The median voter theorem states that in a two-party plurality election with single-peaked voter preferences, both parties will converge to the policy position of the median voter—the voter whose preferences lie exactly in the middle of the distribution. Downs argued that because parties want to maximize votes, they will move toward the center to capture the median voter, leaving little policy differentiation. This prediction has been challenged by polarization and primary elections, but it remains the baseline spatial model for understanding party competition and explains why centrist platforms are often electorally successful.

Downs identified the paradox that rational self-interest predicts near-zero voting: the probability that one vote will decide an election is infinitesimal, while the costs of voting (time, information, transportation) are positive, so the expected benefit is negative. Yet millions vote. Downs resolves this by introducing 'civic duty' or 'consumer surplus'—voters derive intrinsic satisfaction from participating, effectively treating voting as a consumption good. This addition preserves rational choice theory while acknowledging that purely instrumental calculations fail to explain turnout.

Downs rejected explanations based on party identification, social class, or group loyalty—dominant in the Columbia and Michigan schools—arguing instead that voters are instrumentally rational: they choose based on expected policy outcomes. Where sociological models emphasize 'who you are' (class, religion, ethnicity), Downs emphasizes 'what you get' (policies, benefits, income). He reduced voting to an investment decision: citizens expend resources to purchase a future policy stream. This shift from expressive to instrumental voting fundamentally reoriented electoral research toward formal modeling and preference aggregation.

Critics argue Downs assumes unrealistic levels of voter information and rationality, ignoring cognitive biases, low information, and the influence of media and elite framing. The median voter theorem fails under multi-party systems or when preferences are multi-dimensional or non-single-peaked. Empirical research shows partisanship and identity often override policy proximity. Additionally, Downs's theory struggles with expressive voting—people voting for symbolic or moral reasons rather than policy consequences. Despite these criticisms, his framework remains the unavoidable starting point for formal political theory.

Downs's model predicts centrist convergence, yet contemporary U.S. politics exhibits extreme polarization. Extensions of his theory explain this through primaries: when candidates must win a partisan primary before the general election, they face a selectorate to their ideological extreme, reducing the median voter's influence in the general election. Campaign finance and gerrymandering further amplify this effect. While Downs's original model holds for general elections in competitive two-party systems, it requires modification when institutional filters distort the distribution of preferences that candidates actually face.

Downs, along with James Buchanan and Gordon Tullock, founded the public choice tradition—applying economic assumptions to political decision-making. While Buchanan and Tullock focused on collective action problems and constitutional constraints (*The Calculus of Consent*, 1962), Downs specialized in party competition and voter behavior. Buchanan emphasized self-interested bureaucrats and rent-seeking; Downs emphasized information costs and rational ignorance. Together, they established that political actors, like market actors, respond to incentives, undermining the romantic view of politics as purely public-spirited.

Many assume Downs argues that voters are selfish, informed, and make precise calculations. He actually argued the opposite: rational citizens often remain 'rationally ignorant' because information costs outweigh expected benefits. He described voters as having high 'entry costs' into politics and often relying on 'information shortcuts' like party labels and candidate images. His model is not about perfectly informed citizens but about how imperfectly informed citizens make choices under conditions of uncertainty—a nuanced position frequently lost in simplified critiques.