We join the growing consensus in political methodology advocating for approaches to modeling time series that do not hinge on unit root tests. Our approach differs from others in its simplicity. We recommend the use of a general modeling strategy, an autoregressive distributed lag model with multiple lags of the outcome and predictor(s). We demonstrate, mathematically and through simulations, that such a model can account for serial correlation and avoid spurious inferences. This means a versa- tile model specification that eschews unit root testing. We also demonstrate that our suggested modeling approach provides more accurate inferences than the bounds approach suggested by Webb, Linn and Lebo (2019, 2020), the latest in a series of papers addressing the problems created by unreliability in unit root testing.