Tired of complaining about “electability”
So let’s get back to some politics, for a little while at least.
Lots of ink has been spilled about what’s going on with the Democrats
right now, as they focus on selecting a candidate to run against Bush
in the general election, and much of the discussion has centered on the
focus on electability and what the Democrats “should” be doing.
I’d like to talk about why the discussion of electability in the
Democratic candidate is reasonable and probably valuable to have.
In political science, we have a theory that applies very well to
two-party systems such as ours, called the median voter theorem.
(It can also apply to multi-party systems, with simple
modifications.) What it says, in brief, is that in a two-party
system, the “median voter” will win. Imagine, if you will, a
line, representing the ideological spectrum from left-to-right.
We can place every voter in our system along this spectrum. The
voter who falls in the middle of the spectrum (not ideologically, but
the one with equal numbers of voters on either side of him or her) is
the “median voter,” and in a two-party system, the party that gets this
voter will win an election.
(Note to my professional colleagues: This is not intended to be a
formal demonstration of the MVT — I could do that, but many of you
could do it better. I’m just trying to explain it stylistically
and show a general readership how it might apply to what we’re facing
in the elections right now.)
(Note to those interested: more formal definitions and discussion of this theorem can be found here, here, and here.)
Let’s consider a simple system of seven voters, as in this illustration.
There are seven voters here. The way this works is that voters
will pick the candidate closest to them on the preference
spectrum. Whoever captures four (or more) of the voters will win
the election issue here. So place two candidates anywhere you
like in the spectrum. One of them will be “closer” on the
spectrum to a majority of the voters, and that’s the candidate who will
win. We like to call this “capturing the median voter,” for the
candidate who captures the exact middle person — or the 50 percent
plus one voter, as you may have learned in a civics class — will
win. Another diagram will make this a bit clearer.
In this illustration, we’ve got a more complete system. Same diagram
as above, but now I have added candidates (1 and 2), a vertical line to
represent the exact middle of the preference spectrum (the bright green
line — you could consider it the dividing line between left and right,
or you can call it the “center”), and a thing called the tipping point
(which we will ignore for now).
Who will win the election? Remember, whichever candidate is
closer to the majority of voters, whoever can convince the median voter
to vote for him, he will be the winner. Once a candidate is close
enough to the median voter to convince this voter to select the
candidate, the candidate will also be even closer to a bunch of other
voters. Median voters are powerful, in a sense, because capturing
them produces an electoral majority. So if you’re a candidate,
you want to get the median voter to vote for you, because everyone else
to your side of the median voter will also vote for you.
In this election, candidate 2 wins. He is closer to the median
voter than candidate 1, so the median voter will choose to vote for 2
over 1, as 2’s preferences (which you might consider as “positions on
the issues”) are closer to the median voter’s than 1’s are.
Now take a look at this thing I have called the “tipping point.”
This
represents a point equally as distant from the median voter as
candidate 2 is. In other words, the distance from the median
voter to
the tipping point is the same as the distance from the median voter to
candidate 2. In order for candidate 1 to win, he will have to
come to the right of the tipping point, to get closer to the median
voter than candidate 1 is.
Can you guess how we might explain the tendency for both major
political parties in the US to look fairly similar, even on the
abbreviated ideological system that exists in this country? Yeah,
in pursuit of the median voter (who, in electoral opinion surveys of
the US, we have found is pretty much in the middle of the road
ideologically), the two parties have generally moved toward the middle
in pursuit of the MV.
Note some interesting features of this system: First, the
system
is somewhat middle of the road, as four of the seven voters could be
chracterized as preferring outcomes in the middle 50 percent of the
spectrum. Second, note that candidate 1 is closer to the (green)
center than candidate 2. Also note that the median voter is
pretty close to the middle of the system. Even so, the candidate
of the right will win, because his preferences are closer to those of
more of the people in the system.
O.K., so what does this mean for the current election? Here’s
what I think the election looks like to your average Democratic voter
right now.
