Editor’s Note: Learn more about WDDO in our March issue profile on the author, online here.
If you’re not familiar with the phrase “randomized, double blind” testing, it basically describes a scent detector dog team proficiency test method, wherein no one present during the test — including the tested handler — knows the solution to the testing problem. In randomized, double blind testing, the placement of hides and the number of hides used is randomly chosen for each test setup.
In my opinion, based on my decades of experience in working with, testing and training detection dogs for police and fire rescue work as well as mold and pest management, the randomized, double blind testing represents the fairest and most objective test available for evaluating the proficiency of detector dog teams. By comparison, in traditional detector dog certification tests, the test monitor(s) always know how many hides are present and know exactly where they are located. In many cases, the monitors are two individuals who are present in the test area during testing, in addition to the tested handler. The location(s) of hides is not randomized in traditional testing. Therefore, recognizable patterns of the placement of hides, by testing officials, run the risk of becoming obvious to repeatedly tested handlers.
In traditional test set-ups, tested handlers are informed as to how many locations they must correctly identify as containing odors (how many hides were placed) in order to be successful, thereby revealing a significant part of the solution to the testing problem. The problem is that persons with this knowledge, who are present during testing, can introduce bias by unintentionally, or possibly intentionally, offering subtle cues to tested handlers and to tested dogs. For example, he or she might subconsciously choose to stand in the (usually small) test area. He or she might also pointedly move away from or block access to certain areas or pieces of furniture during a test. He or she might even stand in the doorway of a room until the handler has successfully found the hide, or pointedly keeps the doorway clear when the searched room is blank.
Other cues can include subtle changes in respiration rates by the monitors in the room, such as subtle gasps when the dog approaches an area of interest or sighs of relief when the dog moves away from a blank area without false alerting. Even facial expressions from test monitors can represent subtle cues. Any of these could help or hurt the team’s chances of success.
Next week, I’ll discuss research that shows other ways that we can inadvertently offer cues to tested dogs and handlers. In the meantime, if you’d like to learn more about the non-profit organization I founded, the World Detector Dog Organization (WDDO) and randomized, double blind testing, please visit WDDO.org. You’re also invited to join WDDO as a member and attend our April 5-7, 2018 conference, the details of which can be found here. If you have comments or questions about this post, please sound off in the comments section or email me directly at latimer_david@yahoo.com.
I agree in principle of the benefits of this type of testing but wonder how do you keep from rewarding the K9 during a false positive? If the dog indicates the presence of the odor and the handler doesn’t know if it is actually there or not, couldn’t the handler reward the dog for a false indication?
Hi Jim,
Thanks for your question, it’s good one. First, let me ask you, how do you decide when to reward when you and your dog are searching you can do the same thing during the test. If you’re working in the field, I’m sure your dog is reliable enough that you aren’t overly concerned about the dog false alerting, so you aren’t likely to be rewarding on a false alert anyway, right?
Now, if you are rewarding your dog on a variable reward schedule, as most trainers and canine behaviorists recommend, then not rewarding on a couple of alerts won’t be anything out of the ordinary for the dog. If you aren’t rewarding on a variable schedule, just start skipping a reward or two once in awhile during training and your dog will think nothing of it happening during the test. I’m going to talk more about variable rewards in upcoming posts, so stay tuned.
One thing you could do, if you’re really concerned and neither of the above seems right for your program, is ask the test administrator if you could bring your dog back into the scenario after you and all the other teams have completed testing, so that you can correct any erroneous rewards that you think you may have offered during the test. Most officials won’t have a problem with allowing you to do that if time isn’t a great factor.
Given your concern and the fact that you know enough to ask that question, I’m assuming that you train regularly and that your dog isn’t very likely to false alert anyway. I always made a practice of avoiding rewards on deployments unless I knew the target was present and I teach handlers the same thing. But, again skipping a couple of rewards, once a year, even if you don’t do that normally, isn’t going to negate all the quality training you do through out the rest of the year.
Dogs should be trained using the variable reward system. When they have been trained that way, withholding a reward doesn’t harm the dog at all.
Rewarding a false alert always hurts the dogs training.
Hi Jay,
You’re right, rewarding on false alerts can cause problems and I’d avoid that as much as possible. During training, when a new conditioned stimulus is being taught, I recommend treating for each correct response. Once the response to the command is reliable, reward delivery should be randomized. Randomizing rewards, in my experience, makes for a more energetic and focused search dog.
Great stuff
Thanks, Chris