Animal behaviour / Science Education / Zoology

Aggressive dogs, poor training techniques, sound statistics and some good advice in the discussion – shame about the methodology.

ResearchBlogging.orgI came across a new study the other day, recently published in the journal of Applied Animal Behaviour Science

Herron, M., Schofer, F., & Reisner, I. (2009). Survey of the use and outcome of confrontational and non-confrontational training methods in client-owned dogs showing undesired behaviors. Applied Animal Behaviour Science, 117, 47–54.

dog_behaviourIntroduction and Methodology

The study looked at the responses of domestic dogs (Canis familiaris) to “interventions”—things done directly or indirectly to the dogs by owners and/or handlers—with the intention of reducing aggressive behaviour. Interventions themselves could range from the physically confrontational (e.g., kneeing the dog in the chest if it jumps up), to rewarding positive behaviours with treats. In the researcher’s own words…

The purpose of this study was to describe the frequency of use, the recommending source, and the owner-reported effect on canine behavior of interventions that owners of dogs with undesired behaviors had used on their dogs. This study also aimed to report aggressive responses from the dogs subsequent to the use of aversive and non-aversive interventions.

I like what this study is attempting to do – laudable and much needed. As an animal behaviourist and someone who started training dogs with his mother before he went to primary school, I’ve seen my fare share of ignorant and downright dangerous (to animal and human), training techniques (e.g., suspending an aggressive dog consisted of, basically, hanging it for a few seconds on the end of a “check chain” and lead). This study tries to point out how such techniques, animal welfare issues aside, can be dangerous for handlers, just don’t work and could even be counter-productive.

Great, so far so good – you can tell there’s a “but” coming can’t you?

Owners of dogs scheduled for an appointment with the Behavior Service at the Matthew J. Ryan Veterinary Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania, USA, were sent a survey via email, fax, or postal mail designed to identify and briefly note the behavioral outcome of a variety of treatment interventions.

This survey was pre-tested for clarity by 10 dog-owning hospital employees.

Okay, let’s say that the dog-owners who were sent the surveys had the same level of understanding as the 10 dog-owning hospital employees, and that they gave more than a split-second’s thought to filling out the survey (be honest, how much effort did you spend on filling out the last questionnaire you completed?).

The treatment interventions are noted in table 1, below (click on tables for enlarged versions)…

Table 1:


Once again, good, clear stuff – techniques I’ve seen used thousands of times over the years, all with varying degrees of success.

Anecdote Alert – Anecdotes are NOT evidence: The next time you watch a world-class, dog obedience championship (if you ever do), take a long, hard look at the dogs in the finals. They will all look like they are thoroughly enjoying what they are doing. Any trainer worth a dog choc, will tell you that training by fear and intimidation doesn’t get good results.

Anyway, back to the study… so here we have owners answering survey questions on their own efforts with their own dogs. And herein lies the rub.

Take a look at table 1, how willing do you think people would be to admit that they had “hit or kicked” their dog? That they had “kneed [the] dog in the chest for jumping”? That they had “abruptly jab[bed] the dog in the neck or side”?

If they did admit to such techniques, one person’s kick is anothers “push with their foot”.

When the word “muzzle” is used, does that mean one of the old-fashioned muzzles or one of the myriad of new devices with “kinder”, “softer” names? Are they muzzles?

What did the researchers mean by “yell” exactly? Or “growl at the dog”? I didn’t growl per se, I just lowered my voice a lot and sounded angry.

The people filling out these surveys were lay people, not scientists with clear descriptions and definitions for their terms. The respondents didn’t have a standardized template of what these interventions were, in order to compare their own behaviour and make a decision as to what constitutes a yell or a kick.

We have no way of knowing how accurate the responses to this survey were. Which unfortunately makes all that follows a bit of a wasted effort really.

The aggressive behaviour of the dogs (complaints)…

were then categorized as follows: ‘‘aggression to familiar people,’’ targeted to household members or people with whom the dog spent significant time, ‘‘aggression to unfamiliar people’’ targeted to non-household members, ‘‘aggression to dogs’’ if owners described their dogs as aggressive to dogs either within or outside the household, ‘‘separation anxiety’’ if the dog exhibited problems in the owner’s absence, ‘‘specific fears or anxiety’’ if the owners described fear of noises or other environmental stimuli, such as thunderstorms, and ‘‘other’’.

Once again, much of the above categorization is extremely subjective and based upon the opinions of the owners. Here we run into the anthropomorphism of pet owners and the fallibility of human memory, as well as simple accuracy of reporting original data.

In each case, owners were asked whether they had attempted the technique or intervention, the recommending source, whether the method used had had a ‘‘positive’’, ‘‘negative’’ or ‘‘no effect’’ on their dog’s behavior, and whether or not it elicited a ‘‘growl/bare teeth’’, ‘‘snap/lunge’’, or ‘‘bite’’ from the dog. For purposes of analysis and because any display of aggression was considered a safety risk to the owner, the responses ‘‘growl/bare teeth’’, ‘‘snap/lunge’’, and ‘‘bite’’ were collapsed into one ‘‘aggressive’’ response.

Once again, owners are asked for the data… “Now let’s see, who was it told me to kick the dog when he attacks uncle Bob? Was it the trainer or did I get that one off the TV?” And of course, this doesn’t account for their partner who, having never liked uncle Bob to start with, encourages the dog to bark at the door whenever the poor guy shows up.

