hscheng and Li-Lang Yang both wrote
> I am using the Stata commands -factor- (factor analysis)
> and -score- (to
> create scores for different factors) for my study. I have
> 12 items and
> the factor analysis results show 4 different factors. I
> can generate
> scores for the 4 different factors in 2 different ways. First,
>
> . factor item1-item12
> . score f1 f2 f3 f4 // so basically, do all of these at once
>
> Second,
>
> . factor item1-item12 //say, item1-3 for F1; item4-6 for F2, etc
> . factor item1-item3
> . score f1 // This generates the score for Factor 1
> . factor item4-item6
> . score f2 // This generates the score for Factor 2
> . factor item7-item9
> . score f3 // This generates the score for Factor 3
> . factor item10-item12
> . score f4 // This generates the score for Factor 4
>
> I know the Stata menu uses the first way, but is the second
> way wrong?
> The first way and the second way give different results.
> Which one is
> more accurate?
>
These commands produce completely different results
in general, as they refer to different models. Accuracy
is not an issue.
First, note that -score- always picks up on the
last -factor- or -pca- performed. It has that much
memory, and no more.
Given 12 variables, -factor- produces a model
with up to 12 factors.
Evidently in your case the first 4 of those
seem important and those can be put in variables -f1-f4-
by a single command. However, your other
commands refer to separate factor analyses.
So for example
. factor item1-item3
carries out a factor analysis on just three
variables and any -score- command following
refers to that analysis.
Nick
n.j.cox@durham.ac.uk
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/