Re: st: Centering variables for interactions and using mi regress

[Richard Williams <richardwilliams.ndu@gmail.com>]

At 09:06 PM 1/16/2011, Amy S wrote:
> Richard, Thanks for the advice and passing along Paul's previous 
> advice on the subject! That is interesting - a little 
> counterintuitive, but certainly easier to implement.

Incidentally, here is the earlier discussion on how to impute values 
when you have interaction terms:

http://www.stata.com/statalist/archive/2009-02/msg00602.html

http://www.stata.com/statalist/archive/2009-02/msg00613.html

I interpret this as meaning that Allison would NOT approve of using 
commands like -mi passive-.  For example, the Stata help for -mi passive- says
---------
A passive variable is a variable that is a function of imputed 
variables or of other passive variables. For instance, if variable 
age were imputed and you created lnage from it, the lnage variable 
would be passive. The right way to create lnage is to type

. mi passive: generate lnage = ln(age)

Simply typing

. generate lnage = ln(age)

is not sufficient because that would create lnage in the m = 0 data, 
and age, being imputed, varies across m.
----------

I think Allison would disagree and say that, before imputing, do

. generate lnage = ln(age)

and then include lnage in the list of variables you impute values 
for. I don't know if everyone agrees with Allison, but if he is right 
it seems to me you would probably never use -mi passive-.

-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
HOME:   (574)289-5227
EMAIL:  Richard.A.Williams.5@ND.Edu
WWW:    http://www.nd.edu/~rwilliam

*
*   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/