Cand Right is George W. Bush. Candidates Left and Center-Left are
currently running in the Democratic primaries. If you are a
Democratic primary voter, and you think the election in November will
be close, you want to find the candidate who will come closer to the
median voter (whom I have located in the middle of the scale only for
convenience) than George W. Bush will. Candidate Left is just
about as far away from the MV as Candidate Right is. If your
estimation of the location of the median voter is off, you could lose
the election, and since we’re talking about a possibly close election,
you don’t want to be off. So you might choose to take someone who
doesn’t entirely match your own preferences (e.g., the cadidate is more
to the right than you are) in preference to having Candidate Right win.
So what the Democrats are doing right now, as they pursue the more
“electable” candidate, may be more “rational” than voting purely for
the guy whom they prefer to all others. And they realize that
this is not a one-shot game — it’s got two stages. Democrats who
are thinking of who’s more “electable” are perhaps remembering that
they have to win a presidency, not just a nomination. Democrats
pursuing an “electability strategy” are acting out a very complex
analytical situation in a very intuitive way. For just about any
Democratic primary voter, any Democratic candidate is preferable to
George W. Bush, and if the electorate is so tightly divided, it might
do to start with lots of people closer in their preferences to a
party’s candidate (as would be the case with Candidate
Center-Left). Otherwise, in the case when Candidate Left runs as
the nominee, Democrats have a much harder job of positioning to do to
get the votes necessary to obtain their majority.
So the move toward Kerry is not surprising (to me, at least). It may
represent very rational action by large numbers of Democratic primary
voters. But let’s see what happens in Tuesday’s caucuses and
elections….
Caveats: Yeah, this isn’t a perfect model of our system, as we have an
electoral college, possible shifts in preferences by candidates and
voters over the course of the run-up to the election, and so
forth. But it’s a stylized and simplified version of a complex
story that, I think, helps us to understand a bit more what’s going on
out there.
Also, this doesn’t represent my preferences. =-)
Ecto blogging software
3 February 2004 at 8:19 pm.
Nate, Nice synopsis, though I am curious if this concept assumes an equal distribution of ideology across the spectrum. That is, are there more right leaning or left leaning folks and does running to the middle actually mean running more to “right” or “left”?
Also, the electoral map calculator I mentioned can be found at the American Research Group web site. Good catch all site for primary polls, etc. According to my messing with the calculator, IIRC, if the Democrats take every state they took in 2000 and add AZ, the Dems win. Interesting scenarios abound.
3 February 2004 at 11:49 pm.
To what extent does the model incorporate voter preference for an incumbent? I get the sense that some voters may prefer the known rather than the unknown (and we’ll ignore Rumsfeld’s ruminations on these notions for the moment), so that the positioning of the voters to be garnered in the election is itself influenced by which one of those candidates is the incument. In other words, the positions of the candidates and the voters are not independent.
6 February 2004 at 9:02 pm.
Dan and Alex,
Good questions. Let me try to answer.
The “middle of the spectrum” does not have to necessarily be a politically centrist viewpoint. The middle is wherever that median voter is. So if the body politic tends to skew right or left, the middle will also skew right or left, respectively. The American voter, in social science public opinion and voting studies, has tended pretty much toward the middle of the political scale, so running to the middle (median voter) generally means running to the center.
The general model does not incorporate incumbent preference. The preferences are generally modelled as a set of choices about outcomes. Each point on the spectrum represents some set of preferred outcomes and benefits to the voter. Thus, the median voter might prefer B over A over C. But the far right voter might prefer C over B over A. And the leftist voter might prefer A over B over C. Each voter picks the point on the preference spectrum that matches his ov her own preferences. (The very generalized model I described is based upon a single-issue preference sprectrum. We could extend the spectrum to cover more than one issue [i.e., make it multidimensional], but the math gets more complex and the basic underlying story is roughly the same.
Incumbent preference would likely be modelled by a median voter shift toward the incumbent, whichever side that person might be on. In this case, it would skew the median voter further right than I placed the median voter, and so the Dems would probably want to move accordingly in some proportion. But you’re right that the relative incumbency and the voter preferences are probably not entirely independent.