Who Matches? Propensity Scores and Bias in the Causal Effects of Education on Participation
Henderson J, Chatfield S · 2011
grade Ccritiqueindependentnot-applicable
Sample
Re-analysis of the Political Socialization Panel Study data used by Kam and Palmer
Population
Same as Kam and Palmer 2008.
Design
Re-analysis showing that Kam and Palmer's propensity-score result is sensitive to matching specification, and that on reasonable alternatives an education effect reappears. Berinsky and Lenz cite it as agreeing with them that neither side can say anything definitive statistically.
Key findings
Kam and Palmer's null is not robust to matching-specification choices; propensity scores cannot settle whether education causes participation. Useful mainly as evidence that the matching-based arm of this debate is undecidable and that the field needed instruments and lotteries.
Genetic confound
Not assessed in detail; see exclusion note.
Replication notes
A re-analysis rather than an independent study; it disputes Kam and Palmer 2008 rather than replicating it.
Effects
| Outcome | Metric | Value | Measure | Timing | Vs | Horizon | Class |
|---|---|---|---|---|---|---|---|
| Political participation | re-analysis of matched estimates | education effect reappears under alternative specifications; no definitive answer | self-report (panel survey) | adulthood | none | adulthood | behaviour |