Hello and TIA,
I am running a fuzzy RD Design with multiple DVs. The treatedment and control variables remain the same for each analysis. The only variable that changes is the dependent variable. When I use the bandwidth selection command (rdbwselect), I receive bandwidths that are quite a bit different.
As you can tell, If I were to use the optimal bandwidth for each DV, I would change my sample and I would have varying treated/untreated. If I wanted to keep my treated/untreated the same for each analysis, which bandwidth of the optimal bandwidths would be optimal? This is my co-author and I's first go for an RD and we were somewhat surprised the optimal bandwidth would change depending on the dependent variable. As we assumed the value of the optimal bandwidth would rely on assignment of treated/untreated based on covariates, the cutpoint, and our fuzzy variable that remain the same in each analysis, we appear to be wrong.
Any insight will help.
I am running a fuzzy RD Design with multiple DVs. The treatedment and control variables remain the same for each analysis. The only variable that changes is the dependent variable. When I use the bandwidth selection command (rdbwselect), I receive bandwidths that are quite a bit different.
DV1
mserd | 1244.699 1244.699 | 2303.498 2303.498
DV2
mserd | 946.039 946.039 | 1878.249 1878.249
DV3
mserd | 909.258 909.258 | 1759.896 1759.896
DV4
mserd | 1329.356 1329.356 | 2373.603 2373.603
DV5
mserd | 1489.024 1489.024 | 2768.890 2768.890
mserd | 1244.699 1244.699 | 2303.498 2303.498
DV2
mserd | 946.039 946.039 | 1878.249 1878.249
DV3
mserd | 909.258 909.258 | 1759.896 1759.896
DV4
mserd | 1329.356 1329.356 | 2373.603 2373.603
DV5
mserd | 1489.024 1489.024 | 2768.890 2768.890
Any insight will help.
