Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: nlcom question


From   Stas Kolenikov <skolenik@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: nlcom question
Date   Wed, 3 Mar 2010 21:46:31 -0600

If the distributions of the dependent variable are the same for two levels
of the categorical factor, then they will be the same no matter whether you
transformed them or you did not. Hence it suffices to use -test- rather than
-nlcom-. The latter will be answering a more subtle question of whether the
means are the same (you implicitly put zeroes for all other variables, which
may or may not be appropriate); you still may have differences in variance,
skewness, kurtosis, etc. between groups even if you find the means to be the
same.

Stata 11 has new -margins- command; have you looked at it?

On Wed, Mar 3, 2010 at 9:31 PM, john metcalfe <johnzmetcalfe@gmail.com>wrote:

> Dear Statalist,
> I have a simple question I hope someone can help me with.
> I am using OLS with robust variance estimators to model a continuous,
> log-transformed DV ranging from 0 to 10 in increments of 0.01 (this is
> an immunologic test in common use in the U.S.). My goal is to
> determine whether or not there are differences in this test
> performance according to a categorical independent variable (rax, 4
> levels) with an interaction term (nt, 3 levels) and other categorical
> covariates, as below:
>
> Linear regression                              Number of obs =    2734
>                                                       F( 17,  1716) =
>  133.04
>                                                       Prob > F      =
>  0.0000
>                                                       R-squared     =
>  0.4007
>                                                       Root MSE      =
>  1.6254
>
>
> ------------------------------------------------------------------------------
>             |               Robust
>       ln_ag |      Coef.   Std. Err.      t    P>|t|     [95% Conf.
> Interval]
>
> -------------+----------------------------------------------------------------
>     _Irax_1 |  -.3175203   .1598442    -1.99   0.047    -.6310302
> -.0040104
>     _Irax_2 |  -.5266611     .22524    -2.34   0.019     -.968435
> -.0848873
>     _Irax_3 |  -.0842108   .1791368    -0.47   0.638    -.4355602
>  .2671386
>      _Int_1 |   3.201428   .1158472    27.63   0.000     2.974212
>  3.428645
>      _Int_2 |   2.228758   .1441987    15.46   0.000     1.945934
>  2.511582
> _IraxXnt~1_1 |     .05004   .2237556     0.22   0.823    -.3888224
>  .4889025
> _IraxXnt~1_2 |   1.110956   .4651867     2.39   0.017     .1985636
>  2.023349
> _IraxXnt~2_1 |   1.094455   .2752091     3.98   0.000     .5546741
>  1.634236
> _IraxXnt~2_2 |   1.225901   .4853438     2.53   0.012      .273973
>  2.177829
> _IraxXnt~3_1 |   .5545474   .2237545     2.48   0.013     .1156871
>  .9934077
> _IraxXnt~3_2 |   1.250381   .3833228     3.26   0.001     .4985518
> 2.00221
>    age_cntr |   .0047114   .0028156     1.67   0.094    -.0008109
>  .0102338
>      female |  -.1865305   .0811696    -2.30   0.022    -.3457323
> -.0273287
>        jka1 |  -.3056762   .0994589    -3.07   0.002    -.5007496
> -.1106027
>        jka2 |  -.3657116   .1416161    -2.58   0.010      -.64347
> -.0879533
>       prevt |   .3526168   .1933732     1.82   0.068    -.0266552
>  .7318889
>          dm |   .3199483   .1509238     2.12   0.034     .0239343
>  .6159623
>       _cons |   -3.02986   .1154191   -26.25   0.000    -3.256237
> -2.803483
>
> ------------------------------------------------------------------------------
>
> To estimate the difference in the backtransformed DV between rax3 and
> rax0, I am using:
>
> scalar rmse = e(rmse)
> nlcom exp(_b[_cons]+_b[_Irax_3] + _b[_Int_2] + _b[_IraxXnt_3_2]
> +rmse^2/2)-exp(_b[_cons] + _b[_Int_2] + rmse^2/2), but I think I need
> to also add in the coefficients for the other predictors, multiplied
> by the average value of the covariate in both sides of the nlcom
> statement. Is there an easy way to go about doing this?
> Thanks in advance,
> John
> *
> *   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/
>



-- 
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: I use this email account for mailing lists only.

*
*   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/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index