Dear Statalist Users,
I am working with a judiciary dataset, with my dependent variable being 'Days for Decision'. My data spans from 2010 till 2018 and contains cases filed every year.
I have two broad aims:
1. To find what factors affect the days for decision - such as genders of the litigants etc., and
2. To run a DID on a change in this 'speed' since an amendment passed in early 2013, hence 2013-2018 is the post period. This amendment targeted only specific crimes, hence those crimes are my treatment group.
Since this is a count variable, I want to use a Poisson regression for both of my analyses. The dataset is not panel structure to my knowledge.
I want to include fixed effects of 1. district, and 2. crime sections, since longer cases might be biasing the estimates.
I am not sure which one out of -ppmlhdfe-, -xtpqml-, and -xtpoisson- is suitable for my analysis. I think that the latter two allow only for the panel variables fixed effects. I also want to cluster my standard errors at a district level since the data is of district and sub-district courts. Again, -xtpoisson- does not allow for this and -xtpqml- says that group variable (i) must be nested within clusters.
In this regard, which command would be the most suitable to use? Additionally, if I add something like i.section to -xtpqml-, is there going to be a difference between the estimates from -xtpqml- and -ppmlhdfe-?
Apologies if I haven't been clear, I would be happy to clarify on any aspect as required.
Attaching a small sample of the data just in case:
I am working with a judiciary dataset, with my dependent variable being 'Days for Decision'. My data spans from 2010 till 2018 and contains cases filed every year.
I have two broad aims:
1. To find what factors affect the days for decision - such as genders of the litigants etc., and
2. To run a DID on a change in this 'speed' since an amendment passed in early 2013, hence 2013-2018 is the post period. This amendment targeted only specific crimes, hence those crimes are my treatment group.
Since this is a count variable, I want to use a Poisson regression for both of my analyses. The dataset is not panel structure to my knowledge.
I want to include fixed effects of 1. district, and 2. crime sections, since longer cases might be biasing the estimates.
I am not sure which one out of -ppmlhdfe-, -xtpqml-, and -xtpoisson- is suitable for my analysis. I think that the latter two allow only for the panel variables fixed effects. I also want to cluster my standard errors at a district level since the data is of district and sub-district courts. Again, -xtpoisson- does not allow for this and -xtpqml- says that group variable (i) must be nested within clusters.
In this regard, which command would be the most suitable to use? Additionally, if I add something like i.section to -xtpqml-, is there going to be a difference between the estimates from -xtpqml- and -ppmlhdfe-?
Apologies if I haven't been clear, I would be happy to clarify on any aspect as required.
Attaching a small sample of the data just in case:
Code:
* Example generated by -dataex-. For more info, type help dataex clear input float decision_time1 int(female_defendant female_petitioner section_no) long dist_code 2979 0 . 379 4 1 0 . 366 4 1905 0 . 290 27 1905 1 . 290 27 1902 0 . 290 27 2502 0 0 420 27 2508 . 0 392 31 1356 0 1 506 8 1041 0 . 290 8 1604 0 1 353 8 1696 0 0 506 8 1052 0 . 290 8 904 0 . 278 27 1011 0 . 290 27 1 0 . 290 27 1384 . . 325 14 1 0 1 506 56 1 . 0 438 43 1 0 . 290 27 1 . . 290 27 1195 . . 406 14 1 . . 290 27 381 0 . 8 6 381 0 . 17 6 1321 0 . 294 14 381 1 . 70 6 1928 0 . 279 9 1959 . . 358 31 367 0 . 307 22 367 1 . 399 22 1026 0 . 283 9 2377 0 0 188 28 2503 0 0 324 28 2060 1 0 384 28 1997 1 . 379 28 1861 0 1 354 28 2095 0 1 452 28 2782 0 0 148 28 1998 0 0 394 28 1920 0 1 380 28 1991 0 0 307 28 1984 0 1 325 28 2184 0 0 420 28 1976 0 0 380 28 1983 0 0 452 28 2016 0 0 506 28 2487 0 0 452 28 1956 1 1 307 28 2457 1 0 380 28 1887 0 1 307 28 1807 0 0 323 28 1877 1 0 379 28 1899 0 0 420 28 1975 1 1 452 28 2065 0 0 467 28 1831 0 0 420 28 1925 0 0 467 28 1 . . 290 27 1 0 . 290 27 1 0 . 290 27 1 . . 290 27 1 0 . 290 27 1 . . 290 27 1 0 . 290 27 1 0 . 290 27 1 1 . 290 27 1 . . 290 27 1 0 . 290 27 1 0 . 290 27 1 1 . 290 27 1 0 . 290 27 1 0 . 290 27 1 1 . 290 27 1 1 . 290 27 1 0 . 290 27 1 0 . 290 27 1 0 . 290 27 1 0 . 290 27 1 0 . 290 27 1 1 . 290 27 1 0 . 290 27 1 0 . 290 27 1 0 . 290 27 1 0 . 290 27 548 0 . 273 1 2558 0 . 379 50 1 0 . 290 27 1 0 . 290 27 1 0 . 290 27 1 0 . 290 27 1 . . 290 27 1 0 . 290 27 1 0 . 290 27 1 0 . 290 27 1 0 . 290 27 1 . . 290 27 1 0 . 290 27 1 1 . 290 27 494 0 . 273 1 1 1 . 290 27 end label values female_defendant female label values female_petitioner female label def female 0 "0 male", modify label def female 1 "1 female", modify label values dist_code dist_code label def dist_code 1 "01", modify label def dist_code 4 "04", modify label def dist_code 6 "06", modify label def dist_code 8 "08", modify label def dist_code 9 "09", modify label def dist_code 14 "14", modify label def dist_code 22 "22", modify label def dist_code 27 "27", modify label def dist_code 28 "28", modify label def dist_code 31 "31", modify label def dist_code 43 "43", modify label def dist_code 50 "50", modify label def dist_code 56 "56", modify
Comment