[Rasch] Rasch analysis then EFA

Swank, Paul R Paul.R.Swank at uth.tmc.edu
Sat Nov 19 03:11:38 EST 2011


The residuals contain information not accounted for by the Rasch scale. If there is coherent information in those residuals, it could lead to exploring additional scales. The variance in the person scale is that which is attributable to the underlying construct. Given that the factors were correlated there at least two potential models of explanation. First there could be a second order factor to account for those correlations. Or you could have a general factor with a number of specific factors. If it is the common variance among the items that you are interested in then the Rasch scale should capture that even in the presence of specific factors.

Dr. Paul R. Swank,
Children's Learning Institute
Professor, Department of Pediatrics, Medical School
Adjunct Professor, School of Public Health
University of Texas Health Science Center-Houston

From: Juanito Talili [mailto:talilij at yahoo.com]
Sent: Thursday, November 17, 2011 10:46 PM
To: Swank, Paul R; Rasch Forum
Subject: Re: [Rasch] Rasch analysis then EFA

My apology for the low level questions:

1. What is the logic why the residuals of the Rasch analysis are subjected for EFA?  Why not the person or item measures?
2. The factors in the previous analysis were correlated. Should I do Rasch on the combined items?

Thank you.
J

From: "Swank, Paul R" <Paul.R.Swank at uth.tmc.edu>
To: Juanito Talili <talilij at yahoo.com>; Rasch Forum <rasch at acer.edu.au>
Sent: Thursday, November 17, 2011 10:41 PM
Subject: RE: [Rasch] Rasch analysis then EFA
I would do the Rasch on all classes in order to increase heterogeneity of responses. And I would suggest EFA on the residuals of the Rasch analysis. It all depends on whether the factors are correlated so that it is possible there is an underlying latent variable with perhaps some specific factors. If the factors in the previous analysis were uncorrelated, then I would think twice before using Rasch on the combined items.

Dr. Paul R. Swank,
Children's Learning Institute
Professor, Department of Pediatrics, Medical School
Adjunct Professor, School of Public Health
University of Texas Health Science Center-Houston

From: rasch-bounces at acer.edu.au [mailto:rasch-bounces at acer.edu.au] On Behalf Of Juanito Talili
Sent: Wednesday, November 16, 2011 11:57 PM
To: Rasch Forum
Subject: [Rasch] Rasch analysis then EFA


Opps!! I forgot to indicate the subject.
Dear All,

A self-administered "Computer Laboratory Usability Questionnaire" consists of 43 five-point ordinal scale items (1=strongly disagree to 4 strongly agree).  The instrument, moreover, consists of six factors with the following number of items: 3, 3, 4, 7, 12, and 15.  The questionnaire was developed using CTT.

My Concerns:
1) Now, I want to re-validate the instrument using Rasch Model.  My sample now is already more than 3000.  The samples are students who rated six different computer laboratories (LAB1, LAB2....LAB6).  Shall I use only sample from one LAB?

2. Can I do the Rasch analysis for the 43 items as if unidimensional?  Then, do an EFA using the resulting measures?  The EFA will be used to check if the items are really loaded to the expected factor.

Your comments are highly solicited.

Thank you.
J


-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailinglist.acer.edu.au/pipermail/rasch/attachments/20111118/fa3e1012/attachment.html 


More information about the Rasch mailing list