It helps if you translate the Stata commands. Not everybody is fluent in those. It would even help more if you would enlight us about the function you used to fit the model. Getting the marginal effects is not that hard at all, but how depends a bit on the function you used to estimate the model.
You can try predict(your_model,type="terms",terms="the_term_you're_interested_in") For exact information, look at the respective predict function, eg if you use lme, do ?predict.lme Be aware of the fact that R normally choses the correct predict function without you having to specify it. predict() works for most model objects. Yet, depending on the model eacht predict function can have different options or different functionality. That information is in the help files of the specific function. Cheers Joris On Wed, Jun 9, 2010 at 11:28 AM, mike mick <[email protected]> wrote: > > Dear all, > > I need to use R for one estimation, and i have readily available stata > command, but i need also the R version of the same command. > the estimation in stata is as following: > 1. Compute mean values of relevant variables > > > > . sum inno lnE lnM > > > > Variable | Obs Mean Std. Dev. Min Max > > -------------+-------------------------------------------------------- > > inno | 146574 .0880374 .2833503 0 1 > > lnE | 146353 .9256239 1.732912 -4.473922 10.51298 > > lnM | 146209 4.281903 1.862192 -4.847253 13.71969 > > > > 2. Estimate model > > > > . xi: xtreg lnLP lnC lnL lnE lnM eco inno eco_inno eco_lnE eco_lnM i.year, fe > i(stno) > > i.year _Iyear_1997-1999 (naturally coded; _Iyear_1997 omitted) > > > > Fixed-effects (within) regression Number of obs = 146167 > > Group variable (i): stno Number of groups = 48855 > > > > R-sq: within = 0.9908 Obs per group: min = 1 > > between = 0.9122 avg = 3.0 > > overall = 0.9635 max = 3 > > > > F(11,97301) = 949024.29 > > corr(u_i, Xb) = 0.2166 Prob > F = 0.0000 > > > > ------------------------------------------------------------------------------ > > lnLP | Coef. Std. Err. t P>|t| [95% Conf. Interval] > > -------------+---------------------------------------------------------------- > > lnC | .0304896 .0009509 32.06 0.000 .0286258 .0323533 > > lnL | -.9835998 .0006899 -1425.74 0.000 -.984952 -.9822476 > > lnE | .0652658 .0009439 69.14 0.000 .0634158 .0671159 > > lnM | .6729931 .0012158 553.53 0.000 .67061 .6753761 > > eco | .0610348 .0177048 3.45 0.001 .0263336 .095736 > > inno | .0173824 .0058224 2.99 0.003 .0059706 .0287943 > > eco_inno | .0080325 .0110815 0.72 0.469 -.0136872 .0297522 > > eco_lnE | .0276226 .004059 6.81 0.000 .019667 .0355781 > > eco_lnM | -.0214237 .0039927 -5.37 0.000 -.0292494 -.0135981 > > _Iyear_1998 | -.0317684 .0013978 -22.73 0.000 -.034508 -.0290287 > > _Iyear_1999 | -.0647261 .0027674 -23.39 0.000 -.0701501 -.0593021 > > _cons | 1.802112 .009304 193.69 0.000 1.783876 1.820348 > > -------------+---------------------------------------------------------------- > > sigma_u | .38142386 > > sigma_e | .2173114 > > rho | .75494455 (fraction of variance due to u_i) > > ------------------------------------------------------------------------------ > > F test that all u_i=0: F(48854, 97301) = 3.30 Prob > F = 0.0000 > > > > 3. Compute marginal effect of eco at sample mean > > > > . nlcom (_b[eco]+_b[inno]*.0880374+_b[eco_lnE]*.9256239+_b[eco_lnM]*4.281903) > > > > _nl_1: > _b[eco]+_b[inno]*.0880374+_b[eco_lnE]*.9256239+_b[eco_lnM]*4.281903 > > > > ------------------------------------------------------------------------------ > > lnLP | Coef. Std. Err. t P>|t| [95% Conf. Interval] > > -------------+---------------------------------------------------------------- > > _nl_1 | -.0036011 .008167 -0.44 0.659 -.0196084 .0124061 > > ------------------------------------------------------------------------------ > > > > in fact i can find the mean of the variables ( step 1) and extimate the model > (step 2) but i couldnt find the equivalent of step 3 (compute marginal effect > of eco at sample mean). Can someone help me for this issue? > > Cheers! > > > _________________________________________________________________ > > > [[alternative HTML version deleted]] > > ______________________________________________ > [email protected] mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. > -- Joris Meys Statistical consultant Ghent University Faculty of Bioscience Engineering Department of Applied mathematics, biometrics and process control tel : +32 9 264 59 87 [email protected] ------------------------------- Disclaimer : http://helpdesk.ugent.be/e-maildisclaimer.php ______________________________________________ [email protected] mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
