Median Voter Theorem vs Condorcet Paradox
Median Voter Theorem and Condorcet Paradox are two Game Theory & Information concepts in AP Economics that students often mix up. The median voter theorem says that under majority rule with single-peaked preferences, the outcome chosen matches the preference of the median voter. The Condorcet paradox is when majority preferences cycle (A beats B, B beats C, C beats A) even though each individual voter has consistent rankings. Here is how they compare side by side.
When voters' preferences over a one-dimensional issue (like a budget size) are single-peaked, the option most preferred by the median voter beats every alternative in pairwise majority voting. This predicts that two competing candidates converge toward the center to capture that pivotal voter, a key result in public choice popularized by Anthony Downs. It breaks down when preferences are multi-peaked or the issue space is multidimensional.
Majority rule can produce intransitive group preferences: a majority prefers A to B, another majority prefers B to C, yet another prefers C to A, so no option is a stable winner. This 'voting cycle' means the outcome can depend on the agenda or order of votes rather than true preferences. It is the simplest illustration of why fair aggregation of preferences is hard and is the seed of Arrow's impossibility theorem.
Median Voter Theorem vs Condorcet Paradox: When Majority Rule Settles and When It Circles
| Median Voter Theorem | Condorcet Paradox | |
|---|---|---|
| What it predicts | One stable winner, sitting at the median voter's ideal point | No stable winner, because majorities cycle from one option to the next |
| Preferences it assumes | Single-peaked preferences over a single dimension | Arises once preferences are not single-peaked or the issue has more than one dimension |
| What a head-to-head vote shows | The median option beats every rival it is paired against | Every option loses to some other option |
| What settles the outcome | Voter preferences on their own | The order in which the pairwise votes are scheduled |
| Where the power sits | With the voter in the middle | With whoever writes the agenda |
| Lesson for candidates | Move toward the centre of the distribution | Fight over the running order before you fight over the policy |
The median wins because any challenger gives away one whole side of the room
Put five council members on one question, how much to spend on a new library, and suppose each has an ideal figure and likes any budget less the further it sits from that figure. Their ideal amounts are 0, 2, 5, 8 and 20 million dollars, all illustrative. The median is 5. Run 5 against a proposal of 4 and the members at 5, 8 and 20 all prefer 5, so it wins three votes to two. Run 5 against 8 and the members at 0, 2 and 5 prefer 5, three to two again. Any challenger you name sits on one side of the median, so it loses every voter on the far side plus the median voter, and three of five is already a majority. Two conditions are doing the work: a single dimension to argue over, and single-peaked preferences, meaning each voter's support falls away steadily on both sides of their own ideal point. Notice how little the extremes matter. Move the last member from 20 to 200 and nothing changes, because the median is still 5. That is why rival candidates tend to crowd toward the middle instead of toward their own strongest supporters, and why intensity of feeling buys nothing under plain majority rule.
Take away the single peak and the majority can prefer everything to something else
Those conditions are not decoration. Take three members choosing among three plans: A is a library, B is a pool, C is no building at all. Member one ranks A above B above C. Member two ranks B above C above A. Member three ranks C above A above B. Every individual ranking is perfectly consistent. Now vote in pairs. A beats B two votes to one, since members one and three both put A first of that pair. B beats C two to one, since members one and two both prefer B. So A looks strongest. Yet C beats A two votes to one, because members two and three both put C above A. Majority preference has closed a circle, and no plan can win every contest it enters. The practical sting is that the running order decides the winner. Pair A against B and send the survivor against C, and C wins. Pair B against C and send the survivor against A, and A wins. Same voters, same honest counting, different outcome. Vote trading of the kind described at /glossary/logrolling grows out of the same soil, and the general version of the problem is set out at /glossary/arrow-s-impossibility-theorem.
Frequently asked questions
What is the difference between the median voter theorem and the Condorcet paradox?
The median voter theorem says majority rule produces a single stable winner at the median voter's position, while the Condorcet paradox shows cases where majority rule produces no stable winner at all because preferences cycle. They are the two possible fates of pairwise majority voting, and which one you get depends on the shape of voter preferences.
Does the median voter theorem always hold?
No. It needs voters to have single-peaked preferences along one dimension, meaning each voter has an ideal point and likes options less as they move away from it in either direction. Add a second dimension, or let a voter like both extremes more than the middle, and the prediction can fail.
Why do majority votes sometimes go in a circle?
Because a group ranking assembled from pairwise majorities does not have to be transitive, even when every individual ranking is. Different majorities decide each pair, so the coalition that puts A above B can be a completely different set of people from the one that puts C above A.
Get AP Econ exam tips in your inbox
Occasional emails with study tips, new interactive graphs, and exam-season reminders. Free, no spam.
No spam. Unsubscribe anytime. Read our privacy policy.
Keep track of what you have studied
A free EconLearn account adds progress tracking, your quiz history, and achievements. Studying here is free either way, and there is nothing to pay for as a student.
Create a free accountAlready have one? Sign in
Last updated