On Appealing to the “Median Voter”
Because I think too much about this stuff.

Ok, so it occurs to me (and I know everyone has been wondering about this!) why I find almost all popular press usage of the term “median voter” to grate on my nerves (this includes outlets like FiveThirtyEight, which should be more precise than they are). I think the main problem is that pundits have a habit of talking about appealing to the median national voter, but there are, alas, no elections in the United States wherein competition would be for the median American.
As a general matter, I think that the problem is that the binary nature of our party system coupled with the single-seat, winner-take-all nature of our politics makes everyone think in 1s and 0s and also assumes that undecided or independent must mean “in the moderate middle.” That is, being near the median doesn’t make you undecided, and being undecided doesn’t necessarily mean you are near the median voter’s position. You can be a radical left independent or a far-right undecided.
For those unfamiliar with the concept, I wrote about it here (On the Median Voter) back in November.
First, let’s remember that the word “median” means the middle of a distribution in the sense that half of the distribution is on the left-hand side and half is on the right-hand side. Note the simple set of examples below (source). If what is being measured is ideological policy preferences (or simply partisan distribution of a district) the policy preferences of the “median voter” would be different in all three districts.

The simple point I am trying to make is that being the median voter does not mean you would adhere to a specific set of ideological positions. It means that your positions are relative to the distribution of voter preferences in the district (or country) in question.
So, let me underscore two thoughts.
First, the ideological position of the “median voter” depends on the distribution being discussed.
The median voter in TX04, which voted 74.9% for Trump in 2020, is ideologically quite different than the median voter in NY07, which voted 81.8% for Biden in 2020.
In other words, “median voter” does not mean “moderate” save in the context of the given distribution.
Second, an undecided voter is not necessarily proximate to the “median voter.”
There is an accepted wisdom that undecided voters are moderate voters near to the median position of the country (or district). But there is no reason why an undecided voter might not be a far-left or far-right voter (or have heterodox views). Indeed, it should be pretty clear that a decent slice of undecided voters really may not have a clear ideological position. If they were more ideologically coherent, they probably wouldn’t be undecided. This isn’t to say that some undecided aren’t high-information moderates. Some are. But if you listen to most discourse on this topic, there is an assumption that most (if not all) undecideds fall into that category.
Along these lines I have written about this before:
- In 2019 I wrote On “Independent” Voters, which looked at who self-identified “independents” were and found that most of them are partisan leaners and only 9% of the overall electorate, according to Pew Research, and that group is less inclined to vote than other groups identified.
- In 2022 I noted a piece by Lee Drutman that discussed moderate/independent voters, Speaking of Moderates and Independents…, which underscored, again, that the notion that those folks are just waiting in “the middle” to be plucked up by the side with the “correct” strategy is a myth.
- Also in 2022, I noted the disjuncture between stated partisan preferences and actual voting: Partisan ID in the United States.
- I discussed some of the other problems with the median voter theorem here in April of this year: Nope Labels.
dl;dr: Independents don’t necessarily mean Moderates nor Centrists and may or may not be close to the Median Voter. There may well be overlap, but each one of those things is not a synonym for the other.
A last thought: it is far easier to track and understand how a voter fits in terms of closeness to the median on a given issue such as abortion (as I noted in the post On the Median Voter). We have seen in single-issue elections that the median voter is likely to support abortion rights. But many of those voters who vote to support abortion rights in a given state also vote Republican for office, even though the Democratic candidate is closer to them (and the median voter) on the topic of abortion.
As such, the problem is that voters contain multitudes when it comes to policy and political preference, so admonitions by pundits that politicians should “appeal to the median voter” are just wildly simplistic. Indeed, this problem accrues to the notion of “appeal to the moderates” as well.
(And apologies, as I expect people would have preferred something about the DNC, but kudos for getting to this apology).
Great point, needs to be repeated often. One practical consequence is to show how misguided it is to concoct some national messaging that’s supposed to work with all undecided voters in all states.
Maybe if Plato and Socrates had been less disdainful of democracy, they’d have found the Form of the median voter 🙂
Harris campaign seems more oriented to GOTV than median voter theory.
