Herein I propose a moratorium on the use of the word “simple” immediately preceding the word “model” in any communication, written or verbal, between and among political scientists.

I am going to assume that we have all seen this particular locution enough that I do not have to trot out a string of examples to demonstrate that it is endemic. Simple models are everywhere. By account of political scientists who describe what they do as modeling— formal or statistical or otherwise— models that are not simple are hard to find. “Simple” has become an automatic modifier of “model” in a way much like how broadcast news reporters attached “Panamanian strongman” to Manuel Antonio Noriega in the early 1990s, or “Yugoslav rump republics” to Serbia and Montenegro later in that decade. The use of “simple” before “model” is, I suspect, practically unconscious by now among many political scientists.

Let us dispense with a couple of caveats at the outset. First, all models are simple, in the sense that they are stripped-down representations of a more complex world. Whatever parameters the modeler leaves in the model are those she or he holds to be the most essential for explaining the phenomenon at hand. Any undergraduate in an introductory course in any science (political or other) should recognize this. If, by putting the word “simple” before “model,” then, one is attempting to remind the audience that models are simplifications, the phrase is overly redundant (joke!), and should be avoided.

Second, some models are simpler than others. Some models that can be written down in mathematical notation can be translated to English in a straightforward and concise fashion. Other such models require substantially more words, and sometimes a good deal of hand-waving and hopping around, to describe without mathematical notation. That's OK, and the effort might well be worthwhile, but it is a sign the model is less (or more) than simple.

The thing is that use of the phrase “simple model” has been completely divorced from whether models are, in fact, simple by this yardstick. I was struck by this most recently when reviewing a paper that uses the phrase twice in the space of three sentences to describe a model that is actually a hierarchical extension of a Bayesian generalized linear model to estimate three separate parameters—the subtle differences among two of which, according to the authors, have eluded some of our best methodologists for a decade—describing the effects of electoral reforms on party competition in sub-national governments. Neither the baseline model nor the extension in question is anywhere near simple. They are good models of an important phenomenon and well worth modeling in my estimation, but not simple.

So, if “simple” is redundant as a generic qualifier of models, and is not, in practice, being used to distinguish models that can be translated in straightforward fashion into English from those that cannot, what is going on? Cutting straight to the chase, I suspect “simple model” often translates something like this:

If I am correct, then users of the phrase might be congratulated for communicating so much so economically, but a patronizing attitude—even communicated with some subtlety—is still obnoxious. So cut it out.