Do Dogs Resemble Their Owners?

Abstract

We examined whether the frequent casual reports of people resembling their pets are accurate by having observers attempt to match dogs with their owners. We further explored whether any ability of observers to make such matches is due to people selecting dogs who resemble them, in which case the resemblance should be greater for predictable purebreds than for nonpurebreds, or is due to convergence, in which case the resemblance should grow with duration of ownership. Forty-five dogs and their owners were photographed separately, and judges were shown one owner, that owner's dog, and one other dog, with the task of picking out the true match. The results were consistent with a selection account: Observers were able to match only purebred dogs with their owners, and there was no relation between the ability to pair a person with his or her pet and the time they had cohabited. The ability to match people and pets did not seem to rely on any simple trait matching (e.g., size or hairiness). The results suggest that when people pick a pet, they seek one that, at some level, resembles them, and when they get a purebred, they get what they want.

Full text

Research Report
Do Dogs Resemble Their
Owners?
Michael M. Roy and Nicholas J.S. Christenfeld
University of California, San Diego
ABSTRACT—We examined whether the frequent casual reports of
people resembling their pets are accurate by having observers
attempt to match dogs with their owners. We further explored
whether any ability of observers to make such matches is due to
people selecting dogs who resemble them, in which case the re-
semblance should be greater for predictable purebreds than for
nonpurebreds, or is due to convergence, in which case the re-
semblance should grow with duration of ownership. Forty-five
dogs and their owners were photographed separately, and
judges were shown one owner, that owner’s dog, and one other
dog, with the task of picking out the true match. The results
were consistent with a selection account: Observers were able to
match only purebred dogs with their owners, and there was no
relation between the ability to pair a person with his or her pet
and the time they had cohabited. The ability to match people
and pets did not seem to rely on any simple trait matching (e.g.,
size or hairiness). The results suggest that when people pick a
pet, they seek one that, at some level, resembles them, and when
they get a purebred, they get what they want.
It has been asserted, by children’s-book illustrators, at dog shows, and
by strangers passing on the street, that people often bear a striking
resemblance to their pets. We examined whether such observations
are empirically warranted by testing whether people can match dogs
with their owners.
The fact that people remark on the resemblance between people
and their dogs is not in itself evidence that such a similarity generally
exists, as people may attend to the few cases in which there is some
overlap in appearance. People may also take advantage of the fact that
they know which people and pets are paired, given that they are often
connected by a leash, and may then search, post hoc, for features that
match.
There are two basic mechanisms that could create an actual sim-
ilarity between people and their dogs. People might select dogs
who are like them, or the features of people and their dogs might
converge over time. There is some evidence that the facial appear-
ance of married couples converges over time (Zajonc, Adelmann,
Murphy, & Niedenthal, 1987); judges have rated couples physically
more similar after 25 years of marriage than as newlyweds. If a sim-
ilar process were to occur with people and their pets, then one might
find that the ability to match them grows with the length of owner-
ship. In contrast, if owners resemble their dogs because of selection
rather than convergence, then there should be no correlation between
the degree of resemblance and the time spent together. However,
there ought to be, in this case, a greater similarity between people
and their purebred dogs than between people and their non-
purebreds. A nonpurebred puppy’s final appearance is un-
predictable, and so the resemblance between owner and dog should
be confined to the much more predictable purebreds. It is also likely
that the acquisition of a purebred dog involves more foresight and
deliberation than the adoption of a nonpurebred, and so should better
reflect the owner’s desires.
There is some evidence that people’s taste in animals is asso-
ciated with their own appearance: Coren (1999) found that women
with long hair gave higher ratings of attractiveness, friendliness,
loyalty, and intelligence to dogs with floppy, rather than pricked,
ears. There is also evidence that people have at least rudimentary
stereotypes of the type of pet that a person might own (Budge,
Spicer, St. George, & Jones, 1997). In pairing pictures of humans
and a variety of animals, judges were more likely to match cats and
small dogs with women and large dogs with males, though the re-
searchers did not examine whether this stereotype reflects actual
ownership patterns.
To examine whether people do look like their pets, and to explore
the mechanism underlying any such resemblance, we asked judges to
match photographs of owners with their dogs, both purebreds and
nonpurebreds.
METHOD
Pictures of 45 dogs and, separately, their owners were taken at three
dog parks. Fifteen such pairs were photographed in each location.
Owners were approached at random and asked if they would be willing
to help us with a psychology experiment examining the relation be-
tween owners and their dogs. The pictures were taken so that the
background was different for dog and owner. This ensured that raters
would not be able to match dog and owner by simply comparing the
backgrounds in the photographs. Owners were photographed from the
waist up, facing forward, wearing whatever clothes they had chosen for
Address correspondence to Nicholas J.S. Christenfeld, Department
of Psychology, University of California, San Diego, 9500 Gilman Dr.,
La Jolla, CA 92093-0109; e-mail: nchristenfeld@ucsd.edu.
PSYCHOLOGICAL SCIENCE
Volume 15—Number 5 361Copyright r2004 American Psychological Society

