[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: RE: ice and constructed variables
S. Strauss is refering to an issue with multiple imputation and
interaction variables (or other 'constructed variables'). Consider an
imputation model for price foreign rep78 and rep78Xforeign. You could
use -ice- and impute all four variables, but that way you would get
random draws for the variable rep78 and different random draws for
rep78Xforeign. Better would be to get random draws for rep78 and use
those to create rep78Xforeign. This can be done with -ice- using the eq
and passive options. S. Strauss can read how to do that in (Royston
Patrick Royston (2005) "Multiple Imputation of missing values: update"
The Stata Journal, 5(2), pp. 188-201.
--- Nick Cox <email@example.com> wrote:
> I guess that you shouldn't represent it in -ice-
> at all. Imputing the components is one thing. Producing
> a construct downstream of the imputation is another.
> Otherwise, at best, it is hard to disentangle what
> to blame for what.
> In any case, what is your formula here? Dying to know,
> > I'm trying to undertake multiple imputation with "ice" and have a
> > questions about how to account for the relationship between a
> > constructed variable and its multiple component parts. I'm aware
> > that I should use "passive" and, at times, "eq", but not exactly
> > sure of how. For example, I have created a variable I'll call
> > "sexuality" from the "gender" variable (0,1) and "attracted to
> > males" (0,1) and "attracted to females" (0,1). How would I
> > represent this in "ice"? I'd appreciate some help figuring this
> > out.
> > Thanks
> > S. Strauss
Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
1081 HV Amsterdam
Buitenveldertselaan 3 (Metropolitan), room Z434
+31 20 5986715
All New Yahoo! Mail – Tired of Vi@gr@! come-ons? Let our SpamGuard protect you. http://uk.docs.yahoo.com/nowyoucan.html
* For searches and help try: