🚀 go-pugleaf

Dear All,
 This is to ask for  your professional advice on obtaining factor
scores after multiple imputation of missing values.
When I  tried  to get factor scores in a  factor analysis after
Multiple Imputation, I encounter a problem.  SAS manual (SAS, 2008,
PP. 937) provides detailed explanations on how to get factor scores
with the following SAS statements.

proc calis data=raw outstat=ostat;
lineqs
v1 = a1 f1 + e1,
v2 = a2 f1 + e2,
v3 = a3 f1 + e3;
std
f1 = 1.,
e1-e3 = evar1-evar3;
run;


proc score data=raw score=ostat out=scores;
run;


but the OUTSTAT in my factor analysis included five sets of relevant
variables because of Multiple imputation. Consequently, five  factor
scores will be produced, which are not so practical if used as index
my further  analysis.
When I searched the network, I also found another  way to do the
factor analysis after multiple imputation with Proc Calis with the
following statements:
proc calis data=esteem method=ml cov privec pcoves outstat=ostat;
var = s1-s4;
lineqs
  s1 = 1.0 f1 + e1,
  s2 = lam2 f1 + e2,
  s3 = lam3 f1 + e3,
  s4 = lam4 f1 + e4;
std
  f1 = ph1,
  e1-e4 = th1 th2 th3 th4;
cov
  e2 e4 = th5;
by _imputation_;
ods output Estimates=a covmat=b;
run;


proc mianalyze parms=a covb=b ;
modeleffects lam2 lam3 lam4 ph1 th1 th2 th3 th4 th5;
run;


The predicted parameters generated by the above PROC MIANALYZE can be
used to produce FACTOR SCORES.  But I don’t think they are the best
solution, especially, when I have to do multiple multilevel data
factor analysis.
So, I am wondering if  you could give me a more practical solution so
that factor scores can be worked out with the PROC SCORE statements?


Thanks and best regards!


LHQ