Dear all,
I am asking to help me with the following:
In making a sharp RDD for a clear cutoff point (c) at a 150 testscore (running variable): I have panel data, that is 170 events going from t=1 till t=7. Thus, total observations is 7*170=1190. No missing data and all data is continious. 0 duplicates.
I've created, and scattered, testscorecentered (score-150) on treatment, and testscorecentered is like expected exactly 0 when treatment goes from 0 to 1. The range of testscorecentered is from -146 untill +148, and the cutoff is thus at exactly 150. Thus, the original testscore range before centering was +4 untill +298.
After having obtained a "good" p value at the manipulation test (0.57), I am stuck on the Rdbwselect part - the essential part to estimate the bandwith -and thus I am unable to perform the RdRobust and RdPlot afterwords. Given my research question is effect of CEO testscore on firm revenued, my command is "rdbwselect revenue testcentered, kernel(uniform) c(0) p(1) bwselect(mserd) masspoints(check))", i.e. exactly according to the structure of the help book of the RDD manual in stata.
From that command, the three errors I keep getting at Rdbwselect are:
''Invertibility problem in the computation of preliminary bandwidth. Try checking for mass points with option masspoints(check).
Not enough variability to compute the bias bandwidth (b). Try checking for mass points with option masspoints(check).
Not enough variability to compute the loc. poly. bandwidth (h). Try checking for mass points with option masspoints(check).''
I have tried everything: narrowing down the dataset more, different type of masspoint specifications, different type of values of p: nothing worked.
Perhaps logically, when I try to still run Rdplot, RdRobust, I as well encounter errors, altough different ones:
''Not enough variability to compute the preliminary bandwidth. Try checking for mass points with option masspoints(check).
Not enough variability to compute the bias bandwidth (b). Try checking for mass points with option masspoints(check).
Not enough variability to compute the loc. poly. bandwidth (h). Try checking for mass points with option masspoints(check).
Invertibility problem: check variability of running variable around cutoff. Try checking for mass points with option masspoints(check)''
My question is: would anybody have any idea why these errors are happening - despite me having a ''good'' manipulation test and ''good'' scatter of aumcentered on treatment? Is it the small amount of observations, or the original range is too wide, or other reasons you think? I've searched off all Stata forums but no fitting answer found. Below, if it helps, is the Excel raw data that I use.
etc
I really appreciate your time and effort very much!!!
I am asking to help me with the following:
In making a sharp RDD for a clear cutoff point (c) at a 150 testscore (running variable): I have panel data, that is 170 events going from t=1 till t=7. Thus, total observations is 7*170=1190. No missing data and all data is continious. 0 duplicates.
I've created, and scattered, testscorecentered (score-150) on treatment, and testscorecentered is like expected exactly 0 when treatment goes from 0 to 1. The range of testscorecentered is from -146 untill +148, and the cutoff is thus at exactly 150. Thus, the original testscore range before centering was +4 untill +298.
After having obtained a "good" p value at the manipulation test (0.57), I am stuck on the Rdbwselect part - the essential part to estimate the bandwith -and thus I am unable to perform the RdRobust and RdPlot afterwords. Given my research question is effect of CEO testscore on firm revenued, my command is "rdbwselect revenue testcentered, kernel(uniform) c(0) p(1) bwselect(mserd) masspoints(check))", i.e. exactly according to the structure of the help book of the RDD manual in stata.
From that command, the three errors I keep getting at Rdbwselect are:
''Invertibility problem in the computation of preliminary bandwidth. Try checking for mass points with option masspoints(check).
Not enough variability to compute the bias bandwidth (b). Try checking for mass points with option masspoints(check).
Not enough variability to compute the loc. poly. bandwidth (h). Try checking for mass points with option masspoints(check).''
I have tried everything: narrowing down the dataset more, different type of masspoint specifications, different type of values of p: nothing worked.
Perhaps logically, when I try to still run Rdplot, RdRobust, I as well encounter errors, altough different ones:
''Not enough variability to compute the preliminary bandwidth. Try checking for mass points with option masspoints(check).
Not enough variability to compute the bias bandwidth (b). Try checking for mass points with option masspoints(check).
Not enough variability to compute the loc. poly. bandwidth (h). Try checking for mass points with option masspoints(check).
Invertibility problem: check variability of running variable around cutoff. Try checking for mass points with option masspoints(check)''
My question is: would anybody have any idea why these errors are happening - despite me having a ''good'' manipulation test and ''good'' scatter of aumcentered on treatment? Is it the small amount of observations, or the original range is too wide, or other reasons you think? I've searched off all Stata forums but no fitting answer found. Below, if it helps, is the Excel raw data that I use.
| period | event | testscore | revenue |
| 1.00 | 1.00 | 75.02 | 732.62 |
| 2.00 | 1.00 | 106.71 | 804.99 |
| 3.00 | 1.00 | 174.23 | 644.68 |
| 4.00 | 1.00 | 203.62 | 644.30 |
| 5.00 | 1.00 | 215.63 | 956.10 |
| 6.00 | 1.00 | 223.98 | 944.83 |
| 7.00 | 1.00 | 237.19 | 949.54 |
| 1.00 | 2.00 | 173.07 | 62.95 |
| 2.00 | 2.00 | 188.65 | 124.12 |
| 3.00 | 2.00 | 188.17 | 98.12 |
| 4.00 | 2.00 | 237.76 | 117.57 |
| 5.00 | 2.00 | 211.61 | 34.40 |
| 6.00 | 2.00 | 234.88 | 21.93 |
| 7.00 | 2.00 | 147.92 | 69.98 |
| 1.00 | 3.00 | 180.21 | 48.86 |
| 2.00 | 3.00 | 105.64 | 69.60 |
| 3.00 | 3.00 | 165.10 | 67.88 |
| 4.00 | 3.00 | 157.60 | 66.96 |
| 5.00 | 3.00 | 248.04 | 186.41 |
| 6.00 | 3.00 | 197.94 | 181.84 |
| 7.00 | 3.00 | 236.26 | 149.52 |
I really appreciate your time and effort very much!!!
