting biases and the dog’s
treatment by its owner (e.g., use as fighting dogs21). It is worth noting that fatal dog attacks
in some areas of Canada are attributed mainly to sled dogs and Siberian Huskies,56
presumably due to the regional prevalence of these breeds. See Table 1 for a summary of
breed data related to bite injuries.”
!
Footnotes 21 and 23 refer to data collected in the previous century (1985; 1997-98). Disfiguring
and fatal attacks by pit bull type dogs have increased dramatically since then.
!
Their Table 1 is put together in such a way that it is difficult to figure out what it is telling us
without consulting the cited sources at length. The table tells us N (total number of records), but
it doesn’t tell what percent of N the listed dog types claimed. Close examination reveals that 20
of the 35 listed references are outdated. Of the 15 based on data after 2000, 10 look at foreign
countries, including 5 countries where pit bulls are banned or regulated – and thus where they are
indeed not regularly attacking people and animals. This table also shows that in the US, only pit
bull types and Rottweilers are the top two in committing fatal attacks.
!
Back to the second paragraph text. This paragraph introduces alternative explanations for the
deadly position of the pit bull type in the fatality statistics. In itself, this is good scientific
practice. Unfortunately, none of the alternative explanations the AVMA puts forth are valid:
!
1) Popularity of the pit bull type – In fact, the CDC study (cited in pamphlet footnote 23)
concludes, on page 4: “Despite these limitations and concerns [about identifying exactly
which ‘breed’ of pit bull type dog was involved], the data indicate that Rottweilers and pit
bull-type dogs accounted for 67% of human DBRF in the United States between 1997 and
1998. It is extremely unlikely that they accounted for anywhere near 60% of dogs in the
United States during that same period and, thus, there appears to be a breed-specific
problem with fatalities.” It must be noted that after 1998, pit bull type dogs and their
mixes jumped to causing 82% of all dog bite-related fatalities in 2015. It remains doubtful
that this type and its mixes are anywhere near 80% of the dog population. It is deceptive
that the AVMA omits this information. Furthermore, in the U.S., the pit bull types have
been responsible for more fatal attacks than all other ‘breeds’ and types of dogs combined,
accumulatively, since the 1980s regardless of their ‘prevalence’ in the dog population.
2) Reporting bias – This is a conspiracy theory. The AVMA suggests that all parties
concerned – from press to doctors to coroners – consistently refrain from reporting deaths
by (for example) poodles. If you want to offer such a theory, you must also offer
evidence.
3) Treatment by the owner (eg, use in dogfighting) – The AVMA footnote here refers to data
from 1985, when pit bull types were indeed mostly kept only by dogfighters. If you go to
the source, you find that nevertheless two of the five pit bull types involved in attacks
were kept as family pets. As of 2016, almost all pit bull types are kept as household pets,
and only a tiny portion is involved in dogfighting. These ‘family pet’ pit bulls are the only
ones mauling and killing in our present century.
4) Treatment by the owner – This is a variable that randomizes out – all dogs types are
subject to good and bad owners – so this is irrelevant and invalid as an alternative
explanation for the high rate of catastrophic attacks by specifically pit bull type dogs.
!
The reference in this paragraph to Canadian data is also dated (data up to 2007). Only in recent
years have American pit bull activists intensified efforts to export US pit bulls to Canada – and in
2016, pit bull type dogs have so far killed twice as many Canadians as any sled dog type,
shows data collected by Animals 24-7. Hopefully the AVMA will speedily update this pamphlet
to reflect that fact.
!
Note: The disclaimer note at the bottom of the first page of this pamphlet is deceptive. It calls this
document ‘peer-reviewed’, a term reserved for scholarly works, usually at submission to
scientific journals. The term is misplaced here. This is a political pamphlet; it is not a peer-
reviewed scholarly article or a review of the scientific literature.
!
Section Two: “Controlled studies”
!
This section tries to claim that you can only calculate ‘breed’ risk if you know exactly how
prevalent a ‘breed’ of dog is in the canine population. As the CDC pointed out, one doesn’t need
this precise information. It was enough for the CDC to know that one or two types of dogs were
claiming 67% of all fatal dog attacks and that it was “extremely unlikely that they accounted for
anywhere near 60% of dogs in the United States during that same period.” The same still holds
for the present.
