Dear Paul: this is an interesting issue. For the specific case you outline, the X1Y interaction should be included; a simple strategy would be to simply stratify on X1 and impute Y and X2 separately in the two strata. That strategy only applies in limited situations though. Rod
On Tue, 8 Jun 2004, Paul von Hippel wrote: > Suppose I intend to regress continuous Y on binary X1, continuous X2 and > the interaction X1X2. > > People seem to agree that Y should be imputed conditionally on X1, X2 and > X1X2. > > Is it also the case, though, that X2 should be imputed conditionally on X1, > Y and X1Y? In other words, since the slope of Y on X2 depends on X1, is the > same true for the slope of X2 on Y? > > For that matter, should X1 be imputed conditionally on X2, Y and X2Y? > > I'd be grateful for your thoughts on this. In a more complex settings, this > perspective leads to quite a lot of interactions being included in the > imputation model..... > > Best wishes, > Paul von Hippel > > Paul von Hippel > Department of Sociology / Initiative in Population Research > Ohio State University > 300 Bricker Hall > 190 N. Oval Mall > Columbus OH 43210 > > > _______________________________________________ > Impute mailing list > [email protected] > http://lists.utsouthwestern.edu/mailman/listinfo/impute > ___________________________________________________________________________________ Roderick Little Richard D. Remington Collegiate Professor (734) 936-1003 Department of Biostatistics Fax: (734) 763-2215 U-M School of Public Health M4045 SPH II [email protected] 1420 Washington Hgts http://www.sph.umich.edu/~rlittle/ Ann Arbor, MI 48109-2029
