Uniform Punishment and Conditional Reward: Ideological Extremism and Senatorial Approval in the United States

Do constituents hold elected officials accountable for their ideological positions? Using over 1 million Senatorial evaluations from the 2006–2024 Cooperative Election Study, we employ a within‐respondent design that exploits the fact that each constituent evaluates two Senators, absorbing unmeasured respondent characteristics. We find “uniform punishment, conditional reward”: outpartisan constituents penalize extremism by roughly 14 percentage points of Senatorial approval, while only ideologically aligned copartisans (liberal Democrats and conservative Republicans) reward extremism by 22 percentage points in approval. These results are robust to the inclusion of Senator fixed effects, which identify extremism effects from within‐Senator variation over time. Ideological accountability is evident and operates through offsetting channels, which can explain null findings in election‐level research.