Statistics as If Politics Mattered: A Reply to Fowler and Hall

Anthony Fowler and Andrew Hall question both the statistical validity and the broader significance of our analysis of the electoral impact of shark attacks along the Jersey Shore in 1916. Setting aside the politics and history of the Progressive period, they focus on methodological considerations, carrying out an extensive set of tests intended to cast all-purpose doubt on our interpretation of the electoral evidence. However, we show that this style of analysis leads them not just into historical misjudgments but into statistical lapses as well. Correcting those missteps, and setting aside the substantial share of their evidence that has no bearing on our argument, leaves our conclusions handsomely supported. In the end, we argue, statistical calculations ignoring the relevant politics and history contribute little to scientific understanding.