Date: Wed, 15 Nov 2000 13:11:50 -0800
Reply-To: c dommeyer <vcmkt001@CSUN.EDU>
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From: c dommeyer <vcmkt001@CSUN.EDU>
Subject: Re: Reliability Analysis with Nominal Data
Content-Type: TEXT/PLAIN; charset=US-ASCII
Thanks much for your response. I, however, do not have binary
variables. Each of my items has 3 options, not 2. Therefore, I don't
think I can legitmately use the KR20 approach. Correct?
On Wed, 15 Nov 2000, Winkel, Gary wrote:
> Hi Mr. Dommeyer!
> Kuder-Richardson 20 is a special case of Cronbach's alpha for binary
> variables. So you can use the reliability module in SPSS for your analysis.
> Gary Winkel
> City University of New York
> -----Original Message-----
> From: c dommeyer [mailto:vcmkt001@CSUN.EDU]
> Sent: Wednesday, November 15, 2000 3:57 PM
> To: SPSSX-L@LISTSERV.UGA.EDU
> Subject: Reliability Analysis with Nominal Data
> I asked this questions a few days ago but got no response. Hopefully today
> someone will be inspired to respond.
> I have a ten item scale where each item allows for three choices. The
> choices are coded 1, 2, & 3, and represent nominal data. Each item has
> one "correct" answer where I award a point to the respondent. If the
> "correct" choice is not marked, no points are awarded. A respondent can
> score from 0 to 10 on this scale. From what I understand, I can't
> "legally" perform a reliability analysis on this scale (similar to a
> Cronbach's alpha). Am I right or wrong?
> Curt Dommeyer