I am working on large panel data set (consist on 2000 firms spanning 20 years data). My running variable is continuous and its values lies from 0 to 20 (like, 0.00, 0.15, 0.20, 1.20, ......17.20, 2.22, 3.30, 19.27.....etc.).
In my context the cutoff-value is 60th percentile value and in our case 60th percentile value of running variable = 0.54, even almost 20% the running variable observations are lies between 1 to 20. So, when I use the comogram command by using c(0.54) for identification of discontinuity in outcome variable before-or-after cutoff. Get the following below image as output, but its not look appropriate due to cutoff line. It will looks appropriate if cutoff appears in center or near to the center of the graph.
So, my Question is that, is it possible use percentiles score (i.e. 10th, 20th, ........90th, 100th) on xaxis instead of RV values? which our graph most appropriate, or any other suggestion which help for identification of discontinuity in Treatment and Control Group.
Thanks
