st: Defining the omitted reference group with mi estimate: mlogit

[Erin Bumgarner <ekb2130@columbia.edu>]

Dear Stata users,

I am having problems with the "mi estimate: mlogit" command. My model
includes a series of dummy variables (e.g. white, black, hispanic,
asian, other). When running "mi estimate: mlogit" I omit one category
(e.g. white), which I intend to use as the reference group. However,
sometimes Stata will automatically omit a second group (e.g. white and
asian) -- which I obviously don't want to happen.

When imputing my missing data I used the "mi ice" command. When
imputing categorical variables (e.g. race), I used the ".o" or ".m"
prefixes. Then, after the imputation model was finished, I used "mi
passive: gen" to create the corresponding dummy variables (white,
black, hispanic, asian, other).

I am looking for suggestions about how to prevent Stata from omitting
a second group. I'm not sure if this is a problem related to how I did
my imputation -- or if it's caused by some other problem I'm unaware
of. When I leave it up to Stata to choose the default reference group,
it does only kick out one group, but it is not the reference group I
want to exclude.

I would appreciate any help you could offer.

Best,
Erin
ekb2130@columbia.edu
...
Erin Bumgarner, MA
Graduate Research Fellow
National Center for Children & Families
Columbia University
www.policyforchildren.org
*
*   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/