Strategic voting or confounding?

Survey research yields the surprising conclusion that voters strategically abandon small parties to a similar degree in first-past-the-post (FPTP) and proportional representation (PR) systems. If true, this challenges Duverger's classic theory that strategic voting helps explain why PR systems have more parties. We argue instead that observational analyses of FPTP and PR elections produce similar patterns of apparently strategic voter behavior because they suffer from the same methodological flaw. After reproducing previous findings in a more comprehensive dataset, we show via placebo tests that the same research design would erroneously imply strategic behavior in non-strategic outcomes like party identification. We also show that a more rigorous design reduces but does not eliminate the problem. Our findings suggest that isolating the effect of party popularity on vote choice using observational data may be fundamentally more difficult than the literature has recognized, and that alternative approaches are needed.