Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | "Martin Weiss" <martin.weiss1@gmx.de> |
To | <statalist@hsphsun2.harvard.edu> |
Subject | RE: st: RE: problem with factor variable and margins. |
Date | Wed, 17 Mar 2010 19:24:19 +0100 |
<> " Should I report it to stata's tech staff, or is that something you do, Martin?" Nope, it is your query, and I cannot possibly know as much about it as you do, including, but not limited to, the question whether there is the possibility of a bug or not. Instructions are at http://www.stata.com/support/tech-support/tech2.html HTH Martin -----Original Message----- From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Rich Steinberg Sent: Mittwoch, 17. März 2010 18:26 To: statalist@hsphsun2.harvard.edu Subject: Re: st: RE: problem with factor variable and margins. I think I have tracked down the problem. In addition to the factor variable nainc, I have a continuous variable nainc_ainc as a regressor (which is the indicator times a continuous variable). The margins command then got confused and assumed that nainc was an abbreviation for nainc_ainc, which did appear in the results and subsequent tests. When I rename nainc_ainc as n2ainc_ainc, my problem goes away -- a marginal effect for the discrete change in the factor variable nainc does indeed show up in the results. If I am right in diagnosing why the rename solves the problem, this means there is a bug in stata's margins. It is not reporting an ambiguous abbreviation, it is simply picking one. Should I report it to stata's tech staff, or is that something you do, Martin? Martin Weiss wrote: > <> > > So "1.nainc" is not in there. Is there any message after the -margins, post- > command to indicate that it got dropped? > > > HTH > Martin > > > -----Original Message----- > From: owner-statalist@hsphsun2.harvard.edu > [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Rich Steinberg > Sent: Mittwoch, 17. März 2010 01:52 > To: statalist@hsphsun2.harvard.edu > Subject: Re: st: RE: problem with factor variable and margins. > > Martin -- thanks for quick reply. > Here's what I get when I follow your directions. Note that it makes no > mention of the "n" indicators. > e(b)[1,24] > > > 0b. 1. > linc2005 ainc2005 welinc2005 trinc2005 > pdv_tot~2005 home2005 owlth2005 zlinc zlinc > y1 .00361868 .00423484 -.02269471 -.00096762 > .00184778 .0001901 .00027922 0 -316.88901 > > 0b. 1. 0b. 1. > 0b. 1. 0b. 1. 0b. > zainc zainc zwelinc zwelinc > ztrinc ztrinc zinh zinh zhome > y1 0 -390.58658 0 22.671088 > 0 68.796578 0 -93.26363 0 > > 1. 0b. > 1. > zhome zowlth zowlth nainc_ainc > nhome_home nowlth_owlth > y1 -411.67932 0 -870.19597 .00711539 > .00115194 .00159752 > > . > > > Martin Weiss wrote: > >> <> >> >> What does >> >> ************* >> mat l e(b) >> ************* >> >> after the -margins,post- line give you? >> >> >> HTH >> Martin >> >> >> -----Original Message----- >> From: owner-statalist@hsphsun2.harvard.edu >> [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Rich Steinberg >> Sent: Dienstag, 16. März 2010 23:50 >> To: statalist@hsphsun2.harvard.edu >> Subject: st: problem with factor variable and margins. >> >> Running a tobit, then getting average marginal effects (dy/dx) using the >> margins command. I ask it to compute the margins for some continuous >> variables and 2 kinds of factor variables. Results are only reported >> for one of the two kinds of factor variables even though raw tobit >> coefficients are computed for both kinds and no error messages appear. >> I "post" the margins results to do some testing, and all is well for >> testing continuous variables and one kind of factor variable (using >> "test"), but I get an error message