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.

# Re: st: Specify omitted group in regression

 From Phil Clayton To statalist@hsphsun2.harvard.edu Subject Re: st: Specify omitted group in regression Date Wed, 21 Nov 2012 10:01:55 +1100

```Instead of:
... i.agencytype bps i.agencytype#i.bps ...

Try:
... i.agencytype i.bps i.agencytype#i.bps ...

(so that bps is always modelled as a factor variable; in your original syntax it's modelled once as a continuous variable and once as a factor variable)

Slightly more efficient is:
... i.agencytype##i.bps ...

Phil

On 21/11/2012, at 9:52 AM, Simone Peart Boyce <simonepeart@yahoo.com> wrote:

> Hi,
>
> I am running a regression with interacted factors and would like to specify the factors to use as the base.  I do the following, but it seems only to work when the factor is not interacted.
>
> fvset base 1 agencytype
> xtgee breast i.agencytype bps i.agencytype#i.bps, family(binomial) link(logit) corr(exc) vce(robust)
>
> The output as below omits agencytype=6 when interacted with BPS rather than agencytype=1.  How do I get Stata to omit agencytype 1 instead?
>
> Simone
>
>
> note: 2.agencytype#1.BPS identifies no observations in the sample
> note: 6.agencytype#1.BPS omitted because of collinearity
> Iteration 1: tolerance = .02508496
> Iteration 2: tolerance = .00012234
> Iteration 3: tolerance = 4.649e-07
> GEE population-averaged model                   Number of obs      =     16426
> Group variable:                 patient_id      Number of groups   =     14253
> Link:                                logit      Obs per group: min =         1
> Family:                           binomial                     avg =       1.2
> Correlation:                  exchangeable                     max =         6
>                                                 Wald chi2(10)      =    220.51
> Scale parameter:                         1      Prob > chi2        =    0.0000
>                                (Std. Err. adjusted for clustering on patient_id)
> --------------------------------------------------------------------------------
>                |               Robust
> breast_compl~r |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
> ---------------+----------------------------------------------------------------
>     agencytype |
>             1  |          0  (base)
>             2  |  -.3476051   .4017287    -0.87   0.387    -1.134979    .4397687
>             3  |   .6622408   .1231947     5.38   0.000     .4207837    .9036979
>             4  |    .303693   .3774096     0.80   0.421    -.4360163    1.043402
>             5  |   .4469481   .1493272     2.99   0.003     .1542722    .7396241
>             6  |  -.2703296   .1281538    -2.11   0.035    -.5215065   -.0191527
>                |
>            BPS |   .2035705   .1517018     1.34   0.180    -.0937596    .5009006
>                |
> agencytype#BPS |
>           1 1  |   .6664529   .1794499     3.71   0.000     .3147376    1.018168
>           2 1  |          0  (empty)
>           3 1  |     .77403   .2479991     3.12   0.002     .2879608    1.260099
>           4 1  |   .8264168   .5002503     1.65   0.099    -.1540558    1.806889
>           5 1  |   .1369255   .2521904     0.54   0.587    -.3573586    .6312096
>           6 1  |          0  (omitted)
>                |
>          _cons |   2.319875   .0603756    38.42   0.000     2.201541    2.438209
>
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/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/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/
```