Back when she was a California senator, she had positions (e.g., fracking) that would appeal to the median CA Democratic voter. Now, like Biden, she is more aligned to the center of the Democratic Party and its Democratic voters. (So, e.g., backing off on the anti-fracking thingy).
The big change from earlier Democratic campaigns is de-emphasizing communicating through the MSM and reaching voters instead with newer methods – Tic-tok, influencers, etc.
I started off today’s forum with a link to a TNR piece on this, and yesterday with this link:
https://the.ink/p/new-democratic-party-brat-pack
Another difficulty is that people tend to see median/moderate positions as representing ideal/best/morally superior/ winning ones.
Good observations. Thanks! And no, I wouldn’t have wanted to read a post about the DNC. Not even if it had been about the median/moderate reaction to it. 😉
@charontwo:
I just noticed the link to Semafor in the Tabs post, there is extensive fisking of the Semafor piece over at BJ.
Side note: I am listening to the 538 reaction to Day 2 of the DNC and Galen Druke keeps talking about “the median voter.”
I’ve long noted that in much polling the “independents” generally poll pretty close to the midpoint between Rs and Ds. This tells me they’re pretty evenly split between D leaners and R leaners. It seems to me lot of people like to feel they’re smart for being above the fray, not taken in by any party.
This post is more about “independents” than the “median voter”. The basic problem is that there are multiple issues and therefore multiple dimensions and therefore no one median. As you note, a lot of Republican voters support abortion rights. IIRC a majority of NRA members support expanded background checks. If you play to the median on one issue, you may antagonize the median on another.
I forget where, and it’s past my expertise, but I’ve read that in politics not only can a party not define a set of positions that appeal to a hypothetical median voter, it can’t, as in it’s proven mathematically impossible, to define a Pareto Optimum set of positions across a set of voters.
In the search for a “median” position, pundits of course revert to the Pundit fallacy.
@gVOR10:
What is the Pundit fallacy?
I Googled and this is what was returned: “Yglesias coined the term “pundit’s fallacy” to denote “the belief that what a politician needs to do to improve his or her political standing is do what the pundit wants substantively.” (end copy/paste)
Yeah, that unfortunately does seem to sum up how the majority of Pundits think over the past 20 or so years.
@inhumans99: I think that every self-proclaimed pundit should be forced into 5 years of silence after 5 years of activity.
@Steven L. Taylor:
Looks like the Platonic Form of the Median Voter is a crutch.
Visually and linguistically, our conception of political preferences is a 2D image. But people are 4D.*
Most of the models that try to account for this make the mistake of reducing by too many dimensions. Or at least do not properly account for it.
That does not even get into other distinctions: ir/rational voting behavior and differences between expressed views and behavior.
Our epistemology is all screwed up.
*Maybe even 5D, in terms of a model, assuming time would be the 4th dimension. Ranked order preferences conceived as a rolled up dimension?
@Kurtz: I’ve started thinking in terms of 3D Venn arrays with multiple alternative policy options. People’s views could embrace several options across the void.
@Steven L. Taylor:
Interesting piece, Steven.
This ‘median voter’ idea is, no doubt, great raw material for those who spend their time completely breaking down the numbers, the vote, in districts across the country, but determination of the value and utility of such information is another matter.
There may or may not be actual median voters but I suspect you could come up with something that represents a sum of the median policy positions. It would probably be something like, just to choose a few issues, supports abortion with some restrictions, is concerned about immigration and thinks it would be better if people came legally, supports health care reform but doesnt really want the government running all of it.
Steve
If you rated all of the voters in the US on the left/right scale that Randall Munroe used in his famous infographic, you could find a median voter on that scale. That voter would almost certainly be someone who holds a mix of fairly extreme liberal and conservative views — either socially liberal and fiscally conservative, or a religious conservative with more-Christian-than-most views on wealth redistribution. Not a moderate in any sense of the word.
High-information undecided voters are people who haven’t yet decided whether it’s more important to punish women who have abortions or to have access to healthcare — or some equivalent multidimensional version of that quandary.
Low-information undecided voters are a random sample of low-information people in general, and believe pretty much any combination of things, including all of the conspiracy theories ever.