Political scientists commonly use Grambsch and Therneau’s (1994, Biometrika 81, 515–526) ubiquitous Schoenfeld-based test to diagnose proportional hazard violations in Cox duration models. However, some statistical packages have changed how they implement the test’s calculation. The traditional implementation makes a simplifying assumption about the test’s variance–covariance matrix, while the newer implementation does not. Recent work suggests the test’s performance differs, depending on its implementation. I use Monte Carlo simulations to more thoroughly investigate whether the test’s implementation affects its performance. Surprisingly, I find the newer implementation performs very poorly with correlated covariates, with a false positive rate far above 5%. By contrast, the traditional implementation has no such issues in the same situations. This shocking finding raises new, complex questions for researchers moving forward. It appears to suggest, for now, researchers should favor the traditional implementation in situations where its simplifying assumption is likely met, but researchers must also be mindful that this implementation’s false positive rate can be high in misspecified models.

Keele (2010, Political Analysis 18:189–205) emphasizes that the incumbent test for detecting proportional hazard (PH) violations in Cox duration models can be adversely affected by misspecified covariate functional form(s). In this note, I reevaluate Keele’s evidence by running a full set of Monte Carlo simulations using the original article’s illustrative data-generating processes (DGPs). I make use of the updated PH test calculation available in R’s survival package starting with v3.0-10. Importantly, I find the updated PH test calculation performs better for Keele’s DGPs, suggesting its scope conditions are distinct and worth further investigating. I also uncover some evidence for the traditional calculation suggesting it, too, may have additional scope conditions that could impact practitioners’ interpretation of Keele (2010). On the whole, while we should always be attentive to model misspecification, my results suggest we should also become more attentive to how frequently the PH test’s performance is affected in practice, and that the answer may depend on the calculation’s implementation.