Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
| From | LUCIA SUMMERS <summers_lucia@hotmail.com> |
| To | <statalist@hsphsun2.harvard.edu> |
| Subject | st: RE: How do the Wald test chi-square statistic and the pseudo r-square relate to each other in clogit? |
| Date | Thu, 2 Feb 2012 19:17:42 +0000 |
Hello, I am trying to understand some results I got (see below) and any assistance will be greatly appreciated. I have run a number of conditional logit models (with cluster-adjusted robust SEs) and usually, if you get a greater Wald chi-square value, you also get a greater pseudo r-square. But the models below are strange in that the first one has a great pseudo r-square value while the second one has a greater Wald chi-square value. The only thing that changes from one model to the next are the third and fourth variables starting from the bottom (i.e., in the second model, a_lneuclkm and m_lneuclkm have been substituted by a_lneuclmins and m_lneuclmins). Does anyone have any information about how the Wald test statistic is calculated so that I can relate this to how the log-likelihood and the pseudo r-square are calculated (and work out why they might be behaving in this way)? Thanks very much, Lucia. Model 1 . clogit choice commerc100 nostations popdens unemp10 imdincome imdhealth imdeducation imdhousing imdliving age15to29p10 seh100 eh100 mob10 a_lneuclkm m_lneuclkm sameethnic10 barrier, group(accno) vce(cluster crid) or Iteration 0: log pseudolikelihood = -3001.5126 Iteration 1: log pseudolikelihood = -2981.3624 Iteration 2: log pseudolikelihood = -2981.1934 Iteration 3: log pseudolikelihood = -2981.1934 Conditional (fixed-effects) logistic regression Number of obs = 2434915 Wald chi2(17) = 2123.56 Prob > chi2 = 0.0000Log pseudolikelihood = -2981.1934 Pseudo R2 = 0.3111 (Std. Err. adjusted for 276 clusters in crid)------------------------------------------------------------------------------ | Robust choice | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval]-------------+---------------------------------------------------------------- commerc100 | 1.085783 .0265569 3.36 0.001 1.03496 1.139101 nostations | 1.031757 .1310436 0.25 0.806 .8043889 1.323392 popdens | .9933222 .0018225 -3.65 0.000 .9897565 .9969008 unemp10 | 2.113067 1.908653 0.83 0.408 .3597953 12.40998 imdincome | 8.174588 14.58476 1.18 0.239 .2476213 269.8633 imdhealth | 1.317499 .3145333 1.15 0.248 .8251607 2.103595imdeducation | .9797458 .0111904 -1.79 0.073 .9580568 1.001926 imdhousing | .9871881 .0132498 -0.96 0.337 .9615576 1.013502 imdliving | 1! .007555 .0072077 1.05 0.293 .993527 1.021781age15to29p10 | .9085186 .2040912 -0.43 0.669 .5849496 1.411072 seh100 | 1.080342 .0875577 0.95 0.340 .9216681 1.266333 eh100 | 1.002131 .0102836 0.21 0.836 .9821768 1.022491 mob10 | 1.5589 .4010742 1.73 0.084 .9415009 2.581164 a_lneuclkm | .1852571 .0089984 -34.71 0.000 .168434 .2037605 m_lneuclkm | .1642576 .0171001 -17.35 0.000 .1339401 .2014374sameethnic10 | 1.320282 .0669419 5.48 0.000 1.195387 1.458226 barrier | .4008838 .0867893 -4.22 0.000 .2622634 .6127725------------------------------------------------------------------------------ Model 2 . clogit choice commerc100 nostations popdens unemp10 imdincome imdhealth imdeducation imdhousing imdliving age15to29p10 seh100 eh100 mob10 a_lneuclmins m_lneuclmins sameethnic10 barrier, group(accno) vce(cluster crid) or Iteration 0: log pseudolikelihood = -3055.2028 Iteration 1: log pseudolikelihood = -3035.5308 Iteration 2: log pseudolikelihood = -3035.4117 Iteration 3: log pseudolikelihood = -3035.4117 Conditional (fixed-effects) logistic regression Number of obs = 2434915 Wald chi2(17) = 3070.44 Prob > chi2 = 0.0000 Log pseudolikelihood = -3035.4117 Pseudo R2 = 0.2986 (Std. Err. adjusted for 276 clusters in crid) ------------------------------------------------------------------------------ | Robust choice | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- commerc100 | 1.080073 .0234639 3.55 0.000 1.03505 1.127055 nostations | .973521 .1170455 -0.22 0.823 .7691413 1.232209 popdens | .9921836 .0018916 -4.12 0.000 .988483 .995898 unemp10 | 2.280424 2.013854 0.93 0.351 .4039385 12.87407 imdincome | 5.966666 10.39074 1.03 0.305 .196514 181.1632 imdhealth | 1.307199 .289918 1.21 0.227 .8463645 2.018952 imdeducation | .9814027 .0110888 -1.66 0.097 .9599079 1.003379 imdhousing | .984771 .0126269 -1.20 0.231 .9603312 1.009833 imdliving | 1.005372 .0071359 0.75 0.450 .991483 1.019456 age15to29p10 | .9137366 .2075709 -0.40 0.691 .5854045 1.426218 seh100 | 1.080684 .0880311 0.95 0.341 .9212152 1.267759 eh100 | 1.003857 .0098699 0.39 0.695 .9846979 1.023389 mob10 | 1.478928 .3834865 1.51 0.131 .8896727 2.458464 a_lneuclmins | .100379 .0054367 -42.44 0.000 .0902694 .1116209 m_lneuclmins | .0900161 .0094458 -22.95 0.000 .0732825 .1105708 sameethnic10 | 1.329318 .0638578 5.93 0.000 1.20987 1.460558 barrier | .2776155 .0561903 -6.33 0.000 .1867057 .4127905 ------------------------------------------------------------------------------ * * 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/