New estimating method reduces “wobble” factors & confirms accuracy of the old method
How many animals are killed or badly injured by dogs each year?
How many of those animals are killed or badly injured by pit bulls?
Quick & easy
If you just want quick-and-easy projections, four years’ worth of data, 2013 through 2016, indicates that on average:
- About 31,500 dogs per year kill or severely injure other animals, including about 27,500 pit bulls (87%).
- About 10,000 dogs per year are killed by other dogs, 85% of them killed by pit bulls.
- About 11,500 dogs per year are disabled or disfigured by other dogs, 10,500 of them disabled or disfigured by pit bulls (92%).
- About 2,500 cats per year are killed by dogs, about 2,000 of them (80%) killed by pit bulls.
- About 7,500 hoofed animals, poultry, and wildlife per year are killed by dogs, about 6,400 of them (85%) by pit bulls.
The 2016 numbers appear to have been significantly higher.
(See table below.)
If you want to know where the numbers come from, why we just spent four days doing all the math necessary to derive them and prove them through cross-checks, and why we ended up changing the estimation method we used in 2013-2015, please read on below the table.
Why do the math?
As in determining the numbers of humans who are killed or disfigured by dogs each year, and the percentage of victims killed or disfigured by pit bulls, attempts to find the answers run afoul of the reality that information about incidents is not routinely reported from local first responders to any law enforcement or public health agency at the state or national level, nor to any state or national medical, veterinary or insurance association.
No one has ever tracked human dog attack data year in and year out in a consistent manner, except ANIMALS 24-7, collating the information since 1982 from local media reports, the only consistently accessible source of input data.
Because human deaths almost always occasion at least some local media notice, we can be reasonably sure that we collect the particulars of almost all of them.
Because human injuries of a permanently disabling or disfiguring degree of severity are also often reported, we can be confident that our data on the most severe nonfatal dog attacks on humans provides at least a very strongly representative indication of the trends.
But dog attacks on other animals, we have learned over the years from tabulating thousands of accounts, rarely receive media notice, no matter how many other animals die. Indeed, dog attacks on other animals are almost never reported by mass media, even those serving the smallest communities, unless any (or all) of three further circumstances occur as result of the same incident:
- A human is killed or disfigured;
- Law enforcement or another intervenor kills the attacking dog or dogs;
- or livestock are killed of market value exceeding $1,000.
Estimating the under-reporting
ANIMALS 24-7 has since 2013 sought to estimate the total numbers of dog attacks on other animals, and the contribution of pit bulls to the total, by estimating the under-reporting factors and then projecting the available data by whatever those factors are.
Based on the thousands of dog attack accounts we have logged and evaluated, we believe that at minimum the number of reported attacks is only a third of those that are reported, even in the smallest towns, where any event out of the unusual tends to make news.
Least likely to be reported are dog attacks on other animals in which the victim animal or animals belongs to the same household as the attacking dog.
Yet these are the households, the data involving human deaths and injuries indicates, where dogs are most likely to kill and disfigure other animals, simply as a matter of proximity and opportunity.
Also tending to go unreported are dog attacks on other animals in which no human is injured, no one kills the attacking dog, and the economic value of the victim animals is not high enough to encourage the animals’ owners to pursue litigation.
Accordingly, we believe that at minimum the numbers of reported dog attack victim animals must be multiplied by three to get a representative total.
Beyond these considerations, there is the consideration that any local event tends to be of diminishing newsworthiness as the audience expands. From 2013 through 2015, ANIMALS 24-7 tried to calculate the newsworthiness of dog attacks on other animals by comparing the available reportage to the reportage of police discharging firearms in animal-related cases.
Taking all accessible data into account, ANIMALS 24-7 multiplied reported dog attacks on other animals in communities of under 10,000 people by a compensation factor of three; multiplied reported dog attacks on other animals in communities of 10,000 to 100,000 people by a compensation factor of 10; and multiplied reported dog attacks on other animals in communities of more than 100,000 people by a compensation factor of 147.
This produced numbers that looked reasonable, but also required a tremendous amount of labor to compute. Inputting the information for each reported dog attack on other animals required looking up the size of the community, inputting the population data for the community, and then, at the end of each year, spending three days or more doing calculator work to ensure proportional representation.
Even at that, the data wobbled. For example, even within samplings of hundreds of incidents, the accuracy of our projections could be thrown off by as little as a single incident in which one dog killed an unusually large number of poultry or hooved animals.
