Electoral outcomes significantly influence voter representation, yet this impact often lasts far longer than anticipated. Using dynamic regression discontinuity designs with combined electoral and legislative roll-call data across the US House, Senate, and state legislatures, we demonstrate that even 'coin-flip' election results maintain their effects for a decade or more—and sometimes as long as three decades—in all settings. Incumbent advantages contribute to this durability.

This study reveals two critical insights: first, elected officials resist adapting their roll-call voting patterns over voter preferences; second, voters do not consistently retaliate by replacing incumbents in subsequent elections. The findings underscore a fundamental disconnect between electoral democracy and legislative responsiveness.