What I’m saying of course, is that there’s no real control mechanism here, we don’t know for certain if there are other factors affecting the answers given by the respondents.

In addition, owners are then asked to judge the responses of their dogs. This is all highly subjective.

Sure, we could use it to base a hypothesis on and conduct further research… (wait for it – you can just bet there’s a bit in the discussion about that one).

Fisher’s Exact Test was used to determine if dogs presenting with an aggression to familiar people, and dogs presenting for aggression to any people (either familiar, unfamiliar, or both) were more likely to respond aggressively compared to dogs presenting with other behavior problems.

Significance levels for multiple comparisons were adjusted for using the Bonferroni correction. A P-value of < 0.002 was considered significant.

Good, standard methods of analyzing contingency table data and tightening up significance levels. But is the data valid in the first place?

Results and Discussion

Between April 1 and July 31, 2007, 30 (28%) of 107 distributed surveys were completed and returned.

Between August 1, 2007 and May 1, 2008, an additional 110 completed surveys (98% of 112 distributed) were collected, for a total of 140 completed surveys (64% of the total distributed).

One-hundred and forty respondents is a good number, but it’s stretched out over three months – once again I’d be concerned over issues of memory and accuracy of responses.

Also, I wonder about changes in temperature and environmental stimuli (e.g., frequency / duration of exercise) and how this might affect aggressive behaviour in pet dogs?

Please guys, don’t send out any more surveys – when your dog bit the local cop, was it a hot day or was it a bit cold?

Owners attempted a variety of behavioral interventions, many of which elicited an aggressive response, with their dogs prior to their appointment with a referral Behavior Service. As we expected, the highest frequency of aggression occurred in response to aversive interventions, whether direct or indirect. In contrast, reward-based training elicited aggression in very few dogs, regardless of presenting complaint.

Table 2 (below), details opinion of the respondent dog owners on the behaviour of their dogs, following the interventions identified (click on tables for enlarged versions).

Table 2:


The table shows that “reward-based” interventions generally have a higher “positive effect” – in the opinion of the dog owners. Nobody else, and, sadly, certainly NOT as indicated by objective, scientific observation.

The data analyzed in this study is not conclusive. The methodology for collecting the data was flawed and the analysis which follows is therefore equally compromised.

This is certainly not news to the researchers, who deal with many of my criticisms (plus one or two more) as follows…

There were several limitations in our study. First, the dog owners surveyed were recruited from a population of owners making appointments at a referral behavior clinic; in many cases, the behavior problems were significant.

The frequency of aggressive responses and effectiveness of training methods might have been different if we had sampled a general population of dog owners.

Next, the survey did not request a temporal description of these interventions and many of them may have been applied well before the presenting behavior problems occurred.

It is, therefore, difficult for us to determine whether owners attempted specific interventions to alter aggressive behavior or whether aggression developed as a result of their use.

It is also possible that owners misinterpreted the meaning of the ‘‘effect’’ section of the survey.

The terms ‘‘positive’’, ‘‘negative’’, and ‘‘no effect’’ are subjective, and judging a technique’s effectiveness based on these options may not be accurate.

Not to mention the fact that the definitions of the interventions themselves (was it a “kick” or a “push”?), may have been unclear to some respondents.

Next, owners’ self-reporting may have led to recall bias and/or poor answer reliability.

For example, each owner may have remembered the outcomes of various treatment techniques differently and some owners may have felt reluctant to admit to a veterinary professional that they used physically aversive methods on their dogs.

Finally, the retrospective nature of the survey prevented the possibility for direct comparison of safety and efficacy between aversive and non-aversive techniques.

In other words, because the survey was conducted by having respondents remember interventions and results, it was not really possible to judge whether the results reported were accurate (i.e., positive, negative or no effect).

It would, however, be unethical to put dog owners at risk for injury for a randomized, prospective comparison between the two categories.

Dog owners possibly, but what about professional dog trainers? This study has a good premise and leaves a lot of important questions unanswered – this would be great stuff if it were part of a research proposal, rather than a published study in its own right.

I just don’t feel that the opinions of the dog owners constitutes real evidence and therefore data.

But this isn’t a criticism of the researchers, I for one want more from these people – they are breaking new ground!

This study is the first of its kind to investigate several commonly used behavioral interventions and the potential for aggression as a result of their use.

A larger scale study with a more general population of dogs would be the next step towards evaluating the effects of the various behavioral modification techniques and their associated risks.

Sounds great. I’m hooked and am awaiting the publication!

Who’s got the film rights?

In conclusion, confrontational or aversive behavioral interventions applied by dog owners before their pets were presented for a behavior consultation were associated with aggressive responses in many cases.

Owners of dogs aggressive to family members are especially at risk for injury—and their pets at risk of relinquishment or euthanasia—when certain aversive methods are used.

Ultimately, reward-based training is less stressful or painful for the dog, and, hence, safer for the owner.

It is important for primary care veterinarians to advise owners about risks associated with aversive training methods, despite their prevalence in the popular media, and to provide resources for safe and effective management of behavior problems.

Now, I’m not sure what it’s like in the United States, but over here in the United Kingdom, veterinarians are not necessarily animal behaviourists, and are just as capable of delivering some truly dire advice as any TV pundit you’ll ever tune into.

This is a much needed area of research – for both professionals and a lay audience.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s