Abstract
We analyse the effect of removing re-election incentives on the individual legislative participation of state lawmakers by using an original study based on the random assignment of term length that occurs in the Arkansas Senate, which in turn induces the random assignment of the total number of terms each senator may serve. Across five measures of legislative output—bills introduced, bills passed, bills cosponsored, resolutions and abstention rates, we cannot reject the null hypothesis of no effect. Since our sample is small, we adopt two strategies in our statistical analysis: we perform randomization-based inference to ensure that our tests adequately control size, and we use tests of equivalence to avoid incorrectly concluding that the effects are null because of low power. We also use bounds as a robustness check to address attrition in our original experimental sample.
Original language | English (US) |
---|---|
Pages (from-to) | 351-378 |
Number of pages | 28 |
Journal | Journal of the Royal Statistical Society. Series A: Statistics in Society |
Volume | 181 |
Issue number | 2 |
DOIs | |
State | Published - Feb 1 2018 |
Externally published | Yes |
All Science Journal Classification (ASJC) codes
- Statistics and Probability
- Social Sciences (miscellaneous)
- Economics and Econometrics
- Statistics, Probability and Uncertainty
Keywords
- Bounds
- Fisherian inference
- Legislative behavior
- Term limits
- Tests of equivalence