Date: Fri, 6 Jun 2003 14:02:37 -0400
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From: "Dr. Jeffrey D. Leitzel" <leitzel@ADELPHIA.NET>
Organization: Behavioral Health Research Institute
Subject: Re: Factor Loading
Content-Type: text/plain; charset="us-ascii"
I believe the bad news is that you can't do what you are describing in SPSS
Base. You need to use an SEM package such as LISREL, AMOS, EQS, or M-PLUS
to do confirmatory factor analysis. AMOS is available through SPSS and
integrates well but much be purchased separately.
Jeffrey D. Leitzel, Ph.D.
Behavioral Health Research Institute
326 Adams Avenue
Scranton, PA 18503
ph - 570 650-6286
fax - 570 383-6847
ph - 570 383-6849 (home)
pager - 570 480-5753
e - director@BHRI.net
> -----Original Message-----
> From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU]
> On Behalf Of Daniel B. Swartz, Ph.D., CI, CT
> Sent: Friday, June 06, 2003 11:17 AM
> To: SPSSX-L@LISTSERV.UGA.EDU
> Subject: Factor Loading
> I am psychometrically testing 16 items in a survey (i.e.,
> factor loadings, intrainventory scale correlations, and
> Cronbach's alphas). I am not completely comfortable with how
> to do this in SPSS (I'm using Base 8.0).
> I need instruction on how to:
> 1. Obtain a comparison of factor loadings (lambda) for
> these 16 items, with test statistics (t) for each loading. I
> am trying to make sure that my Lambda Y loadings for each
> question exceed a published test statistic.
> 2. I wish to compare the test statistic for overall model
> fit, following a chi-square distribution, for alternative
> models including uncorrelated, two-factor, and one-factor.
> 3. I wish to assess the reliability of the subscales on my
> 16-item survey (3 subscales), and the relationships among
> them. I believe that scales need to be constructed by
> summing the items loading on each of the factors and dividing
> by the number of items. Correlations need to be generated
> for these subscales.
> 4. I would like to determine internal consistency of the
> subscales and of the composite survey score by calculating
> Cronbach's alpha.
> 5. I would like to test relationships between hypothesized
> antecedents and correlates related to my survey. I believe I
> need to perform a Pearson product-moment correlations (or
> point biserial correlations in the case of dichotomous
> antecedent - work setting). I believe I would need to
> estimate correlations for several aspects of the work setting
> with the survey global score (which is for burnout, by the
> way), including work hours, setting, supervision, and workload.
> 6. I would like to estimate current relations between
> burnout scores (global score from 16-item survey) and
> theoretically relevant correlates - that is, job
> satisfaction, contingency rewards, promotion, operating
> conditions, coworkers, nature of work, communication.
> If anyone can instruct me with the methods to accomplish the
> above in SPSS Base I would be very grateful.
> Daniel B. Swartz, Ph.D., CI, CT