Description

This article is from the rec.pets.dogs: Getting A Dog FAQ, posted to rec.pets.dogs newsgroup. Maintained by Cindy Tittle Moore with numerous contributions by others.

003. What Kind of Dog Should I Get: Purebred or mixed-breed dogs

If you are interested in a purebred dog, you should pick up a book on dog breeds (most libraries will have a good selection) and do some research, with the above questions in mind. There are some breed-specific FAQ's available. Finally, you should SERIOUSLY consider attending a dog show where not only can you potentially contact breeders, but you can see ADULT specimens of the breed you are considering. It's very important to remember that cute little puppies remain cute little puppies only for a matter of weeks. There is a long period of ungainly and rebellious adolescence finally followed by mellow adulthood.

If the dog's breed is not important to you, you should still consider the above list when choosing the dog. You do face a few more unknowns since a mixed-breed puppy (e.g., a "mutt") may or may not clearly exhibit what its adult characteristics will be.

Many people have strong feelings about purebred dogs, especially the characteristics of the breed. Other people feel that the "stereotypes" are overrated. Jon Pastor made some nice comments about the usefulness and caveats of typical breed behaviors:

Are behaviors commonly ascribed to specific breeds based in fact or are they just stereotypes?

They are really a bit of both: they are informal statistical descriptions (i.e., stereotypes), and to the extent that they reflect reality they're also facts. "Stereotypes" -- or, more simply, "types" -- can be, but are not necessarily, evil: it depends on how you use them.

Typical means "characteristic of the type," and is a statistical abstraction; it does not have any normative implications -- i.e., there is no claim that all (or even most) examples of the type in question have the characteristics that are stated to be typical. One of the ways in which people make sense of the world is by comparing entities they encounter with the types they've stored in their memories in order to identify them; it's a remarkably effective way of compiling knowledge about an infinitely complex environment so that it can be accessed quickly enough to (in the extreme) save one's life.

Thus "typical" is a largely ad-hoc, somewhat personal label, until it is agreed-upon by some number of people who share the same notion of what common characteristics identify the "typical" object of a particular kind. If we could eliminate the biases that have been identified in such behavior (e.g., if the last 20 dogs you've seen have been Borzois, you will most likely over-estimate the true number of Borzois in the dog population), we would find that "typical" approximated some statistical tendency in the population we're addressing, typically the mean (average) or mode (most common).

If you pick some characteristic and look in a particular population to see how many individuals have different levels of that characteristic, you'll find that when you graph the results they look like this (more or less):

no.
|                             |
|                             *
|                         *   |   *
|                       *     |     *
|                      *      |      *
|                     *       |       *
|                    *        |        *
|                  *          |          *
|               *             |             *
|          *                  |                  *
| *                           |                          *
--+---------------------------+------------------------------------------------
score

There will be some value that occurs most frequently (the mode); in the case of a perfectly symmetric curve like the one above, this value will also be the average (mean). Symmetric curves like this occur surprisingly frequently, so I'll continue to use it as an example.

For example, let's say that you want to plot the aggressiveness of various breeds. First, you have to come up with a way of ranking dogs on aggressiveness [an exercise left to the reader ;-)], and then for each breed you score a large number of dogs on aggressiveness and plot the results:

   
no. with
score                         |
|                             *                   |
|                         *   |   *               o
|                       *     |     *         o   |    o
|                      *      |      *      o     |     o
|                     *       |       *    o      |      o
|                    *        |        *  o       |       o
|                  *          |         o*        |         o
|               *             |     o       *     |             o
|        *                    |o                 *|                   o
| *                o          |                   |      *                    o
--+---------------------------+-------------------+----------------------------
"aggressiveness" score

Here, breed 1 is represented by '*' and breed 2 is represented by 'o'. Notice a couple of things:

1. the centers of the two curves are clearly separated, from which you'd conclude that the breeds differ to some degree in aggressiveness

2. there is some overlap, so that the most aggressive breed 1 dogs are substantially more agressive than the average breed 2 dog, and the least aggressive breed 2 dogs are substantially less aggressive than the average breed 1 dog.

The significance you attribute to the results depends on the shape and position of the curves, but in most cases there will be substantial variation within groups and at least some overlap between groups.

Now, by doing this in N dimensions you can play the same game on as many characteristics as you wish, and make statistically meaningful statements about tendencies of one particular breed or typical differences between breeds.

By doing so, you are *NOT* saying that

1. all dogs of a particular breed have all -- or, in fact, *any* --of the "typical" levels of each characteristic

2. there is necessarily any real dog that has all of the "typical" levels of each characteristic

3. it is impossible for a dog of breed 2 to have some -- or, in fact, *all* -- of the typical characteristics of breed 1

This is not a True/False situation, it's an infinitely-graded situation. If you get a dog of that particular breed, the modal (typical) value is simply the one you'd be most *likely* to get.

A big caveat: breed traits are not computed scientifically, and are thus not quite subject to the laws of Statistics. However, they do reflect the cumulative wisdom of hundreds (thousands?) of years of human observation and active breeding of dogs.

The bottom line is that if you get an Newfoundland, it is highly likely that it will be a good lifesaving dog; it is possible, although less likely, that it will be a *great* lifesaving dog; and it is also possible, although also less likely, that it will show no aptitude for lifesaving. Similar statements hold for "typical" traits of sight hounds, Rotts, Poodles, GSDs, Goldens, Irish Setters, and any other breed you can think of.

If you use this "stereotype" information to inform your choice of a dog, and make some effort to determine how "typical" a given dog is likely to be of its breed (by looking at parents and siblings, by observing the dog, by asking the owner, etc., etc.), it's innocuous and can be quite useful. If you use it blindly to make blanket judgements of breeds, use of stereotypes can be foolish. In the extreme, if you don't understand the meaning of the characteristics, or have mis-identified or mis-measured them, use of stereotypes can be positively evil, such as when "all Pitbulls" are identified as dangerous and banned.

The only conclusion that this discussion licenses with respect to the purebred-vs.-mixed-breed question is that prediction is easier with a purebred because the number of purebreds is (relatively) small and (relatively) fixed, while the number of possible mixes is essentially infinite; as a result, there has been more observation of individual "pure" breeds, and there is consequently more data to support generalizations about breed characteristics. This is not, by any means, to say that purebreds are necessarily better or worse; they're just more predictable.

So if you want a dog with a particular set of characteristics, you will be more likely to get such a dog if you find a breed that typically has those characteristics and choose a dog of that breed *intelligently* than if you choose a dog of mixed breeds (unless, of course, you're talking about an older dog whose behavioral characteristics are already obvious and therefore observable). This is a statement about probability, not about quality, and anyone who attempts to apply an absolute value-scheme to it is making unwarranted and unjustifiable extrapolations.

Statistics is a powerful weapon. As with any other such weapon, use it ignorantly or indiscriminately at your peril...

 

Continue to:

• prev: 002. What Kind of Dog Should I Get: Factors to consider
• Index
• next: 004. What Kind of Dog Should I Get: Books

Share and Enjoy

Tags

dog, owner, getting