The utility of model dogs for assessing conspecific aggression in fighting dogs

Abstract

This paper describes a study to assess the utility of a life-sized plush model dog as a screening tool for conspecific aggression in a population of dogs bred for use in dogfighting. Because the dogs in this sample experienced similar selective pressures and life histories, we hypothesized conspecific aggression, when present, would be a stable trait readily elicited by the model. Thus, we predicted that aggressive dogs would behave consistently toward both the model and live conspecifics. We tested our prediction with a sample of 292 animals (165 females and 127 males) confiscated during a large dogfighting raid. We found evidence of substantive predictive value with 75% of the dogs exhibiting aggression to the model also exhibited aggression to the same-sex stimulus dog. Similarly, sensitivity was high, with 81% of same-sex-aggressive test dogs detected by the model, indicating the model can discriminate between dogs with and without conspecific aggression. Finally, we found evidence of concurrent validity – with dogs exhibiting similar levels of aggression in all scenarios but with a significant sex*scenario interaction (p = 0.03) driven by higher level of aggression in the same-sex stimulus dog condition for male subjects. Based on our findings, we recommend using a model dog first when evaluating fight-bred dogs. If no physical aggression is exhibited to the model, or the behavior is ambiguous, the dog should be tested with a same-sex stimulus dog. Opposite-sex stimulus dogs alone are insufficient to screen for conspecific aggression because aggression is reduced in this test scenario for male subjects. Screening for conspecific aggression with model dogs does not completely eliminate the need for live conspecifics, but it substantially reduces the risk to stimulus dogs and evaluators alike by screening out the majority of extremely conspecific-aggressive animals. Using this procedure, a small number of dogs will be mistakenly identified as aggressive, but few truly aggressive dogs will be missed. This is the first in-shelter study on model validity to use a large, homogenous sample of dogs with known life histories and to report the predictive value of models to screen for conspecific aggression in fight-bred dogs. Comparisons of a canine model’s validity in other populations of dogs known to exhibit conspecific aggression are needed to determine if fight-bred populations are qualitatively different from other domestic dogs in their response to a model conspecific.