📊 How voter forecasts were collected

Survey questions that ask who will win an election are common, but they capture winners rather than the underlying vote shares. This study compares that standard qualitative approach with a simple alternative: asking respondents to predict candidates' vote shares.

🔢 Can people predict vote shares?

• Respondents are capable of making sensible, quantitative vote-share predictions.
• Asking for voteshares is a viable measurement strategy that provides more precise information than winner-focused questions.

🔍 Where winner predictions lead analysts astray

• Traditional qualitative (winner) forecasts produce distorted inferences when used as data.
• Specifically, qualitative predictions vastly overstate the degree of partisan bias in election forecasts.
• They also produce misleading conclusions about how political knowledge interacts with bias—suggesting effects that disappear or reverse when voteshares are used instead of winners.

🧪 Practical payoff: elections as natural experiments

• Vote-share predictions improve the use of elections as natural experiments by providing finer-grained measures of electoral closeness and expectation.
• An applied example examines the 2012 election’s effect on partisan economic perceptions, showing how vote-share forecasts clarify causal inference that winner measures obscure.

⚖️ Why this matters

• The choice between asking for winners versus voteshares affects substantive conclusions about bias, knowledge, and electoral effects.
• These findings have implications for methodologists, pollsters, political scientists, and interdisciplinary scholars of collective intelligence who rely on mass election predictions.