!
This section acknowledges the fact that other breeds – they name Dalmatians and setters – have
been fashion dogs without the rate of human (or animal) maulings and killings – or indeed, even
‘bites’ – going up. Then they acknowledge that when Rottweilers became popular, they quickly
became a problem due to aggression. What the AVMA actually demonstrates here is that indeed,
it’s only when molosser / war dog types and baiting / fighting dog types become popular that
casualties suddenly rise – that this is a breed-specific and type-specific problem. It’s relevant to
mention here that talk of a ‘dog bite epidemic’ didn’t truly begin until fashion changed from
various human-selected companion dog types to various human-selected aggressive dog types.
And aside from questions of fashion, it remains a fact that pit bull types have always mauled and
killed humans way out of proportion to their presence in the dog population.
!
The AVMA promised in their title to talk about animal welfare, but then left this issue out
entirely. Why did they omit the information that pit bull types account for about 96% of the
savage killing of companion type domestic dogs in our communities? This is cumulative data
over the 3-year period (2013- 2015) collected by Animals 24-7. The pit bull types are certainly
not 96% of any dog population anywhere, in any community.
!
This section claims that only ‘controlled studies’ can reveal which ‘breeds’ are involved in the
most ‘bite’ incidents. In fact, behavioral science demands that field data be included in research.
The AVMA then chooses to cite a single ‘controlled’ study done in the Denver area, where pit
bulls are banned and…surprise, surprise…in Denver the only ‘biters’ left were two other breeds
(both guard dog types) and there were no dog bite related fatalities. It would seem that the
AVMA is telling us here that the one controlled study they know of shows that BSL does in fact
work.
!
The Italian study cited in this second paragraph included only registered dogs (a running joke in
Italy), which will give a gross underestimate of any type of dog in a population. This is the only
study in Table 2 that dates from this century. All the other cited studies use data from before
1994. Again the AVMA doesn’t define “serious bites” the table supposedly includes. In fact,
many of the sources they cite include all reported dog bites. The estimates of “breed prevalence”
are based on dubious sources – few people license their dog; shelter admissions don’t reflect the
prevalence of a dog type in any population; Animal Control, shelters, and veterinary hospitals
aren’t acceptable sources of data on human injury rates and they often hide the fact that a dog is a
pit bull type. On top of this, these 10 sources include a humane society newsletter, a master’s
thesis at a veterinary college, and again almost half are written by pit bull advocates.
!
The use of these outdated sources to say “Look! Other dog breeds bite!” is misleading. The surge
in pit bull ownership (and the accompanying epidemic of catastrophic maulings) didn’t begin
until the late 1980s. The authors of this document furthermore repeatedly misrepresent what the
cited studies say.
!
Section Three: “Aggressive breeds”
!
Contrary to what this section claims, there is no valid data that says small and medium-sized
dogs are more aggressive than large dogs. Three of the four recent studies this section cites are
based on dubious data and were written by people involved in pit bull advocacy; the fourth uses
data from a foreign country. All the other studies cited use data from the previous century; four of
them use old data from non-US countries, and a fifth oldie is only about dog-dog aggression.
!
Section Four: “Pit bull types”
!
Contrary to what this section claims, there is strong jurisprudence and even an ASPCA
experiment to the effect that adult humans of average intelligence can easily identify a pit bull
type dog or a pit bull mix. There is no valid study that proves we can’t. There is no evidence that
pit bull attacks on children have anything to do with ‘breed prevalence’. Most attacks are on
household members where one or more pit bull types are kept. It doesn’t matter whether others in
the neighborhood do or don’t have pit bull type dogs – a population of one in your own
household is enough to cause disaster.
!
Section Four: “Breed bans”
!
This section is deceptive. They open with with a statement that “maulings by dogs can cause
terrible injuries”, then they cite a Detroit study that dealt exclusively with maulings by pit bulls.
It is deceptive to state that a study of specifically pit bull maulings shows what ‘dogs’ in
general do.
!
The 2008 study cited to emphasize within-breed variation was written by pit bull activists. It is
based on a public, uncontrolled and non-valid owner-report questionnaire. This 2008 study also
misrepresents behavioral variation within any given breed. In fact, breed-specific behaviors
vary only little within any breed or type of dog, while breed-specific behaviors do indeed
differ greatly among unrelated breeds or types of dogs.
