Online Correction to “Social Media and Protest Participation: Evidence from Russia”

This document corrects an error in the construction of two control variables used in Enikolopov et al. (2020). While working on a new project that uses data from Enikolopov et al. (2020) it was noticed that the coordinates of 58 out of 625 cities in the sample were incorrect. As a result, two of the control variables used in the paper—spherical distances to Moscow and Saint Petersburg— were computed erroneously. Since these control variables were part of the baseline specification, the authors feel the need to issue a correction. In this online correction, we proceed as follows. In Section 1, we describe the issue of near multicollinearity that arises once one uses the corrected city coordinates and argue that the underlying cause is near multicollinearity rather than a substantive relation between the instrument and the endogenous variable. In Section 2, we propose alternative approaches to solve this issue and indicate our preferred correction; all results remain qualitatively unchanged.