The Median Voter Theorem

The median voter theorem as developed by Anthony Downs in his 1957 book, “An Economic Theory of Democracy,” is an attempt to explain why politicians on both ends of the spectrum tend to gravitate towards the philosophical center.  Downs, as well as economist Duncan Black, who proposed the theory in 1948, argue that politicians take political positions are far as possible near the center in order to appeal to as many potential voters as possible.  Under certain constraints/assumptions, Black says, the median voter “wins,” and the outcome ends up as a Nash equilibrium.  These assumptions are as follows:

  1. Voter preferences are single-peaked:  All voters have a single point along a policy position preference curve at which they would receive the highest utility.  The farther a candidate’s policy positions are from this point, the more dissatisfied the voter will be.  
  2. Voter preferences are one-dimensional: Similar in composition to the Hotelling model, the median voter theorem is only able to find an equilibrium when voter preferences are being measured in one-dimension.
  3. There are 2 candidates or parties competing for voters: Each voter must choose between two different policy positions.  Voters will vote for whichever candidate is closer to their highest preference point from the first assumption. 

The polarizing nature of this most recent presidential election however has many experts doubting how empirically applicable the median voter theorem truly is.  The 2016 election was awash with fringe candidates, politicians such as Trump, Bernie, and Jill Stein represented voters with preferences much farther away from the median than had been seen in a long while.  Many argue that this is a result of party stances drifting farther to the right/left, however the positions of Bernie/Trump were substantially outside those even of their own parties.  Why?  Thomas Romer and Howard Rosenthal point out that cartels of “agenda-setters” are able to exploit the differences between median voter’s policy preferences and the actual status quo in order to articulate policies that come as close as possible to their (the agenda setters’) preferences as possible.  In US presidential politics, agenda setters are party activists and special interest groups, to whom candidates need to win favor in order to receive campaign funding.  For the Republican side Daniel Drezner of the Washington Post hypothesizes that fierce competition among GOP candidates for party favor caused many of them to shift “ever rightward” on foreign policy and many other issues.  He goes on to say that once one party begins to shift to one side, the other party is able to shift more and more to their side while still appearing to be relatively moderate.  This would explain much of the early success of Bernie Sanders and some of the more liberal stances taken by Clinton in the later stages of the election.

All-in-all, the median voter theorem does a satisfactory job at explaining why candidates choose the positions that they do, and how voters choose who to vote for.  However, the theorem will soon stand its strongest test as the Trump presidency rumbles on and partisan politics drift farther and farther from the median.


“The End of the Median Voter in Presidential Politics” – Washington Post

“The Median Voter Theorem and its Applications” – Joshua Palette

“Why Politics is Stuck in the Middle” – NYTimes


10 thoughts on “The Median Voter Theorem

  1. I think the point about the drift to the political extremes by the candidates in this past election is spot on. However, I wonder, after a few more elections, if the candidates will move back towards center as the theorem predicts. It seems to me that this is an entirely plausible idea, as recently the American public seems to grow weary of normalized politics. As far-right and far-left politics become more common, perhaps people will desire a political change that moves back toward the “philosophical center.”

  2. Is there a direct correlation between the funding that the “agenda setters” provide and shifting opinions of voters? I understand that candidates will take the position of those that enable their campaigns, but how does that impact the opinions of the people? Does this phenomena effect voter turnout negatively, resulting from low numbers of moderate candidates?

  3. The 2016 election environment where Bernie and Trump deviated significantly from the middle ground is definitely an interesting point, and probably an ending scenario very few could have called. My comments on the post express the same skepticism that you and Pierce express, in regards to whether candidates in the future will continue to build their foundations around what is “not normal”. And yes, this theorem’s ultimate and arguably first major test comes in the form of this recent election Only time can tell.

  4. I’m not sure that second assumption made by the theorem is realistic; that voter preferences are one dimensional. While there are definitely many people who have one issue that is of utmost importance when deciding who to vote for, the fact is the majority of politically active citizens care about the candidates overall platform and message. Similar to most models out there, this one is making a simplifying assumption that may not be valid.

