Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | Nick Cox <n.j.cox@durham.ac.uk> |
To | "'statalist@hsphsun2.harvard.edu'" <statalist@hsphsun2.harvard.edu> |
Subject | st: RE: RE: local macro and xtmixed in stata 12 giving diff results |
Date | Fri, 17 Feb 2012 17:21:24 +0000 |
This is clearly and prominently documented. When you write a program using local macros, you need not worry that some other program has been written using local macros with the same names. Local macros are just that: local to your program. [U, p.195] Local macros exist solely within the program or do-file in which they are defined. [P, p.192] Nick n.j.cox@durham.ac.uk -----Original Message----- From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Nick Cox Sent: 17 February 2012 17:11 To: 'statalist@hsphsun2.harvard.edu' Subject: st: RE: local macro and xtmixed in stata 12 giving diff results I don't think the results are "slightly different". In essence, the -xtmixed- command that you invoke with an embedded reference to `variables' interprets that reference as referring to a non-existent macro. Otherwise put, Stata cannot see the macro from where it is implementing the -xtmixed- command. My guess is that there is something you are not telling us, namely that the -local variables <stuff>- is defined in one place and the -xtmixed- command is run from another. In fact, this is not really a guess. The tell-tale line is "end of do-file" meaning that this is not a "complete log" at all, as you ran a do-file to do part of this. A key point about local macros is that they _are_ local, so that local macros defined in one locale are not visible from another. Here a "locale" is defined (by me, not StataCorp) as being one of (1) an interactive session (2) a program (3) a do-file (4) the do-file editor contents. The remedy is easy: if you refer to a local macro within a do-file or do-file editor window, you _must_ define that local macro earlier in the same locale. Nick n.j.cox@durham.ac.uk Clara Barata Any idea why these two models below would give me slightly different results? *Model1 xtmixed ptc_10_ interv ptc_09_ D2-D5 agrup saseAp09 saseBp09 saseCp09 compy_09 web_09 preesc secundario totalunos_09 /// docenttot_09 docentPercfem_09 docidade_09 naodoctot_09 antigu_09 docdoutmest_09 /// AMU10 APU10 || cescola:, cov(un) pweight(pscorewgt) mle var *Model2 local variables D2-D5 agrup saseAp09 saseBp09 saseCp09 compy_09 web_09 preesc secundario totalunos_09 /// docenttot_09 docentPercfem_09 docidade_09 naodoctot_09 antigu_09 docdoutmest_09 /// AMU10 APU10 xtmixed ptc_10_ interv ptc_09_ `variables' || cescola:, cov(un) pweight(pscorewgt) mle var * Complete log . local variables D2-D5 agrup saseAp09 saseBp09 saseCp09 compy_09 web_09 preesc secundario totalunos_09 /// > docenttot_09 docentPercfem_09 docidade_09 naodoctot_09 antigu_09 docdoutmest_09 /// > AMU10 APU10 . xtmixed ptc_10_ interv ptc_09_ `variables' || cescola:, cov(un) pweight(pscorewgt) mle var Note: single-variable random-effects specification; covariance structure set to identity (1625 missing values generated) Obtaining starting values by EM: Performing gradient-based optimization: Iteration 0: log pseudolikelihood = -7629.1893 Iteration 1: log pseudolikelihood = -7628.8716 Iteration 2: log pseudolikelihood = -7628.8715 Computing standard errors: Mixed-effects regression Number of obs = 1359 Group variable: cescola Number of groups = 361 Obs per group: min = 1 avg = 3.8 max = 8 Wald chi2(2) = 329.84 Log pseudolikelihood = -7628.8715 Prob > chi2 = 0.0000 (Std. Err. adjusted for 361 clusters in cescola) ------------------------------------------------------------------------------ | Robust ptc_10_ | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- interv | 4.425498 .7011216 6.31 0.000 3.051325 5.799671 ptc_09_ | .4548144 .0289727 15.70 0.000 .398029 .5115999 _cons | 47.83422 2.600319 