Re: st: A correlation matrix after multiple imputation

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

At 01:24 PM 7/23/2010, Alan Acock wrote:
> Some journals request a correlation matrix and vector of standard 
> deviations. The American Psychological Association stresses this in 
> their journals. When doing multiple imputations, I'm unclear how to proceed.
> ...
> 4. I've seen manuscripts where people say here is the correlation 
> matrix (casewise) before the multiple imputation, but this seems to 
> be inconsistent with the entire logic of doing multiple imputation.

One rationale for presenting means/correlations/standard deviations 
is that it provides a means by which people could replicate your 
regression analyses without having the full data set. Of course, that 
isn't true anymore once you use MI. I kind of like option 4, because 
at least you are still getting descriptive measures. I suppose a 
critical issue, though, is how do the casewise statistics compare 
with the MI statistics?

My own personal intuition is that I should be happiest with MI when 
it increases my N, reduces the standard errors, and makes more of the 
effects significant. If, however, I start seeing sign flips or major 
differences in my substantive conclusions, then I feel more nervous. 
Perhaps my intuition is wrong though?


-------------------------------------------
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/