going to the park and whatever facial expression they chose for the
picture. Dogs were photographed facing forward, with the whole dog
visible; they made whatever facial expression they chose, exhibiting
rather more lolling tongues than the owners.
Owners (24 women and 21 men, with a mean age of 36 years) were
asked to indicate the breed of their dog and how long they had owned
the dog. Because of an error in procedure, data on length of ownership
were available from only two of the parks (n530). Overall, there were
25 purebred dogs (representing 15 different breeds) and 20 non-
purebreds (mean age 53.7 years, range 54 months to 12.5 years).
Owners had been with their dogs for an average of 2.9 years, with 76%
of the owners acquiring their dog in its 1st year. There was no sig-
nificant difference between purebred and nonpurebred owners in
length of ownership or age of the dogs at adoption.
Each set of 15 pictures was viewed by 28 naive undergraduate
judges who were participating for course credit. We constructed
triads of pictures, each consisting of one owner, that owner’s dog, and
one other dog photographed at the same park. Each judge was shown
the 15 owners from one dog park, one at a time, and instructed to
identify which of the two possible dogs belonged to each person. In a
Latin square design, each of the 14 incorrect dogs served as a foil for
each dog, with the order of presentation randomized. Thus, within
each set of 15 pairs, each owner-and-dog pair was presented with
every other dog photographed at that park. A dog was regarded as
resembling its owner if a majority of judges (i.e., more than 14)
matched the pair.
RESULTS
There was no evidence of any resemblance between nonpurebreds
and their owners; of the 20 such dogs, there were 7 matches, 4 ties,
and 9 misses, w
2
(2, N520) 50.64, n.s., r-equivalent 5.14 (Ro-
senthal & Rubin, in press). However, purebreds could be matched
with their owners; of the 25 purebreds, there were 16 matches, 0
ties, and 9 misses, w
2
(2, N525) 56.75, p<.05, r-equivalent5.37.
The difference between the matchability of nonpurebreds and
purebreds was significant, w
2
(2, N545) 57.03, p<.05, f5.40
(Cohen, 1969).
The results were consistent with the notion that the ability to match
is due to selection rather than convergence, as there was no correla-
tion between how long a dog and owner had been together and the
number of judges who correctly picked that dog and owner as be-
longing together (r5.12, n.s., over the 30 pairs for whom this in-
formation was available).
Further analyses explored the possible origin of the resemblance
between purebred dogs and their owners. Three undergraduates
naive to the purpose of the experiment rated several features of the
owners and purebred dogs independently, and we then correlated the
ratings of dogs and their owners on these dimensions. If the judges
hadmatchedpairsonthebasisofthese characteristics, then we
would find significant correlations. The three raters judged hairiness,
size, sharpness of features, attractiveness, perceived friendliness,
and perceived energy level using 6-point Likert-type scales (run-
ning, e.g., from 1, not hairy,to6,very hairy). The reliability of the
ratings on each dimension was examined factor analytically, allowing
us to determine how each individual’s ratings loaded on the latent
concept (Schuster & Smith, 2002). All dimensions displayed ade-
quate reliability for both human and canine photographs: Using
principal components extraction with varimax rotation, we obtained
only one factor for each attribute; eigenvalues were all greater than
1.7, and the factors accounted for between 58 and 88% of the
variability.
For none of the rated characteristics, however, was the correlation
between person and pet significant, although there was a suggestion of
a trend for people and their purebred pets to be rated similarly on
friendliness (r5.31, p5.13). To examine whether any rated similarity
would emerge when all the variables were considered simultaneously,
we performed exploratory canonical correlation analyses. No sig-
nificant relation between person and pet was found for any combi-
nation of the attributes, approximate Pillai’s F(64, 128)50.81, n.s. We
also examined whether, as suggested by the findings of Budge et al.
(1997), the judges’ accuracy with purebreds could be based on dif-
ferences between the pets of men and women. Dogs with male and
female owners did not differ on any of the rated characteristics, all
ts(23) <1.1, n.s. Finally, in light of the findings of Coren (1999), we
examined if the hairstyle of owners (long or short) was related to the
ear style (lop or pricked) of their dogs. There was no association for
female owners (f5.03, n.s.) or female and male owners combined
(f5.19, n.s.).
DISCUSSION
The results suggest that when people pick a pet, they seek one that, at
some level, resembles them, and when they get a purebred, they get
what they want. The resemblance between dogs and their owners was
confined to purebreds, and there was no correlation between length of
ownership and similarity. Both findings are consistent with a selection,
but not a convergence, account of the phenomenon.
The results do not reveal at what level the resemblance between
person and pet exists. It could be at the level of physical attributes,
with owners of wolfhounds being tall and elegant, or at a stylistic level,
with retriever owners being warm and friendly. There are personality
differences between owners of various sorts of pets (Podberscek &
Gosling, 2000), and our judges could have used what they knew about
temperament differences between breeds to help them make matches.
It does not, however, appear that they used any single obvious char-
acteristic to make the matches, because we found no significant cor-
relations between dogs and owners on the six dimensions we
examined. That is, the judges did not make correct matches by simply
matching hairy people with hairy dogs, or big people with big dogs.
There was some suggestion that people and pets were similar in ap-
parent friendliness, but the effect was of modest size, and not statis-
tically significant. It may be that the judges used some other more
subtle trait, or based their judgments on a more configural analysis of
the animals. We also cannot know from these data if people can tell
whether a particular person is an owner of a dog, as opposed to, say, a
weasel. However, it does appear that, as in the case of selecting a
spouse (Berscheid & Reis, 1998), people want a creature like them-
selves.
REFERENCES
Berscheid, E., & Reis, H.T. (1998). Attraction and close relationships. In D.T.
Gilbert, S.T. Fiske, & G. Lindzey (Eds.), The handbook of social psy-
chology (4th ed., Vol. 2, pp. 193–281). New York: McGraw-Hill.
362 Volume 15—Number 5
Dogs Resemble Owners

