Principal Components

[Manning, Linda (UMC-Student)]

Dear SPSS List serve:

I was wondering if you could help me with a question concerning SPSS and Principal Components Analysis.

When you run a Principal Components Analysis, it allows you the option to save the factors as variables. However, it automatically saves the factor with all the items---before I had a chance to go through each item, and throw away those items that had loadings below a critical value.

How can I get SPSS to save a factor as a variable that includes only items above a certain loading score?

Thank you, Linda

[statisticsdoc]

Stephen Brand
www.statisticsdoc.com

Linda,

This is not possible because factor scores are computed using all of the
information that is available in a set of items (this is not a limitation of
SPSS; it is a feature of the factor score estimation procedures).  If you
have not already done so, you may wish to delete items that perform poorly,
re-run the analysis, and save the factor scores from this run.

HTH,

Stephen Brand

For personalized and professional consultation in statistics and research
design, visit
www.statisticsdoc.com


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]]On Behalf Of
Manning, Linda (UMC-Student)
Sent: Wednesday, January 17, 2007 12:26 PM
To: [hidden email]
Subject: Principal Components


Dear SPSS List serve:

I was wondering if you could help me with a question concerning SPSS and
Principal Components Analysis.

When you run a Principal Components Analysis, it allows you the option to
save the factors as variables. However, it automatically saves the factor
with all the items---before I had a chance to go through each item, and
throw away those items that had loadings below a critical value.

How can I get SPSS to save a factor as a variable that includes only items
above a certain loading score?

Thank you, Linda

[Hector Maletta]

        Dear Linda,

        It is nice, though a bit quaint, that you address the SPSS list
server as "dear". We all love it too, in a way. In fact, however, you are
addressing us, the subscribers, and we are not the server but real people
subscribed to the list, to whom the server (a machine) blindly resends your
message.

        Regarding your question:

        When you run factor analysis in SPSS you may choose how many factors
to extract. Starting from K observed variables the FACTOR procedure can
extract between 1 and K unobserved factors or components). Automatically,
the SAVE subcommand saves factor scores for all extracted factors. By
default SPSS extracts all factors with eigenvalues above 1.00 but you may
change that through syntax.

        To achieve what you want you may do one of the following:

        1. Extract as many factors as you like, possibly the whole set of K
factors, save all the scores, then delete all unneeded factor scores from
the working file.

        2. Run the procedure twice. First extract many factors as in the
precedent option; then identify visually the first M factors that you will
use because they have significant loadings; and in a second run extract and
save only the first M factors.

        Hector


        -----Mensaje original-----
De: SPSSX(r) Discussion [mailto:[hidden email]] En nombre de
Manning, Linda (UMC-Student)
Enviado el: 17 January 2007 14:26
Para: [hidden email]
Asunto: Principal Components

        Dear SPSS List serve:

        I was wondering if you could help me with a question concerning SPSS
and Principal Components Analysis.

        When you run a Principal Components Analysis, it allows you the
option to save the factors as variables. However, it automatically saves the
factor with all the items---before I had a chance to go through each item,
and throw away those items that had loadings below a critical value.

        How can I get SPSS to save a factor as a variable that includes only
items above a certain loading score?

        Thank you, Linda

[Hector Maletta]

        In addition to my previous message:

        Besides choosing which factor scores to retain, you seem to be
asking also which items to retain. This of course requires running the
procedure several times. Every run will use all the items you indicate are
to be used. At each new run you would select the best items, discarding
those that perform poorly, until you reach a final set of items you're
satisfied with, obtain the PCA solution, and save the corresponding factor
scores.

        Hector

        -----Mensaje original-----
De: SPSSX(r) Discussion [mailto:[hidden email]] En nombre de
Manning, Linda (UMC-Student)
Enviado el: 17 January 2007 14:26
Para: [hidden email]
Asunto: Principal Components

        Dear SPSS List serve:

        I was wondering if you could help me with a question concerning SPSS
and Principal Components Analysis.

        When you run a Principal Components Analysis, it allows you the
option to save the factors as variables. However, it automatically saves the
factor with all the items---before I had a chance to go through each item,
and throw away those items that had loadings below a critical value.

        How can I get SPSS to save a factor as a variable that includes only
items above a certain loading score?

        Thank you, Linda

[David Hitchin]

Quoting "Manning, Linda (UMC-Student)" <[hidden email]>:

> How can I get SPSS to save a factor as a variable that includes only
> items above a certain loading score?

Another way to do this is to use the Reliability procedure to find out
which variables make a scale with large variance. You do this by
throwing away the less useful variable(s) repeatedly until the only ones
that remain form a good scale.

Then remove the variables which contribute to the first scale, and
repeat the procedure on the rest.

The results of this apparently ad hoc method often (but not always) are
quite similar to those coming out of a component analysis.

David Hitchin