“…Few other concepts in time series cause me as much conceptual and existential uncertainty and as many sleepless nights as that of unit root. There is long-standing argument among methodologists concerning whether political time series contain a 'unit root' (i.e., are integrated) or, instead, are fractional, near-integrated, or stationary processes (e.g., Beck, 1992, Box-Steffensmeier and Smith, 1996, De Boef, 2000, De Boef and Granato, 1997, Lebo et al, 2000, Wlezien, 2000 4 -that is, whether the effect of shocks to variables are additive and are therefore preserved in observations in future time periods. In non-technical jargon, unit root relates to whether a variable -such as public opinion or public policy -has a memory, insomuch as it is the sum of past values of itself plus random variation.…”