Budge, R.C., Spicer, J., St. George, R., & Jones, B.R. (1997). Compatibility
stereotypes of people and pets: A photograph matching study. Anthrozoo
¨s,
10, 37–46.
Cohen, J. (1969). Statistical power analysis for the behavioral sciences. New
York: McGraw-Hill.
Coren, S. (1999). Do people look like their dogs? Anthrozoo
¨s,12, 111–114.
Podberscek, A.L., & Gosling, S.D. (2000). Personality research on pets and
their owners: Conceptual issues and review. In A.L. Podberscek, E.S.
Paul, & J.A. Serpell (Eds.), Companion animals and us (pp. 143–167).
Cambridge, England: Cambridge University Press.
Rosenthal, R., & Rubin, D.B. (in press). r-equivalent: A simple effect size
indicator. Psychological Methods.
Schuster, C., & Smith, D.A. (2002). Indexing systematic rater agreement with a
latent-class model. Psychological Methods,7, 384–395.
Zajonc, R.B., Adelmann, P.K., Murphy, S.T., & Niedenthal, P.M. (1987).
Convergence in the physical appearance of spouses. Motivation and
Emotion,11, 335–346.
(RECEIVED 4/3/03; REVISION ACCEPTED 6/4/03)
Volume 15—Number 5 363
Michael M. Roy and Nicholas J.S. Christenfeld