Reconsidering the Effects of Education on Political Participation

Political Socialization Panel Study (1965 high-school seniors re-interviewed 1973)

US high-school seniors of 1965 followed into adulthood.

Propensity-score matching on pre-adult characteristics, treating college attendance as the treatment. Concludes that higher education does not cause participation once pre-adult experiences are matched. The design cannot rule out matching on the wrong covariates, and the rebuttal literature says it does not survive alternative specifications.

Once pre-adult characteristics are matched, college attendance has little or no effect on political participation - education is a proxy for prior disposition. Influential as the opening move of the 'education as proxy' revisionist literature that Berinsky and Lenz later carried with a stronger instrument.

Not assessed in detail; see exclusion note.

Directly contested by Henderson and Chatfield 2011, who argue the matching approach is sensitive to specification and that the null does not hold up. Berinsky and Lenz 2011 reach a similar conclusion by a stronger design.