Date: Tue, 19 Mar 2002 09:39:46 +1100
Reply-To: Sharon Morris <firstname.lastname@example.org>
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From: Sharon Morris <email@example.com>
Subject: Re: Different types of variables
Content-Type: text/plain; charset="Windows-1252"
I had a similar problem some time back with an evaluation we did of some
chocolate products. I dealt with the problem by re-coding the "just right"
type responses into a 3 point scale. For example, if you have very sweet,
fairly sweet, just about right, not very sweet,
not at all sweet, I would re-code them as follows:
very sweet --> poor
fairly sweet --> average
just about right --> just right
not very sweet --> average
not at all sweet --> poor
(I wouldn't use these particular labels, but just to give you an idea)
I then was able to use these recoded variables in a regression analysis. It
is, of course, true that you are losing information by doing this, but in my
particular case we had always done regressions for this particular client,
and it was expected.
Just one way of using the data - I'm sure there are many more.
Senior Project Manager
DBM Consultants Pty Ltd
5-7 Guest Street
Hawthorn VIC 3122
ph: 03 9819 1555 fax: 03 9819 9333
From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU]On Behalf Of
Sent: Monday, 18 March, 2002 9:29 PM
Subject: Different types of variables
In a product test there is an evaluation scale from "extremely good" to
"extremely poor". There are also a number of scales that evaluate
certain aspects of the product like colour that go from "very good" to
However there are a number of scales where the optimum is around the
middle e.g. Very sweet, fairly sweet, just about right, not very sweet,
not at all sweet.
I want to determine which of the scales drive/ correlate with the
overall evaluation scale, but I realise that main scale is increasing
so that optimum is at the end, whereas some the others have an optimum
in the middle.
What analyses can I do?