NetNews Usenet Archive 1992 #20/NN_1992_20.iso/spool/bit/listserv/statl/1521

Text File | 1992-09-09 | 1.6 KB | 35 lines

Comments: Gated by [e-mail]
Path: sparky!uunet!paladin.american.edu!auvm!CHARLIE.USD.EDU!MGRANAAS
X-Vmsmail-To: SMTP%"[e-mail]"
Message-ID: <[e-mail]>
Newsgroups: bit.listserv.stat-l
Date:         Wed, 9 Sep 1992 10:16:54 -0500
Sender:       "STATISTICAL CONSULTING" <[e-mail]>
From:         [e-mail]
Subject:      continuous likert scale items--summary
Lines: 23

I want to thank members of the list for responding to my question of
yesterday regarding continuous scoring of likert scale items by way of
visual analog.  Since the practice of questions askers on this list seems
to be to summarize responses for the list, I thought I should take a
shot at it.  Unfortunatly I lack the expertise needed to cut and paste from
other folks email, so I will just have to type my summary.

Several people provided references and/or reports of personal experience.
For the most part there was a consensus that what I suggested could indeed,
and has been done.  However, most (all?) with opinions on the matter
said that it was simply not worth the effort.  One individual provided me
with a summary from Feldt, L. (1958) _Psychometrika_ (table attached to
reprint, not in original article) which indicated that most of the variance
of scores in a continuously scored item was contained in the scores from
a discretely scored item.  (Of course I am very quick to delete my email
once I read it, so I can't give the sender's name or the exact percentages
of variance that he quoted for me.)

Again, my thanks to all who responded.

mike

[e-mail]