!
Contrary to what this section states, population genetics and field data both tell us that the nature
of a ‘breed’ of dog does not vary over time. Population genetics show that breed-specific
characteristics, including behavior, remain in place unless there is strong active human selection
against those traits. There has never been strong human selection against the type of explosive
aggression that typifies the pit bull type. As for variation in time, there has been none – pit bull
types have killed more people than all other breeds and types of dogs combined for over 175
years now.
!
Also contrary to what this section states, the nature of a ‘breed’ does not vary geographically.
The dog trade is international – every ‘breed’ or type of dog represents a single, world-wide gene
pool. The pit bull type disproportionately kills more people (and animals) than all other
types and breeds of dogs in every country where it is kept.
!
As for alternative causes, the AVMA presents only outdated pit bull lobby talking points. Pit bull
type dogs and mixes of pit bulls claim about 80% of catastrophic maulings and fatal attacks in the
past three years (2013-2015), data from Animals 24-7 and DogsBite.org shows respectively. They
are only about 6% of the canine population. All the alternative causes for this that the AVMA
offers randomize out and must therefore be marked irrelevant background noise.
!
This AVMA document closes first with an outright falsehood. In fact, all valid data tell us that
breed or type of dog is the best predictor of the risk a dog represents with respect to
sustained mauling, disfiguring, dismembering and/or fatal attacks – which is what BSL is
meant to control.
!
The AVMA follows this falsehood with a truth: if BSL is to work, a cluster of dog ‘breeds’ will
have to be targeted. This follows from population genetics. The targeted cluster will consist of all
of the ‘breeds’ that are in fact members of the biological and genetic population descended,
directly or by mixing, from the bull-baiting, dogfighting types of the 17th century and were
widely used for those purposes until only some fifty years ago, when they began to be widely
marketed as family pets. The AVMA couldn’t resist following this truth with one more falsehood:
In fact, the GSD and other large dogs would not be targeted, since they are different genetic
populations altogether and disfigure, maim and kill at only a small fraction of the rate shown by
pit bull and war dog types.
!
Closing remark: Detailed analysis of this pamphlet is intended partly to help the public
understand how to see whether a document has any valid content. I am personally puzzled as to
why a veterinary association would choose to lobby for continuation of the immense animal
suffering – both for ordinary pets and for pit bull type dogs themselves – that BSL is successful in
preventing. I am puzzled as to why the AVMA would ignore the veterinary oath that compels it
and its members to guard public safety and animal welfare. I am puzzled as to why the AVMA is
willing to use such deceptive tactics as the ones displayed in this pamphlet. It’s unclear whether
this is a failure of ethics or a failure of veterinary education.
!
Reference: Literature Review on the Welfare Implications of The Role of Breed in Dog Bite Risk
and Prevention
!
Alexandra Semyonova
Newark, NJ
October 2016
!
Alexandra Semyonova is an internationally acclaimed animal behaviorist and author of The 100
Silliest Things People Say About Dogs. Academically educated in behavioral science and
specialized in animal behavior, she has worked with dogs and their owners on a daily basis for
more than 30 years. Visit her website at Nonlinear Dogs.
!
!
SUGGESTED BACKGROUND READING:
!
On Bullshit, by Harry Frankfurt at Princeton University, see especially pages 13 – 16, accessible
at: www.stoa.org.uk/topics/bullshit/pdf/on-bullshit.pdf
!
Betrayers of the Truth; Fraud and Deceit in the Halls of Science, by William Broad and Nicholas
Wade, Century Publishing Company Ltd, London, 1983.
!
Doubt Is Their Product; How Industry’s Assault on Science Threatens Your Health, by David
Michaels, Oxford University Press, New York, 2008.
!
While We Were Sleeping; Success Stories in Injury and Violence Prevention, by David
Hemenway, University of California Press, Berkeley and Los Angeles, 2009.
!
The Expert Witness Scam, by Leon S. Robertson, PhD, Yale University (retired), 2000, 2006.
!
Methods in Behavioral Research, by Paul C, Cozby, McGraw-Hill Companies, Inc, New York,
2007.
!