18.40 0.000 42.73769 52.93075 ------------------------------------------------------------------------------ ------------------------------------------------------------------------------ | Robust Random-effects Parameters | Estimate Std. Err. [95% Conf. Interval] -----------------------------+------------------------------------------------ cescola: Identity | var(_cons) | 9.664571 2.245821 6.128909 15.2399 -----------------------------+------------------------------------------------ var(Residual) | 42.38466 2.659331 37.4802 47.93088 ------------------------------------------------------------------------------ . . xtmixed ptc_10_ interv ptc_09_ D2-D5 agrup saseAp09 saseBp09 saseCp09 compy_09 web_09 preesc secundario totalunos_09 /// > docenttot_09 docentPercfem_09 docidade_09 naodoctot_09 antigu_09 docdoutmest_09 /// > AMU10 APU10 || cescola:, cov(un) pweight(pscorewgt) mle var Note: single-variable random-effects specification; covariance structure set to identity (1641 missing values generated) Obtaining starting values by EM: Performing gradient-based optimization: Iteration 0: log pseudolikelihood = -7567.187 Iteration 1: log pseudolikelihood = -7565.6155 Iteration 2: log pseudolikelihood = -7565.6135 Iteration 3: log pseudolikelihood = -7565.6135 Computing standard errors: Mixed-effects regression Number of obs = 1359 Group variable: cescola Number of groups = 361 Obs per group: min = 1 avg = 3.8 max = 8 Wald chi2(23) = 493.65 Log pseudolikelihood = -7565.6135 Prob > chi2 = 0.0000 (Std. Err. adjusted for 361 clusters in cescola) ---------------------------------------------------------------------------------- | Robust ptc_10_ | Coef. Std. Err. z P>|z| [95% Conf. Interval] -----------------+---------------------------------------------------------------- interv | 4.098699 .5971533 6.86 0.000 2.9283 5.269098 ptc_09_ | .4162623 .0267885 15.54 0.000 .3637578 .4687669 D2 | -.5525483 .8820861 -0.63 0.531 -2.281405 1.176309 D3 | -3.002324 .8082203 -3.71 0.000 -4.586407 -1.418241 D4 | -3.811223 1.05004 -3.63 0.000 -5.869264 -1.753181 D5 | -4.668435 1.141422 -4.09 0.000 -6.90558 -2.431289 agrup | 1.900642 .9238745 2.06 0.040 .0898816 3.711403 saseAp09 | -5.459656 2.802486 -1.95 0.051 -10.95243 .0331154 saseBp09 | 5.247487 4.820744 1.09 0.276 -4.200997 14.69597 saseCp09 | -5.374767 8.870566 -0.61 0.545 -22.76076 12.01122 compy_09 | -.0563566 .0343605 -1.64 0.101 -.123702 .0109889 web_09 | .0823975 .0430518 1.91 0.056 -.0019826 .1667775 preesc | 1.373344 1.237249 1.11 0.267 -1.051619 3.798308 secundario | .3618894 .6426638 0.56 0.573 -.8977086 1.621487 totalunos_09 | .0011173 .0018731 0.60 0.551 -.0025538 .0047884 docenttot_09 | -.0349247 .023305 -1.50 0.134 -.0806017 .0107524 docentPercfem_09 | -.0630209 .0412631 -1.53 0.127 -.1438951 .0178534 docidade_09 | -.3729159 .1935731 -1.93 0.054 -.7523122 .0064805 naodoctot_09 | -.0240927 .0380972 -0.63 0.527 -.0987619 .0505765 antigu_09 | .1846323 .1690689 1.09 0.275 -.1467367 .5160014 docdoutmest_09 | .093919 .0949468 0.99 0.323 -.0921734 .2800113 AMU10 | -1.187541 1.247426 -0.95 0.341 -3.632452 1.25737 APU10 | -1.663058 1.27383 -1.31 0.192 -4.15972 .833603 _cons | 73.14461 8.145537 8.98 0.000 57.17965 89.10957 ---------------------------------------------------------------------------------- ------------------------------------------------------------------------------ | Robust Random-effects Parameters | Estimate Std. Err. [95% Conf. Interval] -----------------------------+------------------------------------------------ cescola: Identity | var(_cons) | 7.148994 1.617105 4.588839 11.13748 -----------------------------+------------------------------------------------ var(Residual) | 41.30718 2.560637 36.58133 46.64355 ------------------------------------------------------------------------------ . end of do-file * * 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/