Date:         Wed, 3 May 2000 15:18:22 -0700
Reply-To:     glaser@pacific-science.com
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Dale Glaser <glaser@PACIFIC-SCIENCE.COM>
Subject:      Re: Alpha for dichotomous variables
Comments: To: Jocelyn Bisson <jbisson@VIF.CA>
In-Reply-To:  <000901bfb54a$836e1420$933d0dd8@jbisson.dsuper.net>
Content-Type: multipart/alternative;

I've computed by hand the Kuder-Richardson 20 (KR-20) for a cluster of
items, and when compared to the coefficient alpha from the SPSS output,
they're virtually the same, so I haven't had any problems with using the
reliability procedure in SPSS for dichotomous items......however, other's
may have different opinions!!.........dale glaser
    -----Original Message-----
    From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU]On Behalf Of
Jocelyn Bisson
    Sent: Wednesday, May 03, 2000 2:57 PM
    To: SPSSX-L@LISTSERV.UGA.EDU
    Subject: Alpha for dichotomous variables


    Hello list.

    I'd like to assess the internal consistency of scale made of dichotomous
variables (yes / no),
    ans I was wondering if the alpha statistic as computed in "Reliability"
was appropriate for the purpose.

    Is it outright inappropriate, somewhat appropriate (why), or totally
appropriate ?

    Would there be a better test for the purpose ?

    And if the scale contains dichotomous and Likert scales ... ?


    Thank you very much for your comments.

    J. Bisson
    Research associate
    U. of Montreal


[text/html]