Social Sciences History Seminar
Abstract: An influential literature uses "distance from" as an instrumental or a control variable. I show that "distance from Wittenberg" and "distance from Mainz," two prominent instruments for the adoption of Protestantism during the Reformation and printing, respectively, have historical and econometric drawbacks that generate misleading results. First, printing data challenges the idea that Protestant ideas (or printing technology) diffused outward from Wittenberg (or Mainz) in concentric circles. Second, replication exercises and Monte Carlo simulations suggest that the statistical significance of "distance from Wittenberg" in first stage and OLS regressions reflects spatial autocorrelation. "Distance from" variables often exhibit statistical significance even when they have no substantive impact on an outcome. This property affects the second stage results. My findings suggest caution in using "distance from" as an instrument for the adoption of ideas and technologies.