In a late‑night kitchen experiment, owner Adam Wespiser pitted his 83‑lb Greyhound, Bebop, against five treats using the Bradley‑Terry model. Daily pairings let the dog choose between two snacks, generating data for a statistical ranking.

The model assigns each treat a strength score, turning win–loss counts into probabilities. After 30 trials, Pur Luv Chicken emerged as the leader, beating the favored Greenies and Pork Chomps in most pairings.

A bootstrap analysis reinforced the result: Pur Luv Chicken topped 63 % of resampled models, while the next best, MON2SUN, reached only 33 %. A right‑side bias in the kitchen setup noted, but did not alter overall ranking.

The study shows how simple pairwise comparisons can quantify preferences, a method already used in Elo chess ratings and AI model contests. For dog owners, the takeaway is clear: the dehydrated chicken treat wins the statistical vote.