Hi,
I am finding a causal effect on education on fertility by using regression discontinuity design. First, we need to check whether it is sharp or Fuzzy using this command. where X is running variable and treatment is the treatment variable.
* 1. Visual inspection
rdplot treatment X, c(0)
The graph is attached.
It shows clear sharp RDD
Further, I did by these codes
***Check for perfect compliance
sum X if treatment==0
sum X if treatment==1
sum X if treatment==0
Variable | Obs Mean Std. dev. Min Max
-------------+---------------------------------------------------------
X | 329 -25.42249 12.23204 -43 -1
. sum X if treatment==1
Variable | Obs Mean Std. dev. Min Max
-------------+---------------------------------------------------------
X | 103 15.8835 13.11959 0 56
These results also show it is a sharp RDD.
But the point is that because my outcome variable is fertility and independent variable is education attainment and the running variable is the woman's year of birth (capture the time of education reform). I have specific cutoff dates. Nevertheless, may have people on the left and the right side of the cutoffs with similar levels of education. for this reason, can I use the fuzzy option? considering education, I have people, both treated and untreated, who acquired similar levels of education because of this reason, can I use the fuzzy option?
I am finding a causal effect on education on fertility by using regression discontinuity design. First, we need to check whether it is sharp or Fuzzy using this command. where X is running variable and treatment is the treatment variable.
* 1. Visual inspection
rdplot treatment X, c(0)
The graph is attached.
It shows clear sharp RDD
Further, I did by these codes
***Check for perfect compliance
sum X if treatment==0
sum X if treatment==1
sum X if treatment==0
Variable | Obs Mean Std. dev. Min Max
-------------+---------------------------------------------------------
X | 329 -25.42249 12.23204 -43 -1
. sum X if treatment==1
Variable | Obs Mean Std. dev. Min Max
-------------+---------------------------------------------------------
X | 103 15.8835 13.11959 0 56
These results also show it is a sharp RDD.
But the point is that because my outcome variable is fertility and independent variable is education attainment and the running variable is the woman's year of birth (capture the time of education reform). I have specific cutoff dates. Nevertheless, may have people on the left and the right side of the cutoffs with similar levels of education. for this reason, can I use the fuzzy option? considering education, I have people, both treated and untreated, who acquired similar levels of education because of this reason, can I use the fuzzy option?
Attached Files