for any tests involving the other >> kind of factor variable indicating it is "not found". Is it me or >> stata, I say hesitantly, realizing that every time in the past that I >> thought I found a bug it turned out to be a stupid error on my part. >> >> Specifically, we have all sorts of income and wealth covariates and some >> indicators created from them: variables beginning with 'z' are >> indicators that =1 if the corresponding income source =0, 0 otherwise. >> I have no problem with the z variables. Variables beginning with n are >> indicators that = 1 if the corresponding income source takes a negative >> value, 0 otherwise. Here are the relevant command lines, with a few >> comments inserted: >> >> tobit Atotgiv $income i.zlinc i.zainc i.zwelinc i.ztrinc i.zinh i.zhome >> i.zowlth i.nainc nainc_ainc i.nhome nhome_home i.nowlth nowlth_owlth >> $control, ll(24.99) vce(cluster fid1968) >> *runs fine, produces coefficients for the three 'n' factor variables: >> i.nainc, i.nhome, and i.nowlth >> margins, predict(ystar(0,.)) dydx( linc2005 ainc2005 welinc2005 >> trinc2005 pdv_totinh2005 home2005 owlth2005 zlinc zainc nainc nainc_ainc >> zwelinc ztrinc zinh zhome nhome nhome_home zowlth nowlth nowlth_owlth) >> > post > >> *produces the following translation and table, where the 'n' factor >> variables disappear >> Average marginal effects Number of obs >> = 4790 >> Model VCE : Robust >> >> Expression : E(Atotgiv*|Atotgiv>0), predict(ystar(0,.)) >> dy/dx w.r.t. : linc2005 ainc2005 welinc2005 trinc2005 pdv_totinh2005 >> home2005 owlth2005 1.zlinc >> 1.zainc 1.zwelinc 1.ztrinc 1.zinh 1.zhome 1.zowlth >> nainc_ainc nhome_home nowlth_owlth >> >> >> Delta-method >> dy/dx Std. Err. z P>z [95% Conf. Interval] >> >> linc2005 .0036341 .0018126 2.00 0.045 .0000814 .0071867 >> ainc2005 .0042529 .0054043 0.79 0.431 -.0063394 .0148451 >> welinc2005 -.0227912 .0274355 -0.83 0.406 -.0765638 >> .0309814 >> trinc2005 -.0009717 .0044668 -0.22 0.828 -.0097265 >> > .007783 > >> pdv_tot~2005 .0018556 .0016007 1.16 0.246 -.0012817 >> .004993 >> home2005 .0001909 .0004784 0.40 0.690 -.0007468 .0011286 >> owlth2005 .0002804 .0001727 1.62 0.104 -.0000581 >> > .0006189 > >> 1.zlinc -318.2919 220.6601 -1.44 0.149 -750.7777 114.194 >> 1.zainc -392.2745 79.73669 -4.92 0.000 -548.5555 -235.9935 >> 1.zwelinc 22.76784 236.5038 0.10 0.923 -440.7712 >> > 486.3069 > >> 1.ztrinc 69.08922 71.11036 0.97 0.331 -70.28453 208.463 >> 1.zinh -93.65764 124.6813 -0.75 0.453 -338.0284 150.7131 >> 1.zhome -413.516 85.3139 -4.85 0.000 -580.7282 -246.3038 >> 1.zowlth -874.5375 135.0824 -6.47 0.000 -1139.294 >> > -609.7809 > >> nainc_ainc .0071456 .0273387 0.26 0.794 -.0464373 >> > .0607286 > >> nhome_home .0011568 .0057747 0.20 0.841 -.0101613 >> > .012475 > >> nowlth_owlth .0016043 .0015222 1.05 0.292 -.0013791 >> .0045878 >> >> Note: dy/dx for factor levels is the discrete change from the base >> level. >> >> *And then I do many tests, including the following that work and don't >> > work: > >> test (ainc)(1.zainc) >> >> ( 1) ainc2005 = 0 >> ( 2) 1.zainc = 0 >> >> chi2( 2) = 28.57 >> Prob > chi2 = 0.0000 >> test (1.nainc) (nainc_ainc) >> 1.nainc not found >> r(111); >> >> >> >> >> > * > * For searches and help try: > * http://www.stata.com/help.cgi?search > * http://www.stata.com/support/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/statalist/faq > * http://www.ats.ucla.edu/stat/stata/ > -- Rich Steinberg Department of Economics, 516 Cavanaugh Hall IUPUI Indianapolis, IN 46202-5140 317-278-7221 "Unanswered email since 1955" * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/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/statalist/faq * http://www.ats.ucla.edu/stat/stata/