Field Notes

concept

Median voter

StatusStableUpdatedStands on4 sourcesSupportsnotes

The median voter model is the workhorse result of spatial voting theory: when voters have single-peaked preferences over a one-dimensional policy space and two office-seeking candidates compete under majority rule, the outcome converges to the policy preferred by the median voter. The result anchors the intuition that democratic competition pulls policy toward the center, and it gives the vault a precise way to talk about who the politically decisive voter is — including, in Boomers and democracy, how that voter ages.

The theorem and its conditions

Duncan Black proved the core result in 1948: if every voter’s preferences are single-peaked along one dimension — each voter has one ideal point and prefers alternatives closer to it on either side — then the median voter’s ideal point is a Condorcet winner, an option that beats every rival in a pairwise majority vote.1 Formally, with voters ordered by ideal point and an odd electorate,

xmed≻y for every y≠xmed

because the median voter plus everyone on the far side of any rival proposal forms a majority against it. Harold Hotelling had already shown in 1929 that two competing sellers on a line drift to the center, and Anthony Downs transposed the argument to elections in An Economic Theory of Democracy (1957): two vote-maximizing candidates who can commit to platforms both adopt the median position, since any other stance lets the rival win by moving slightly closer to the middle.2

What the model explains is convergence and moderation: why two-party races so often produce near-identical platforms, why parties chase the swing region rather than mobilize their edges, and why shifts in the electorate’s composition — its income distribution, age structure, or turnout pattern — predict shifts in policy even with no change in anyone’s principles.

Where the model breaks

Every clause of the theorem is load-bearing, and each fails in practice:

  • More than one dimension. Once policy bundles two or more dimensions, a Condorcet winner generically fails to exist: McKelvey’s and Schofield’s chaos theorems show that majority rule can cycle from any point to any other, handing the outcome to whoever controls the agenda.3
  • More than two candidates. With three candidates, locating at the median lets an entrant win by sitting just beside it, so convergence is no longer an equilibrium.
  • Who turns out. The decisive voter is the median of voters, not of citizens; groups with lower turnout shift the effective median away from the demographic median without moving anyone’s preferences.
  • Primaries and party activists. Candidates must first survive selectorates that sit off the general-election median, pulling platforms toward each party’s base.
  • Money, valence, and commitment. Campaign resources, candidate quality, and the inability to credibly commit to a platform all loosen the median’s grip.

The robust lesson is narrower than the theorem: competition creates gravitational pull toward whoever can assemble or deny a majority, and identifying that pivotal voter is analytically useful even when no literal median exists.

The vault’s use

Boomers and democracy invokes an ageing median voter: as a large birth cohort and rising longevity push the electorate’s median age upward, the pivotal voter’s position in the life course shifts, and policy follows — protecting pensions and incumbent asset stocks while adjustment lands on later entrants, one mechanism of Intergenerational policy capture. The relationship should be stated precisely. That mechanism needs only pivotality — a cohort that is large relative to its neighbors, turns out reliably, and is organized — not a literal median voter in Black’s sense. Swedish and British politics are multidimensional, multi-party, and coalition-governed, so the theorem’s conditions do not hold; what survives is the generalization that electoral leverage tracks who is pivotal, which is why the same note warns that age-based turnout and wealth patterns alone do not identify a cohort effect.

  1. Duncan Black, On the rationale of group decision-making, Journal of Political Economy 56(1), 1948. ↩

  2. Median voter theorem, overview of the Hotelling–Downs convergence result. ↩

  3. “McKelvey–Schofield chaos theorem”, Wikipedia. A tertiary source standing in for the underlying results of Richard McKelvey (1976) and Norman Schofield (1978), which the wiki has not read directly. ↩

Built on 4 sources (4 external).

Working out connections…