Date:         Fri, 30 Sep 2005 16:38:53 -0500
Reply-To:     "Swank, Paul R" <Paul.R.Swank@uth.tmc.edu>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         "Swank, Paul R" <Paul.R.Swank@uth.tmc.edu>
Subject:      Re: Stability estimation: test-retest alpha reliability
              coefficient
Comments: To: Josh Johnson <joshjohnson88990@yahoo.co.uk>
Content-Type: text/plain; charset="iso-8859-1"

I believe you will find that doing an alpha on a test and retest will give you the reliability of the two tests combined. I think what you want is the simple correlation between the test and retest.


Paul R. Swank, Ph.D.
Professor, Developmental Pediatrics
Director of Research, Center for Improving the Readiness of Children for Learning and Education (C.I.R.C.L.E.)
Medical School
UT Health Science Center at Houston

-----Original Message-----
From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU] On Behalf Of Josh Johnson
Sent: Friday, September 30, 2005 2:38 PM
To: SPSSX-L@LISTSERV.UGA.EDU
Subject: Stability estimation: test-retest alpha reliability coefficient

Hi all,



A question regarding test-retest reliability/stability calculation-



For the simple case of, for example, n respondents measured twice each (say, a month apart between administrations) on only one test/question. The calculation of the coefficient alpha seems to me straight forward in SPSS-



àSelect the two variables (the test and retest) to the variable box, àselect alpha, two way mixed (consistency) as the model.

(This is assuming the n respondents are a random sample of the larger population but the one question/test is the only item we are interested in and not a sample of a larger group of questions)



However how does one calculate the coefficient alpha when more then one measurement/question is asked for each respondent? For example a case where 20 questions are asked twice (test-retest)- is there an overall alpha I can calculate or am I confined to calculating separate alpha's for each question? Assuming again that the respondents are a random sample and the 20 questions are fixed (not a sample of a larger group of questions I am interested in asking)



Any advice, references will be greatly appreciated.



Thanks in advance,

Josh



---------------------------------
Yahoo! Messenger  NEW - crystal clear PC to PC calling worldwide with voicemail