Poll Releases Boost Turnout Intentions by 5%

This letter reports results from a study that combines a unique natural experiment with a local randomization regression discontinuity design to estimate the causal effect of poll releases on turnout intention.

📊 How the effect was identified

A natural experiment generated exogenous variation in the timing of a poll release. A local randomization regression discontinuity approach was used to exploit that variation while avoiding the strong continuity assumptions required by standard RD designs when the running variable is discrete.

🔍 Key findings

• The release of a poll increases turnout intention by 5%.
• The estimated effect is robust to multiple falsification tests of predetermined covariates.
• The effect holds under placebo outcome checks.
• The result remains stable when changing the time window selected to estimate the effect.

🧭 Why it matters

The letter highlights advantages of the local randomization approach over the standard continuity-based RD design for important political science cases with discrete running variables. This method can broaden the set of empirical questions that can be credibly addressed using regression discontinuity techniques.