Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
| From | Nick Cox <njcoxstata@gmail.com> |
| To | "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu> |
| Subject | Re: st: issues with tsset based on more than a time variable and duplicates conflicting results |
| Date | Mon, 20 Jan 2014 18:16:41 +0000 |
Please don't direct further questions on this as if I will answer them. I've already suggested that you should let Stata technical support look at this (and they will want to see you your data). I have already commented that -egen, group()- is a poor way to get a time identifier. Otherwise you seem to be going round and round the same questions and although I sympathise greatly with your puzzlement I am (as already said) out of ideas on what is going on. Naturally, the field is wide open for ideas from anyone else. Nick njcoxstata@gmail.com On 20 January 2014 18:11, Abdalla, Ahmed <ahmed.abdalla@kcl.ac.uk> wrote: > Nick, to follow up. I tried to play around with it again. > I have -firmid=group(permno) and timeid=group(yr mth) > I have duplicates when I run-duplicates report firmid timeid > and I have no duplicates when I run -duplicates report permno yr mth > > Since I can't -tsset- based on two time variables,i.e. tsset permno yr mth - is incorrect and tsset firmid timeid-doesn't work UNTIL I drop the duplicates on firmid and timeid that don't show up at the same time in duplicates report permno yr mth, I tried the command- tsset permno timeid- i.e. I thought to group only the time variables in one variable and no need to do the same with the panel id variable(permno). Surprisingly, though I don't know if this is correct, it worked properly and no duplicates based on permno timeid. > > Is this correct? Can I do that and proceed with the analysis in my panel dat or do I mix things up that way? do you know why this might even happen? > > > > ________________________________________ > From: owner-statalist@hsphsun2.harvard.edu <owner-statalist@hsphsun2.harvard.edu> on behalf of Nick Cox <njcoxstata@gmail.com> > Sent: 20 January 2014 17:55 > To: statalist@hsphsun2.harvard.edu > Subject: Re: st: issues with tsset based on more than a time variable and duplicates conflicting results > > OK; that rules out that idea. No more ideas from me, sorry. > Nick > njcoxstata@gmail.com > > > On 20 January 2014 17:47, Abdalla, Ahmed <ahmed.abdalla@kcl.ac.uk> wrote: >> Sorry I forgot to write that I get >> >> . tabmiss mth if mth<1 >> Variable | Obs Missings Feq.Missings NonMiss Feq.NonMiss >> -------------+--------------------------------------------------------------- >> mth | 0 0 . 0 . >> >> . tabmiss mth if mth>12 >> Variable | Obs Missings Feq.Missings NonMiss Feq.NonMiss >> -------------+--------------------------------------------------------------- >> mth | 0 0 . 0 . >> >> Is that what you mean ? >> Nick, do you think I can any more investigation to track why these results are conflicting ? >> >> >> >> ________________________________________ >> From: owner-statalist@hsphsun2.harvard.edu <owner-statalist@hsphsun2.harvard.edu> on behalf of Nick Cox <njcoxstata@gmail.com> >> Sent: 20 January 2014 17:43 >> To: statalist@hsphsun2.harvard.edu >> Subject: Re: st: issues with tsset based on more than a time variable and duplicates conflicting results >> >> You haven't reported on whether there are non-missing values of -mth- >> other than 1 to 12, i.e. <1 and >12. >> >> There is only one way to -tsset- panel data, with a panel identifier >> and a time identifier. There is no syntax for three variables; the >> -help- tells you that. >> >> Nick >> njcoxstata@gmail.com >> >> >> On 20 January 2014 17:38, Abdalla, Ahmed <ahmed.abdalla@kcl.ac.uk> wrote: >>> I run- tabmiss firmid