Date: Tue, 9 Dec 2008 02:38:06 -0200
Reply-To: Hector Maletta <hmaletta@fibertel.com.ar>
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From: Hector Maletta <hmaletta@fibertel.com.ar>
Subject: Re: Factor Analysis Using Matrix Input
In-Reply-To: <op.ulvi81a7ack3zx@manager-laptop>
Content-Type: text/plain; charset="us-ascii"
Of course it is run on the matrix, not on the new data. SPSS ignores the raw
data and uses the matrix as the sole input.
What you mean by applying the old solution to the new data means calculating
factor scores for each subject in the new dataset, using the coefficients
obtained in the previous data set. For this you do not need the correlation
coefficients matrix from the old factor analysis: what you want is the
factor score coefficient matrix, i.e. the coefficients defining factor
scores as a function of observed (standardized) variables. This matrix of
component score coefficients is a normal outcome of the SPSS FACTOR
procedure. The general form is F(ij)=b1*Zi1+b2*Zi2+...+bk*Zik to obtain the
score of the ith case for the jth factor as a function of the standardized
observed variables Z1 to Zk for the same ith case. Coefficients b1 to bz
obtained from the previous factor analysis. Standardized observed variables
are Z=(X-mean)/STDDEV.
Hector
-----Original Message-----
From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU] On Behalf Of
Keith Roberts
Sent: 09 December 2008 00:45
To: SPSSX-L@LISTSERV.UGA.EDU
Subject: Factor Analysis Using Matrix Input
Hi List,
I'm trying to do a factor analysis using a matrix input from an external
Correlation matrix saved from an earlier Factor Analysis -- So I'm trying
to repeat the factor analysis on "new" data using the same " solution" as
was derived from the initial Factor Analysis.
I can get the Factor Analysis to run using the Correlation matrix as the
input, but the outcome in terms of Communalities, EigenValues, Rotated
factor loadings etc stay resolutely the same regardless of what happens
with the " new" data. It is almost as if the Factor analysis is being run
on the matrix input rather than the new data.
Am I missing something or have I misunderstood the concept completely ??
Eventually the aim is the create factor scores and clustering on the new
data based on input from the "original" solution
Any Help or guidance would be much appreciated -- syntax used is below
DATASET ACTIVATE DataSet1.
FACTOR
/MATRIX IN(COR = 'C:\factor_input_1.sav')
/MISSING LISTWISE
/ANALYSIS redat1 redat10 redat11 redat12 redat13 redat14 redat15 redat16
redat17 redat18 redat19
redat2 redat20 redat21 redat22 redat23 redat24 redat25 redat26 redat27
redat28 redat29 redat3
redat31 redat32 redat33 redat34 redat35 redat36 redat37 redat38
redat39 redat4 redat40
redat41 redat42 redat43 redat44 redat45 redat46 redat47 redat48
redat49 redat5 redat50 redat51
redat52 redat53 redat54 redat55 redat56 redat57 redat58 redat59 redat6
redat60 redat61 redat62
redat63 redat64 redat65 redat66 redat67 redat68 redat69 redat7 redat70
redat71 redat72 redat73
redat74 redat75 redat76 redat77 redat78 redat79 redat8 redat80 redat81
redat82 redat83 redat84
redat85 redat86 redat87 redat88 redat89 redat9 redat90 redat91 redat92
redat93 redat94 redat95
redat96 redat97 redat98
/PRINT INITIAL EXTRACTION ROTATION
/CRITERIA FACTORS(3) ITERATE(100)
/EXTRACTION PC
/CRITERIA ITERATE(25)
/ROTATION VARIMAX
/METHOD=CORRELATION.
Thanks in advancefor your help
--
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/
=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@LISTSERV.UGA.EDU (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@LISTSERV.UGA.EDU (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
