Hi all,
I understand that when you add controls into a RDD, it implies that you think your assumption of continuity might not hold, so need to control for covariates. Does the same apply to controlling for year fixed effects, for example? Does this imply you think there might be some omitted variables/treatments varying at the cutoff so you control for fixed effects to make sure these don't bias your results, ie. giving up on the continuity assumption?
I understand that when you add controls into a RDD, it implies that you think your assumption of continuity might not hold, so need to control for covariates. Does the same apply to controlling for year fixed effects, for example? Does this imply you think there might be some omitted variables/treatments varying at the cutoff so you control for fixed effects to make sure these don't bias your results, ie. giving up on the continuity assumption?
