Dear all,
I have several questions about the option noomit on the xi command.
This option was (finally!) introduced in Stata 9 in order to avoid
dropping the indicator for the first category of a categorical
variable when running a command with xi: . In particular it is useful
if you want to run a regression with all the categories included as
regressors and (of course) dropping the constant term.
My questions are the following:
1) Why using xi, noomit: regress i.group*contvar doesn't work?
In other words the noomit option doesn't do anything if the i.group
categorical variable is interacted with a continuous variable. It
seems as if Stata Corp. realized the usefulness of the noomit option,
but just programmed it half way. Why not allow a fully interactive
model when this option is used combined with the * interaction?
2) Why when creating variables the noomit option is overruled by the
prefix option whenever prefix() comes first?
3) Why when creating variables the prefix() option is overruled by the
noomit option whenever the later comes first?
What's so wrong about being able to generate a full set of prefixed
indicator variables? (i.e. without dropping category 1?)
I know all this things can be easily fixed manually, but I just don't
understand why such a useful option (noomit) works in such a weird
manner. Am I missing something here?
Thanks in advance,
robert
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/