Dear Stata Community


After reading the FAQ to the end, I'll try it again, sorry for that.
I use the RDD Stata package provided by Calonico, Cattaneo and Titiunik (2014), to analyze how the transition to daylight saving time (threshold) affects performance in basketball. Distance to transition to DST in days is my running variable. As outcome variables I use field goal percentage and free throw percentage. The percentages are defined as follows: Successfull attempts/ All attempts.

Does anyone know if a percentage as an outcome variable in the RDD can cause some sort of bias? As far as I understand it some problems arise if the percentages take the value of the lower or upper bound. This would be the case for the free throw percentage that sometimes takes the value 100%.

I cannot find any information in literature that deals with fractional dependent variable in RDD. Do you think taking the log would solve the problem?
Or can I somehow argue that the coefficient is valid as long as the fraction es not expected to be near to 100%?

I would be very grateful for any supprt.

Thank you and kind regards
Giulia


Calonico, Cattaneo and Titiunik (2014): Robust Data-Driven Inference in the Regression-Discontinuity Design, Stata Journal 14(4): 909-946.