Dispatch alerts the K9 Unit that a homeowner sees a man fitting the description of the suspect in a nearby liquor store robbery: Caucasian, male, jeans, red baseball cap, and running into his barn from the direction of the robbery. Perimeter has been set up, and suspect is contained to the homeowner’s property. K9 unit prepares for a building and possible area search. The facts of the robbery include the suspect punching the clerk over the counter knocking him to the floor and continuing to hit the clerk unconscious, opening the register and removing the cash and fleeing out the front of the store. He knocked a woman to the ground as he fled who happened to be entering the store at the moment of the robbery causing her head to hit the pavement. The extent of the head injury is unknown at this time.
Choosing to deploy the K9 Unit to locate and possibly apprehend this unsearched, fleeing, felony suspect is a good choice. In the Supreme Court Decision of Graham v. Connor excessive force must be viewed under the Fourth Amendment “objective and reasonableness test.” This requires careful consideration of the totality of the circumstances, including the following three components to determine if in fact the seizure is reasonable:
- The severity of the crime at issue;
- Whether the suspect poses an immediate threat to the safety of law enforcement officers or others;
- And whether the suspect is actively resisting arrest or attempting to evade arrest by flight.
Further, the court determined that the reasonableness of an officer’s use of force must be judged from the perspective of a reasonable officer on scene, in the moment, rather than with the 20/20 vision of hindsight. In our example, we have circumstances that fit the use of a K9 to make an apprehension according to Graham v. Connor. Felony assault and battery, robbery, evading arrest by flight and an immediate threat as evidenced by the violence perpetrated on two people during the robbery. Let’s assume the K9 unit deploys to conduct a find and bite operation in the barn. Announcements are given on the PA system at proper intervals according to policy, and the hidden suspect is given time to make himself known and come out, otherwise be bitten. If the suspect gives up without sending the dog in to bite, we have an apprehension and no bite. If the suspect does not give up and the dog goes in and finds and bites, we have an apprehension with a bite. However, if the dog were a find and bark dog, there would be a third possibility (and hence whey we assume find and bark dogs are lower force dogs) that is the dog goes in and searches and encounters a suspect out in the open who is still and gives up. This would be an apprehension by hold and bark, and no bite. Statistically the same as a suspect walking out and giving up without the dog even searching for him.
Bite Ratios
The debate over training dogs to find and bite vs. find and bark has been going on for quite some time. The key statistic used both by courts and K9 unit supervisors to separate properly functioning patrol dogs from misbehaving ones is the bite ratio. The bite ratio is defined as the number of apprehensions with bites divided by the total number of apprehensions. Bite ratios are generally required to be less than 30%. The only research that has been done on the subject of bite ratios and how it relates to find and bark vs. find and bite K9s came in an academic paper published in by Dr. Charles Mesloh (Police Practice and Research 7(4):1561- 42631477 · September 2006). His regression analysis research from survey sampling K9s in the state of FL indicated that dogs trained to find and bark had higher bite ratios than dogs trained to find and bite. This would seem to be counter intuitive and does not support the recommendations of the DOJ or the IACP both of which have suggested (without any prior research) that K9s should be trained to find and bark to limit liability and reduce bite ratios.
Many agencies have adopted the find and bark, and normally the reason cited is to bring down bite ratios and limit agency liability. However the only scholarly research that has been done indicates the opposite conclusion. Politics and perception will often dictate policy even in the face of contrary evidence when it comes to policing. Unfortunately it is not just what is going on but how people feel about what is going on. Find and bark seems to be kinder and gentler.
Decision Making & Risk Analysis
All handlers must use the calculus of Graham v. Conner in deciding whether to send their dog to make an apprehension. Even the handler handling the find and bark dog should assume the dog will get a bite and that the reasoning behind sending the dog in on that search to make the apprehension is sound. The main difference between and find and bark trained dog and a find and bite trained dog is in the moment the dog encounters the subject. If the dog encounters a subject still moving (evading or attacking toward the dog) the dog will bite. At the moment of encounter, if the subject is stationary and passive, the dog should, as its training dictates, closely guard and bark at the subject. Some trainers will teach the dog to closely guard and bark, and some trainers teach a “circle and bark” so the dogs keeps a measure of distance from the subject to avoid stabbing weapons that could be brought to bear on the dog if it is up close in the guarding.
So the real difference between find and bark and find and bite in a building search which ends on a passive subject is that the dog makes the determination based on the context to either make the apprehension (bite) or rather hold and bark (detain), essentially leaving the use of force up to the dog in those final seconds. Many canine officers abhor the idea of the dog being saddled with that judgement based on the context of “movement” alone. Faced with the prospect of a dog bite in the moment, a suspect may decide that he wants no part of it, and so stop and give up. Turning away, or curling up to protect oneself will only provide the movement the dog needs to see to make contact and bite. Screaming, jerking, moving hands up at the last second can all trigger a bite. The dog is unable to take social cues so to him any movement is fair game, even a submissive gesture such as dropping on the ground. Thus people wanting to give up may be bitten because they aren’t sure how to behave when the dog contacts them.
