Does corruption information affect voting behavior? This meta-analysis shows stark differences between survey and field experiments.

Surprising Findings:

* Voters punish corrupt politicians differently depending on the experimental setting:

* Field experiments show minimal effects (-0 pts)

* Survey experiments reveal substantial negative impacts (-32 pts)

Key Takeaways:*

* The discrepancy likely arises from methodological differences between survey and field experiments.

Methodological Challenges:

* Small field experiment effects may result from weak treatments or candidate noncompliance.

* Large survey experiment effects could stem from social desirability bias in responses.

* Hypothetical nature of costs in surveys might inflate perceived punishment.

* Conjoint experiments (costly trade-offs) may not capture real-world voter decision-making accurately.

Implications:

This research suggests that while survey experiments provide useful insights, their results may overestimate the actual impact of corruption information on votes due to artificial contexts. Field experiments offer more realistic estimates but still face design constraints.