Practically all animal shelters in Australia administer some kind of ‘temperament test’ to pets in an effort to determine their suitability for rehoming. In the in case of dogs, these tests are purported to be able to reveal dogs who are aggressive and a potential danger to prospective families. They are also used as a ‘personality’ appraisal, to match dogs with their new families.
Failing a temperament test is the single largest killer of shelter dogs in this country. Of the 14,211 dogs killed in RSPCA shelters nationally (2011/12), 10,702 of these were for behaviour. Nearly 40% of dogs tested failed the test and were killed. At individual RSPCA shelters, the number can be as high as 60%.
While these tests are used to defend high kill rates, many tests have been developed without even the input of a canine behaviour expert. There is little standardisation and every shelter does their testing differently. Historically, there has been scant scientific validation of these screening processes. In short, temperament testing has been more of a black art, than a science.
Dog temperament tests tend to require the tester to look for ‘friendly’ and ‘non-friendly’ behaviours. How happy the dog is to be touched and stroked, and its reaction to other animals. Many use a toddler doll to test for aggression towards children. There are tests claimed to reveal dominance, reactivity, resource guarding and food aggression.
The main known obstacle with these kinds of appraisals is accurately quantifying the results. Every single human assessor will interpret the behaviours shown differently. Even when assessors are given a detailed ‘scoring’ system, the enormous spectrum of dog behaviour and responses, mean the test results will potentially vary from day to day. Add the pressures of a high-volume pound (noise, stress and a high turnover) and a lack of staff training, and there is an enormous opportunity for misuse:
Monash University research casts further doubt on the integrity of temperament testing, finding almost one-quarter of shelter staff who conducted the assessments received no formal training in them, and only 56 per cent believed they were given enough time to assess behaviour.
The study, published in 2010, involved 11 shelters and pounds in several states, including NSW.
It concluded that although most shelters did their best with limited resources, staff could be deciding the fate of dogs based on ”inadequate training, potentially invalid assessment protocols and subjective interpretations of behaviour in which they have limited confidence”. It called for a stronger scientific and regulatory approach.
Most temperament include an assessment for food aggression or resource guarding (sometimes using a fake plastic hand) by petting the dog while it eats and moving their food dish. The dog growls, lunges or bites, they fail the test… and are most likely killed.
In September last year a paper was published entitled ‘Food-related aggression in shelter dogs: A comparison of behavior identified by a behavior evaluation in the shelter and owner reports after adoption’.
Scientists tested a group of 97 dogs using a standardised behaviour evaluation that included a fake hand test for food aggression. They found that only 55% of the dogs who objected to the hand, went on to show food aggression in their new homes.
These results potentially mean that in nearly half of all cases, dogs who would have failed their temperament test and would have historically not made it out of the shelter alive, are likely not displaying actual ‘food aggression’. Instead the test is maybe showing ‘fake hand in a stressful pound’ aggression. Much less useful.
– A great examination of the full results of these tests can be found here.
– One of my favourite satirical pieces on the same topic is here.
Australian researchers have been working with canine experts to develop a science-based protocol for assessing dogs. Their results were published in November, in the paper Development of the behavioural assessment for re-homing K9’s (B.A.R.K.) protocol.
The BARK program was a series of tests was designed to assess five primary behaviour traits: anxiety, compliance, fear, friendliness, and activity level. It was hoped that a “a scientifically validated and standardised behavioural assessment protocol for shelter dogs” would give shelters a greater ability to accurately assess dogs.
An effective test would;
– see different human assessors agree on the results
– have the dog receive a consistent score over time, and
– accurately predict the dog’s behaviour in the home
If the program could achieve these aims, temperament testing could finally be considered a valid tool for shelters to use to manage their animals.
Five Australian animal shelters agreed to participate in the BARK validation study. Workshops were conducted, during which at least two staff members were trained in administration and scoring of the BARK protocol. Shelters were asked to assess as many dogs as possible, and to invite adopters to participate in the follow up study.
But unfortunately, data collected by the participating shelters was “extremely poor”. Two shelters did not collect any data. One shelter assessed just one dog. Another assessed two dogs. The highest data collection rate from a participating shelter over 6 months was 13 dogs.
“Follow up telephone calls and emails to the shelter staff offering assistance and support did not improve the rate of data collection and it was decided that the investigators would have to collect the data themselves to ensure that sufficient data were available for analysis.”
So researchers used their own assessors for the test. Both assessors were experienced in evaluating canine behaviour. 48 dogs were assessed on two separate occasions, with all dogs were given at least 3 days to acclimatise to the shelter environment.
– Researchers found that using the protocol, they both rated the dogs traits (anxiety, compliance, fear, friendliness and activity level) similarly. However,
– the dog’s score did not stay stable over time and results saw an inconsistency of behaviour, resulting in poor correlations, between the two assessments.
“The general weakness of the correlations is somewhat surprising considering the two assessments were conducted just 24hrs apart and by the same, experienced, assessor in exactly the same environment.”
– The new owners of the adopted dogs were assessed, with 67 dog owners participating. These results were compared with the dogs’ in-shelter BARK scores. The overall predictive value of the BARK test was found to be poor. Of the five behaviour categories just two (fear, friendliness) had a moderate correlation between in-home reports, and the shelter BARK scores. There were no correlations for the traits of anxiety, compliance, and activity level.
Given these kinds of tests are supposedly being used to predict future behaviour in the home, a result that they don’t even predict behaviour from one day to the next is startling.
Researcher concluded that “further research is urgently required” to investigate the value of shelter temperament testing overall, “especially since the outcomes of such tests often determine which dogs are made available for adoption or whether euthanasia is recommended”.
It is known that dogs entering shelters are traumatised. The capture, transport and confinement, the loss of their family and companions, extremely unsettling surroundings, contagious and aggressive barking and loss of control over their environment, all see dogs behaving in inconsistent ways.
Added to differences among shelters in terms of staff experience and training, time availability and the number of animals processed – not to mention the behaviour and knowledge of the new owners – and it’s not hard to see why the holy grail of developing an accurate temperament test is such a challenge. That’s if it is even possible to design a test that reliably indicates a dog’s future behaviour at all.
Despite advice from a team of canine experts, and the utmost care being taken in the design and implementation of the protocol, such poor performance by the BARK program, shows an enormous deficiencies in temperament testing in shelters. This has huge implications to shelters using, by all accounts inferior and unreliable tests, to kill dogs every single day.
Given that dogs literally live and die by these tests, the need for them to be accurate is critical. We cannot justify continuing the systematic killing of pets under the current flawed temperament testing model. It is time to rethink the entire application of these tests and, given evidence that even with great effort they may never be made reliable, to consider not using them at all.
I can tell you from personal volunteering experience, this is exactly how the Townsville QLD RSPCA runs. Hackles up when a strange dog approaches? Euthanized. Won’t submit to physical manipulation into a sitting position? Euthanized. It’s so sad.