Pairwise Deletion in Linear Regression?

[Feinstein, Zachary]

I am doing some very basic linear regressions.  While in my travels I
noticed that there is a Pairwise deletion option for Linear Regression.
That does not seem possible.
 
Theoretically every IV needs a score to have a predicted score on the
DV.  What system or method is SPSS using for simple Linear Regression?
It would seem to me to always have to be Listwise deletion all the time.
 
Thank you.
 
Zachary
[hidden email]
 

[Swank, Paul R]

The regression parameters are typically determined by using the inverse
of the correlation matrix. The correlation matrix may be estimated based
on listwise or pairwise deletion. This is not to say that this is a good
idean. Such matrices may not have nice properties. One could use
multiple imputation or a program which utilizes full information maximum
likelihood.

Paul R. Swank, Ph.D. Professor
Director of Reseach
Children's Learning Institute
University of Texas Health Science Center-Houston


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Feinstein, Zachary
Sent: Wednesday, August 08, 2007 4:17 PM
To: [hidden email]
Subject: Pairwise Deletion in Linear Regression?

I am doing some very basic linear regressions.  While in my travels I
noticed that there is a Pairwise deletion option for Linear Regression.
That does not seem possible.

Theoretically every IV needs a score to have a predicted score on the
DV.  What system or method is SPSS using for simple Linear Regression?
It would seem to me to always have to be Listwise deletion all the time.

Thank you.

Zachary
[hidden email]

[Peck, Jon]

Pairwise deletion can be used in accumulating the relevant moment matrices, but it should be used cautiously, because the resulting regression estimates do not have all of the usual statistical properties.  In particular, the covariance matrix might not be positive definite.

So I would only use this option is there is very little missing data - but, then, you could probably just use listwise deletion anyway.

HTH,
Jon Peck

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Feinstein, Zachary
Sent: Wednesday, August 08, 2007 4:17 PM
To: [hidden email]
Subject: [SPSSX-L] Pairwise Deletion in Linear Regression?

I am doing some very basic linear regressions.  While in my travels I
noticed that there is a Pairwise deletion option for Linear Regression.
That does not seem possible.

Theoretically every IV needs a score to have a predicted score on the
DV.  What system or method is SPSS using for simple Linear Regression?
It would seem to me to always have to be Listwise deletion all the time.

Thank you.

Zachary
[hidden email]