timeid- and -tabmiss permno yr mth- and find no missing values in both cases. >>> I run again duplicates report permno yr mth-and-duplicates report firmid timeid- in the fist case, I get no duplicates, however in the second case I get 510,000 duplicates !! >>> >>> Can I tsset my panel based on permno yr mth, and avoid the grouping I have done? >>> >>> >>> >>> ________________________________________ >>> From: owner-statalist@hsphsun2.harvard.edu <owner-statalist@hsphsun2.harvard.edu> on behalf of Nick Cox <njcoxstata@gmail.com> >>> Sent: 20 January 2014 17:28 >>> To: statalist@hsphsun2.harvard.edu >>> Subject: Re: st: issues with tsset based on more than a time variable and duplicates conflicting results >>> >>> A wild guess is to check for missing values on these variables, and >>> for rogue values of -mth- (missing, <1, >12). >>> >>> Nick >>> njcoxstata@gmail.com >>> >>> >>> On 20 January 2014 17:10, Abdalla, Ahmed <ahmed.abdalla@kcl.ac.uk> wrote: >>>> Dear Statalist >>>> I want to tsset my data based on permno yr mth: >>>> I tried - tsset permno yr mth - I get the error message "too many varaibles specified" >>>> I tried - gen firmid= group(permno) >>>> gen timeid = ym(yr, mth) >>>> tsset firmid timeid - I get the error " repeated time values with panel >>>> So I tried to investigate my duplicates : >>>> I run the command: >>>> duplicates report firmid timeid >>>> I get >>>> copies | observations surplus >>>> ----------+--------------------------- >>>> 1 | 2181223 0 >>>> 2 | 53712 26856 >>>> 3 | 16515 11010 >>>> 4 | 9556 7167 >>>> 5 | 5510 4408 >>>> 6 | 1698 1415 >>>> 7 | 196 168 >>>> 8 | 48 42 >>>> 9 | 18 16 >>>> >>>> I drop my duplicates and tsset my data, it works properly. But I though to investigate the duplicates again and run this code (of course before dropping my duplicates): >>>> duplicates report permno mth yr, I get: >>>> >>>> copies | observations surplus >>>> ----------+--------------------------- >>>> 1 | 2268476 0 >>>> >>>> >>>> Why both duplicates drop based on firmid and timeid versus permno yr mth are different though the firm id groups permno and the timeid groups yr and mth ? >>>> Is there any other way to tsset my data based on permno yr mth rather than the grouping I have done (firmid and timeid) ? >>>> >>>> >>>> Thanks >>>> >>>> >>>> >>>> >>>> >>>> >>>> * >>>> * For searches and help try: >>>> * http://www.stata.com/help.cgi?search >>>> * http://www.stata.com/support/faqs/resources/statalist-faq/ >>>> * http://www.ats.ucla.edu/stat/stata/ >>> * >>> * For searches and help try: >>> * http://www.stata.com/help.cgi?search >>> * http://www.stata.com/support/faqs/resources/statalist-faq/ >>> * http://www.ats.ucla.edu/stat/stata/ >>> >>> * >>> * For searches and help try: >>> * http://www.stata.com/help.cgi?search >>> * http://www.stata.com/support/faqs/resources/statalist-faq/ >>> * http://www.ats.ucla.edu/stat/stata/ >> * >> * For searches and help try: >> * http://www.stata.com/help.cgi?search >> * http://www.stata.com/support/faqs/resources/statalist-faq/ >> * http://www.ats.ucla.edu/stat/stata/ >> >> * >> * For searches and help try: >> * http://www.stata.com/help.cgi?search >> * http://www.stata.com/support/faqs/resources/statalist-faq/ >> * http://www.ats.ucla.edu/stat/stata/ > * > * For searches and help try: > * http://www.stata.com/help.cgi?search > * http://www.stata.com/support/faqs/resources/statalist-faq/ > * http://www.ats.ucla.edu/stat/stata/ > > * > * For searches and help try: > * http://www.stata.com/help.cgi?search > * http://www.stata.com/support/faqs/resources/statalist-faq/ > * http://www.ats.ucla.edu/stat/stata/ * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/faqs/resources/statalist-faq/ * http://www.ats.ucla.edu/stat/stata/