Kevin,
When you extract factors from one dataset, you obtain a matrix with
the component score coefficient matrix. These coefficients equal the factor
loadings divided by the respective eigenvalue, and are the coefficients
giving the (standardized) factor score of a subject as a function of his/her
(standardized) scores in observed variables: Fki= (sum of)BkjZij, where Zij
is the Z-score of observed variable j for subject i, Bkj is the coefficient
of variable j to compute the score of factor k, and Fki is the score of
factor or component k for the same subject.
Now, you should use those coefficients to compute the factor scores
in the second data set, via a series of COMPUTE statements:
COMPUTE FACTOR_1=beta11 * Z1 + beta12 * z2 + ........
COMPUTE FACTOR_2=beta21 * Z1 + beta22 * z2 + ........
(substitute the numerical value of the coefficients where I wrote
the words beta11, beta12, etc).
You should extract as many factors or components as you need
(minimum one, maximum as many as observed variables).
Hector
-----Original Message-----
From: SPSSX(r) Discussion [mailto:
[hidden email]] On Behalf Of
Kevin Manning
Sent: 10 October 2007 14:18
To:
[hidden email]
Subject: Determine factor scores for a sample (with factors derived from
another sample)?
Hi all,
This is a repeat question from several months ago (The emails I
saved were deleted!).
Can someone tell me how I go about deriving factor scores for a
sample of participants (say Group A) based on a factor model derived from an
entirely different group of participants (Group B)?
So I have a factor structure for Group B but I want to determine
factor scores for group A (based on the factor structure of Group B).
I don't know if I am explaining this very well. Any help would be
most appreciated.
Thanks,
Kevin
Locked
