Length of compulsory education and voter turnout—evidence from a staged reform

Not recorded (individual-level survey analysis plus municipality-level aggregate analysis of Norwegian cohorts)

Norwegian cohorts affected by the staged rollout of the compulsory-schooling reform raising minimum attainment from seven to nine years.

Norway rolled out its compulsory-schooling extension municipality by municipality over years, generating exogenous cohort-by-municipality variation - the same identification logic as the US compulsory-schooling literature but with a cleaner staged design. Pelkonen runs it at BOTH the individual level and the municipality level, which matters because municipality-level turnout is an administrative aggregate rather than self-report, so the null is not an artefact of survey misreporting. READ DEPTH CAVEAT: read at abstract level only, via the published abstract and two independent repository records; coefficients and standard errors were not extracted, and neither were sample sizes. That is recorded debt - the direction of the finding is solid but its precision is not verified here.

Two additional years of compulsory schooling had NO effect on voter turnout, at either the individual or the municipality level. Across a range of other civic outcomes, only the likelihood of SIGNING A PETITION was positively affected. This is the cleanest single refutation of the claim that adding years of schooling produces voters: it is a staged reform with two full extra years of schooling, measured in two independent ways, in a high-turnout democracy, and it moves nothing except the cheapest available form of political expression - which is exactly the pattern Manning and Edwards 2014 report across their systematic review.

Low. The staged municipality-by-cohort rollout is exogenous to individual and family traits, so this null cannot be explained by heritable selection. It is a null under a design that would have detected a real effect.

A direct non-US replication attempt of the Milligan et al. 2004 / Dee 2004 design, and it fails to reproduce the turnout effect. It converges with Milligan et al.'s own UK null, with Lindgren et al. 2019 (Sweden) and with Berinsky and Lenz 2011 (US). The 'American exception' framing in this literature dates largely from this cluster of nulls.