The UCSD psychologists found that when people pick a dog, "they seek one that, at some level, resembles them, and, when they get a purebred, they get what they want."

Forty-five dogs and their owners were separately photographed and judges were shown pictures of an owner, that owner's dog, and one other dog, with the task of picking out the true match. The proof of resemblance was that a majority of the purebred dogs and their owners could be identified by the 28 judges called upon to examine the photographs, with the results showing 16 matches out of the 25 purebreds. There was no evidence of resemblance between the mixed breed dogs and their owners.

"This is a project in which I've been interested for a long time because you hear a lot of casual talk about dog-owner resemblance and we wanted to see what could be learned by formal research," said Professor Christenfeld. Once the researchers were able to confirm, with randomized photo matching techniques, the high incidence of resemblance between owners and purebreds, and none for mixed-breeds, they went on to conclude that the similarity was due to owner selection at time of acquisition.

Resemblance only among the pure-breeds and their owners ruled out another possible explanation, that of convergence--the theory that similarity might grow with duration of ownership. Not only was there no correlation between how long a dog and owner had been together, as to similarity in appearance, but for convergence to be applicable the mixed-breeds would also have a resemblance to their owners. Since the similarity was only among pure-breeds, whose future appearance could be predicted, the study concluded that, in a majority of cases, owners select dogs who resemble themselves.

To examine whether people do look like their pets, and to explore the underlying mechanism, the researchers asked 28 judges (undergraduate college students) to examine photos of 45 dogs and their owners, taken at three dogs parks. Owners were approached at random and asked to help with a psychology experiment. The pictures were taken so that the background was different for dog and owners. Triads of pictures were constructed with one owner, that owner's dog and one other dog. A dog was regarded as resembling its owner if a majority of judges (more than 14) matched the pair.

The findings do not reveal at what level the resemblance between person and pet exists. It could be a similarity of physical attributes or of personality traits. The matches seem to be based less on specific obvious characteristics-the connections were not, for example, between hairy people and hairy dogs or big people with big dogs. The data does not reveal how judges were able match dogs to their owners, but the study concludes, "it does appear that people want a creature like themselves."