InSurOp-Ed: Insurance Dog Bite Claim Data Doesn’t Pass the Critical Thinking Test

After Agency Checklists published its article, “Statutory Report To Legislature Outlines Top Dog-Related Home Insurance Claims By Massachusetts Insurers,” concerning the Division of Insurance’s report to the Legislature under the “Protect Animal Welfare and Safety in Cities & Towns” Act, Stacey Coleman, the Executive Director of the Animal Farm Foundation, wrote us to share her thoughts on the claim data contained in the Division’s report. The following is the “Letter to the Editor” that she wrote to share with our readers:

In an effort to provide Massachusetts legislators with good information to guide decisions regarding regulating insurance company policies, data on dog bite claims has been collected by insurers. Unfortunately, the data selected has little hope of providing such guidance.

The Massachusetts Division of Insurance (DOI) mission statement includes the laudable goal of “protection of consumer interests . . . by providing accurate and unbiased information so consumers may make informed decisions and by intervening on behalf of consumers who believe they have been victimized by unfair business practices.” Insurers have been required to provide information on dog bite related claims to the DOI by The Act to Protect Animal Welfare and Safety in Cities and Towns in a well-intentioned attempt to fulfill the Division’s goal. The Division has tabulated and reported to state legislators on the resulting data, seeking to establish risk factors for dog bite claims, and to address questions, given the mission of the Division, such as whether the practice of excluding homeowner’s policy holders whose dogs are identified as members of particular breeds constitutes an unfair business practice. Since this is housed in a act whose basic purpose is to protect animals, we can also presume that the legislators’ intent includes keeping pets and their owners safe, particularly with regard to preventing dog bites.

A recent AC article provides an accurate and neutral overview of the data pulled by insurers from dog bite liability claims in response to the Act’s requirement. Unfortunately, however, the data as defined and collected, cannot reasonably be expected to help inform consumer decisions or guide policies that will protect them from unfair practices, nor can it help legislators to make informed decisions in the spirit of the Act.

The fatal flaw dooming all the data collected here is that it is of no help in assessing risk. Risk is by its nature comparative. But information about the policy holders who file claims when they face damages when their dogs have bitten someone by itself can tell us nothing about risk. It doesn’t help us to know whether they took their dogs to obedience classes, or had their dogs spayed or neutered. Records of the location and circumstances of the incident, or demographics of the victim tell us even less. The only way that such data could be even hypothetically useful would be to compare it to the same data collected from a representative sample of dog owning policy holders who do not file dog bite claims. So, for example, if more people who had no dog bite claims provided obedience training for their dogs than people who did have claims, then we could say that neglecting obedience training is a risk factor for dog bite claims. But if the 2 groups trained their dogs in about the same proportions, it’s no factor at all. And if more of those with claims trained their dogs, it could even mean that not training could be a protective factor. Without the comparison between the 2 groups, the odds of making the right choice based on this one-sided data set are much worse than, say flipping a coin—two out of three equally likely outcomes negate the risk factor possibility.

See also The Power of Positive Thinking

Beyond the risk analysis shortcoming, the data collected here is unlikely to be accurate in the case of almost all the factors. So even if the risk factor flaw were solvable (by comparing the 2 groups), the data being fed into it would be unreliable.

The most glaringly obvious example is the breed identification data. This is particularly worrisome since this is the factor most likely to result in unfair business practices that victimize consumers, by placing restrictions on insuring clients according to the breed labels affixed to their dogs.

There is a robust body of research showing that even people whose professions require expertise about dogs do very poorly when attempting to identify a dog’s breed by looking at the animal. Most of the labels here are attributed to visual identification. Moreover, most of the dogs in the sample are likely to be of mixed breed ancestry, as are more than half the general dog population in the US. The use of the term, “predominant” to describe the breed label seals the fate of this data, since it collects mixed breed dogs into the same basket as the pedigreed. This descriptor renders all the breed labels here unusable, with the exception of the ~25% who are presumably pedigreed dogs as their identification is attributed to “papers.” The “predominant” term presents at least 3 problems. First, and most simply, “predominant” implies that more than half (or at least more than any other breed) of the dog’s DNA can be traced to the named breed. This can only be established through DNA analysis (and sometimes not even then) or through direct knowledge of the precise, documented heritage of the parents. Without such documentation, the dog’s breed heritage is divided among 3 equally plausible possibilities: she is a pedigreed dog of a single breed for whom no pedigree was reported; she is a mix of an unknown number of breeds in unknown proportions of which the named breed is one; she is a mix of an unknown number of breeds in unknown proportions of which the named breed is not one. A dog’s physical traits actually provide no indication of the relative likelihood of these possibilities. And if all an individual’s forebears are not members of the closed gene pools we call breeds, the number of possible combinations of genetic traits becomes dizzying, far beyond the scale of predicting winning lottery numbers. Nothing whatever can be predicted about a dog based on a “predominant” breed, even if such can be established.

More information on the difficulties with this data can be found on the National Canine Research Council website. We must conclude that a fresh approach is needed. https://nationalcanineresearchcouncil.com/wp-content/uploads/2021/10/FAILURE-Insurance-Dog-Bite-Claims-Data.pdf