Passive Threats
This is the other side of the ignorance coin. It is silly to assume that just because the subject is passive in a building or an area search that the subject is not a threat. In fact one of the major criticism of find and bark is just that, passive subjects are still a threat. We cannot operate in law enforcement as if nobody understands our training or its limits. Human suspects that encounter police dogs will learn over time that if they stand still the dog will not bite, and further because we must create a margin of neutrality to incidental movement some slow, cautious movement can be made without triggering the dog to bite. Subjects are rational actors, in other words they can learn how to slowly move away from the dog to slip through a door. How often this can or does happen is unknown, and most cases are annecdotal. More concerning is that use of force can be brought to bear on the K9 while he performs the hold and bark. Slowly producing a gun will also not trigger the dog to bite, and if the dog is close enough a knife will do well enough, and even a blast of dog mace could allow a subject to get away.
Proponents of find and bite say that when they send a dog in to do its location and apprehension function, satisfying Graham v. Conner prior to the search, nothing is left up to the dog. Passive or active, the suspect being sought will be bitten and apprehended accordingly. The decision is made in advance. Building and area searches are dangerous enough for the officer, and at least if the suspect is accessible to the dog upon making contact the handler will know that the dog is occupying the subject with force as he approaches in the case of find and bite.
Why Higher Bite Ratios?
All this begs the question, “why do find and bark dogs have higher bite ratios?” I think there are a lot of reasons we might find this. Mesloh himself suggested that handlers with find and bark dogs may operate in riskier territory as far as satisfying Graham vs. Conner. Because the dog is perceived as operating at less than certainty for a bite, the handler over predicts this level of safety and the dog is sent on more searches and thus achieves more bites. This is known as a “moral hazard” problem in the economics and psychology literature. It’s like feeling safer to drive faster because seat belts and airbags will significantly lower the personal cost of a crash. The find and bark training is assumed to lower the likelihood of a bite, therefore more latitude is given to the find and bark dog over the find and bite dog that we know will bite if released on a subject. The find and bark training is interpreted as insurance.
Mesloh also suggested that some of the dogs may have been switched from find and bite training to find and bark in the hopes to lower bite ratios, and as a result these misbehaving dogs with high bite ratios simply carried their misbehavior over to the new paradigm, being less reliable in the new behavior due to the previous training.
I suggest another reason for the findings. It could be that find and bark dogs are implemented in areas where there are a high number of canine/suspect encounters, and because it is assumed that this training mitigates bites at the end of searches, these type of dogs are just more present in areas where bites are more common per deployment because of the severity of crime and the experience of the police force in proactively using K9 resources. Crime rate variables could help us control for this explanation in further studies.
There are also two little known and cited conclusions found in the paper. The first is that the variable identified as “prior schutzhund training” had a significant effect on lowering bite ratios. Schutzhund training is simply a high level of sport training that results in a more controlled animal. It would be interesting to see if “professionally trained dogs” vs. “Academy trained dogs” resulted in any significant prediction of lower bite ratios. I would also be interested to see if “pre-titled dogs” (IPO, KNPV, Ring) were significant predictors of lower bite ratios.
The other conclusion which is often not mentioned is that the breed of the dog was a significant determinant of higher or lower bite ratios. Belgian Malinois predicted higher bite ratios, and German shepherds predicted lower ratios. However, agencies may often opt to use Malinois in areas where suspects are generally more aggressive in fighting officers and K9s, so that while correlation may be there, causality may run in the other direction, because of agency selfselection. Malinois are put in areas that will likely be higher crime and require a little more horsepower and as a result we have higher bite ratios.
Conclusion
It has been 10 years since this initial study was done, and it is long past time where another follow up study should be done with a larger sampling of agencies (and some variation in independent variables) to see if the same conclusions stand the test of time, different statistical models, and apply in areas other than FL. Both statistical predictive methods and data collection techniques have improved in the intervening time. Also, some of the variables used in the study, i.e. whether the dog was trained using an e-collar or a “bungee” are largely irrelevant to the conclusions. Other more salient independent variables could be chosen which might have a marked impact on the conclusions. It would also be interesting to see if these
conclusions hold across different samplings of agencies in different geographical areas as well as for the totality of the US and Canada as well.
If find and bark is confirmed to lower bite ratios, technology has improved significantly to where camera systems, though expensive, can come into play to help guard against the K9 and handler safety issues that come along with the hold and bark deployment. If the handler could see on a screen the encounter in real time as the dog comes upon a suspect, remote commands could be given to the dog to engage if a suspect was slowly moving to evade or attack the dog while doing a hold and bark, or reinforce the hold and bark command in the moment.
However if new research confirms the overall conclusions Mesloh drew in his 2006 paper, a radical rethinking of the find and bark deployment should be made. If we are thinking this training will lead to lower bite ratios, and it just isn’t borne out by the facts of statistical analysis, the training should be abandoned, and we should leave handlers and dogs, properly trained to understand their jobs, including the appropriate use of force procedures, to make the determination as to whether a K9 bite apprehension is warranted or not in a find and bite paradigm. We need to operate on facts rather than on what “seems” right.