Shelters often use behavioral tests to discover what a dog or cat’s personality is like. If that animal is good with kids and other dogs, if they will growl when food is taken away, if a dog reacts poorly when a plastic hand is shoved in their face.
And sometimes, depending on a shelter’s resources and commitment — like if there is a behaviorist on staff, or a robust foster program — these tests may determine if an animal gets the chance to be adopted, or if they are euthanized.
The use of temperament evaluations to make these life and death decisions is increasingly coming under scrutiny.
For example Kristen Auerbach, deputy chief animal services officer at Austin’s city shelter, the Austin Animal Center, has gotten a lot of attention this last year for her study showing that dogs who show behavioral problems at the shelter act much differently once they get into foster homes.
Now a new paper, published last month in the Journal of Veterinary Behavior, makes no bones about the authors’ perspective, as you can tell from the title: No Better Than Flipping a Coin: Reconsidering Canine Behavior Evaluations in Animal Shelters.
The authors — veterinarian Gary J. Patronek and Janis Bradley, director of communications and publications for the National Canine Research Council — make the case shelter behavioral evaluations are arbitrary and unreliable, and are indeed “not much better than flipping a coin” when it comes to predicting a dog’s future behavior in a home.
These inaccurate prognostics not only produce unreliable information, which may be troubling for its own sake — but also lead to good dogs, who could make great pets, being killed in the name of a false safety.
How, then, should shelters make sure they aren’t sending dangerous dogs into the community — or even just making bad matches between pets and families? Shelter Me caught up with Bradley by email to find out this and more.
Shelter Me: Why is it important if shelter behavioral evaluations are accurate?
JB: Because they are used to determine whether it is safe to make a dog available for adoption. There are other uses, of course, but this is the one Dr. Patronek and I examined in this paper.
If the test gives the wrong answer, either the dog is not made available and usually dies when he never would have hurt anyone, or a dog who is inclined to hurt someone is sent out into the community.
The first of these unhappy outcomes is by far the most likely, as we know that only a very small percentage of dogs in general — less than 1% a year — ever hurt anyone, even to the point of a Band-Aid scratch.
Assuming that shelters are doing the best they can to make sure they aren’t adopting out dangerous dogs — how should they go about keeping their communities safe, if not by using behavioral evaluations?
By engaging the dogs in as many real life activities as possible — going for walks, playing with toys, playing with dogs, training games — then paying attention to and recording what the dogs actually do during their stay in the shelter, and by collecting careful behavior histories from the people surrendering the dog when this is possible.
Of course, nobody wants to send dogs out into the community who then hurt someone if this could have been predicted. The difficulty is that there is no practical way to use a behavior evaluation to make this kind of prediction.
How did you figure out that these evaluations aren’t accurate — no better than flipping a coin, as you put it? What makes the tests so inaccurate?
We can’t even really establish how inaccurate they are. There have been few attempts to compare shelter evaluation results with real life behavior in a home and since even those have already excluded from adoption the dogs who “failed” the test, there isn’t any way to really validate them.
What we can say is that if you want to predict normal behavior you have replicate normal situations and the artificial contrived context of a shelter behavior evaluation can never come close to this.
But even if there were a way to test the tests, and in the very unlikely event that a test could be shown to have a high level of validity — by the standards, say of human diagnostic tests — at the very least half the dogs who “failed” would actually be misdiagnosed and would not have shown problem behaviors in a home.
This is simply a statistical reality any time you are screen for something — in this case problematic growling, snarling, snapping, and biting behavior — that only occurs in a small percentage of your population. It’s why we included links to easily accessible online calculators so people could run the numbers themselves.
I’ve seen a lot of people in animal welfare and sheltering talking very excitedly about your paper. It seems to confirm what a lot of folks already believed, or suspected. Do you think that it’ll lead to change — or help lead to change?
As someone who just co-authored a paper on the difficulty of predicting dog behavior, you can hardly expect me to go out on a limb predicting human behavior!
But seriously, of course when you write a paper, you hope it will lead to positive change or at least initiate a discussion about possible change. We certainly hope this paper will lead to a re-thinking of this practice and are very gratified that it appears to be resonating with many shelter folks.
A year from now, five years from now, do you think that shelters will no longer be doing behavioral evaluations like we see today?
I think I’ll stick to hopes rather than predictions. I hope that shelters will have developed procedures to refine their abilities to accurately observe interactions between the dogs in their care and the people who care for them and to structure that care in ways that facilitate positive interactions.
And I certainly hope that obviously friendly outgoing dogs will no longer have to wait to be processed through an evaluation system before moving on to a more behaviorally healthy environment, i.e., a human home. And that the ones who seem scared can begin to get reassuring treatment immediately, based on how they actually interact with people.
This whole discussion reminds me of something Dr. Amy Marder, one of our National Canine Research Council expert advisors once said. To paraphrase, she said that dogs in shelters are just dogs who don’t have an owner standing by their side.
I think it’s worthwhile to keep this in mind if we find ourselves thinking that “shelter dogs,” require some sort of extraordinary scrutiny before they can move from one home to another.
This interview has been edited for space
Featured image via Flickr/Terrah, used under a Creative Commons license