proc factor with missing values pairwise deletion

Accepted Solution Solved
Reply
N/A
Posts: 1
Accepted Solution

proc factor with missing values pairwise deletion

I am running factor analysis on a set of hospital patient satisfaction survey data that has missing values that have been recoded from "Can't remember" and "Doesn't Apply", as they are valid answers to the survey but not used in the scale score calculation.  When I run PCA in SPSS I can specify pariwise deletion of missing values.  However the default for SAS is listwise deletion and there are no options I can find to specify pairwise deletion.  This drastically reduces the number of obs read if there is one or more missing value in each row of data.  I have only been able to find suggestions to impute the data, which is not possible for my data set as the missing values are recoded from "Can't remember" and "Doesn't Apply", which cannot be assigned as anything else and must remain missing.  Why am I unable to specify pairwise deletion of missing cases and can this be changed?

*SPSS Syntax

FACTOR

/VARIABLES painrelf, symrelf, restrelf, mancond, hos_outc, compdoc, feelleft

/missing PAIRWISE

/ANALYSIS  painrelf, symrelf, restrelf, mancond, hos_outc, compdoc, feelleft

/PRINT UNIVARIATE INITIAL CORRELATION KMO EXTRACTION ROTATION

/FORMAT BLANK(.4)

/PLOT EIGEN

/CRITERIA MINEIGEN(1) ITERATE(25)

/EXTRACTION PC

/CRITERIA ITERATE(25)

/ROTATION VARIMAX

/METHOD=CORRELATION .

*SAS syntax

proc factor data=hosp method=principal rotate=varimax;

var painrelf symrelf restrelf mancond hos_outc compdoc feelleft;

run;

Thanks


Accepted Solutions
Solution
‎02-06-2013 11:03 AM
Respected Advisor
Posts: 2,655

Re: proc factor with missing values pairwise deletion

It is going to get ugly.  I think to do this you will have to preprocess your data using PROC CORR, where you can specify pairwise deletion, and save the output as TYPE=CORR.  This dataset can then be used as input to PROC FACTOR.  See the part of the FACTOR Procedure documentation titled "Input Data Set".

Steve Denham

View solution in original post


All Replies
Solution
‎02-06-2013 11:03 AM
Respected Advisor
Posts: 2,655

Re: proc factor with missing values pairwise deletion

It is going to get ugly.  I think to do this you will have to preprocess your data using PROC CORR, where you can specify pairwise deletion, and save the output as TYPE=CORR.  This dataset can then be used as input to PROC FACTOR.  See the part of the FACTOR Procedure documentation titled "Input Data Set".

Steve Denham

🔒 This topic is solved and locked.

Need further help from the community? Please ask a new question.

Discussion stats
  • 1 reply
  • 597 views
  • 0 likes
  • 2 in conversation