In the version of this article initially published, there was a coding error in the anonymization of the non-anonymous data for our analysis.Due to the coding error, duplicates of entries with identical geographic coordinates were introduced in the data.Correcting the error reduces the sample size from 517 to 473.We did not recognize this earlier since it happened in the anonymization step before the analysis.We have re-run all analyses and the conclusions are not affected.
Möhring et al. (Mon,) studied this question.