  5. If we had higher turnout in our elections, would we not see results in accord with the median voter theory? As turnout drops, “normal” voters lose salience, and those who are off the chart – literally in this model in another dimension – can and do sway elections.

    Had that not occurred some time back in Republican primaries? The result was the collapse of the entire centrist component of the party. First these voters tended not to be very interested in primaries, they were happy historically with whoever emerged from smoke-filled rooms of party stalwarts. So turnout was low. But low turnout allowed fringe candidates to mobilize fringe voters who might otherwise not have voted in the primary to turn out in numbers sufficient to dominate the process. And now these voters have nowhere to go, unless Democrats could succeed in reaching out to them.

    Bill Clinton managed to do so well enough to do very well. But the Democrats were potentially subject to the same challenge, and indeed lost elections thanks to Ralph Nader and now Bernie Sanders. What is ironic is that the candidates who won were anathema to those outsiders, but they remained obstructionist egoists to the end. The main difference is that to date outsiders have been unable to capture Democratic primaries, whereas obstructionist egoists now dominate the Republican party.

    Apathy has its costs.

  6. I believe this theorem will still hold into the future. I think this election was a phenomenon that occurred because of the individual candidates who ended up in the general. If you look at just the democratic primary, Hillary did win because of this theorem right? While Bernie and Jill Stein went far left, Hillary stayed toward the middle. On the other side this certainly didn’t hold with Trump getting elected, but there were several other factors that made it possible for him to defy the theorem and still win. He knew how to hit on the public’s concerns better than a lot of politicians could. This is a cool article explaining that:

  7. I think an interesting take on this is the application of duverger’s law which is basically a theorem proposing that single-member voting districts lead to two parties which results in a bimodal voting pattern with two peaks at the center of the respective ideologies rather than one peak at the true center. As for this presidency in particular I think you might have seen the peak for the conservative side shift a bit to the right and will correct itself to more center-right rather than true center (which may not exist).

    • In Japan until recently there were many 5-member districts. So you only needed 16.7% of the vote to guarantee election, and when there were a couple popular candidates (2 who combined get 60% of the vote) someone could occasionally get elected with under 10% of the vote. But 20 years after that system ended there are still multiple parties, perhaps because the process of merging small parties didn’t go well, weak leaders and fratricide.

  8. I think that the reason for the large divergence away from the median in the 2016 election can in part be the emergence of social media being many peoples source of news these days. According to 62% of Americans get their news from social media. My opinion is that people only tend to follow what they want to see on social media, which includes their favored news sites/popular celebrities and their viewpoints/internet celebrities etc, and they refrain from following sites/people that have their opposite viewpoint. Because of this, I believe more radical viewpoints are born out of this total immersion of one way of thinking: Left (or right) thinking people see other left (or right) thinking people doing something (e.g. protesting something, pushing agendas on various social issues etc) and thus they either may join the same bandwagon, or look to further it and stand out amongst their social media peers. What seemingly ends up happening is people competing for the most radical or “out there” mindset, and this could be the reason why this years presidential election saw political parties with more extreme views succeed. Thus, in the context of this model, the idea of converging toward the middle may be a thing of the past, unless a scenario like embreyp18 describes above takes place.

  9. I think evidence for the Mean Voter Theorem is still readily prevalent in modern politics. Thinking in terms of issues, I remember the category of “single issue voters” we learned about in high school government. These voters cling to one particular issue like abortion, women’s rights, or gay marriage, and this issue consequently plays a vital role in who a voter selects. All of these issues have a clear distinction between Republicans and Democrats, with some stances being more extreme. One can argue that the further a candidate drifts to the left or right on these issues, there is a greater chance of alienating potential voters, as the Mean Voter Theorem suggests.

    Regarding the past election, I disagree that the Mean Voter Theorem played an unknown role as you suggest. You mention players like Jill Stein, Gary Johnson, Trump, Hillary, and Bernie who all competed for votes, but I think this is too broad of a categorization. One the left for instance, you have Hilary vs. Bernie, and Bernie is no doubt further left, so here the Mean Voter Theorem still applies. Such a huge number of Republican Candidates were present during the primaries that it invalidates the second condition of the theorem, making the theorem irrelevant.

Leave a Comment!