More problematic than that, representative data on police discharging firearms in animal-related cases is not always accessible or up-to-date, while media perceptions of newsworthiness tends to vary, not only relative to the size of any given community, but also relative to whatever else is of public concern.
Most obviously, dog attacks in an election year may be much less newsworthy at even the local level than in a non-election year.
Rural vs. urban
The reality we ended up having to work from, in trying to project the probable 2016 totals of dog attacks on other animals, is that data from communities of under 1,000 people showed one fatal or disfiguring dog attack on other animals per 294 residents, while data from communities of over one million people showed only one fatal or disfiguring dog attack on other animals per 844,774 residents.
A case can be made that the data from the smallest and most rural communities is likely to be the most complete and accurate. On the other hand, people in the smallest and most rural communities are also likely to keep the most animals, while people in urban metropolises keep the least––and relatively few Americans live in small communities, while more than 75% live in cities.
Such wide discrepancies can be resolved by looking carefully at the numbers from communities of in-between size, and we did spend two days doing it.
Projecting the 2016 numbers as we had in 2013, 2014, and 2015 produced eventual estimates of 34,419 dogs, 26,905 of them pit bulls (78%), killing 17,728 other dogs in 2016, including 11,018 other dogs killed by pit bulls (62%), who make up only about 5% of the total U.S. dog population.
In addition, according to this projection method, dogs severely injured 12,982 other dogs; 11,905 of those dogs (92%) were injured by pit bulls.
The most consistent number, projecting across the range of community size, was that 80% of the attacking dogs in towns of under 1,000 people were pit bulls; 83% in towns of 1,000 to 9,999 people; 90% in towns of 10,000 to 99,999 people; 83% in cities of 100,000 to 999,999 people; and 94% in cities of more than a million people.
Wobbles of concern
Though these numbers are plausible and defensible, we were and are troubled by the “wobble” factors that resulted from projecting large numbers from small amounts of input data. While this can be done accurately, albeit not easily, some of the “wobble” factors significantly interfered with projecting the numbers for dog attacks on cats and hoofed animals.
Also of concern was the variation we found in the projected totals for each year, which wobbled opposite to the numbers of dog attacks on other animals for which we had collected information, and of greater concern, opposite to the directions indicated by the insurance claim data reported each year since 2003 by the Insurance Information Institute of America.
Logically, even though relatively few dog attacks on other animals result in insurance claims, the numbers that do should roughly parallel the overall trends.
Accordingly, to project what ANIMALS 24-7 believes to be our most complete, accurate, and representative estimate of the numbers of dog attacks in 2016, among six estimates for which we did all the numbers, we threw out all of our previous estimates and started over by using a new method of calculating the under-reporting factors.
We kept our assessment that the reported numbers from documented attacks should be multiplied by at least three, to compensate for non-reporting of attacks occurring within the dogs’ own households and attacks in which no one is seriously injured, the attacking dog or dogs are not killed, and there is no loss of animals with market value of more than $1,000.
Insurance data more consistent
Instead of using a method of estimating the remaining non-reporting factors based on the population size of communities in which dog attacks on other animals occurred, we switched to comparing the Insurance Information Institute of America claims data to the numbers of human deaths and disfiguring injuries we have tabulated each year from media accounts.
This produced a satisfying consistent baseline for projection: from 2013 through 2015, and estimating that the ratio of claims to reported human deaths and disfiguring injuries will remain the same in 2016 as the 2013-2015 average, about 4% of claims have resulted from cases covered in media accounts, with relatively little wobble up or down.
If 96% of the dog attack injuries resulting in human injury insurance claims go unreported by media, and if those attacks are of approximately the same average level of severity as would kill or disfigure another animal, we can estimate the total number of dog attacks each year by presuming that the media reports we have collected are equal to 4%, multiplied by three.
Methods confirm each other
Based on that, ANIMALS 24-7 has assembled new projections of the numbers of dog attacks on other animals for each year 2013 through 2016, cited above and illustrated in the accompanying tables. The input data is in the table below.
Overall, our new projections of the numbers of animals killed and injured by dogs are lower than in our previous estimates, but the projections tend to be more consistent with what both the volume of input data we have collected and what the Insurance Information Institute of America claims data would indicate, and more consistent from year to year.
More significantly, our “legacy” projection for 2016, using our former method of estimating, is so close to the estimate using the new method that each tends to confirm the accuracy of the other.
ANIMALS 24-7 is continuing to log the data, and of course to continue refine and improve our projection methods, to provide readers with the most complete and accurate information possible.
Please help us continue speaking truth to power: