Date:         Tue, 11 Apr 2006 11:30:52 -0500
Reply-To:     jeremiah@spenceconsulting.net
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Jeremiah Spence <jeremiah@spenceconsulting.net>
Organization: Spence Consulting
Subject:      Re: Dealing with measurement issues and regression analysis
Comments: To: "Beadle, ViAnn" <viann@spss.com>
In-Reply-To:  <AC7C0EC28F84ED43AA2F943F1E62EB6D01036CE7@hqemail2.spss.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed

are you referring to log-linear analysis? or are there other techniques
that more appropriate for handling  this sort of data?

-Jeremiah

Beadle, ViAnn wrote:
> There have been a number of techniques developed for handling categorical and ordinal data in the past 20 years that have been added to stat packages that could be used but aren't. Maybe the practitioners who continue to use regression inappropriately weren't introduced to them in school or are a generation behind.
>
>
> ________________________________
>
> From: SPSSX(r) Discussion on behalf of Sibusiso Moyo
> Sent: Tue 4/11/2006 9:40 AM
> To: SPSSX-L@LISTSERV.UGA.EDU
> Subject: Re: Dealing with measurement issues and regression analysis
>
>
>
> This questions is directed mostly at King Douglas and the "SPSS Big 6" (Hector, Ray, Kirill, Marta, Richard and Art), but I am sure the whole list is bound to benefit.
>
> I am very uncomfortable with what I see everyday in the "real" world, people/companies applying regression analysis to ordinal data as if it was ratio data. I always get the justfication that everyone is doing it, so why not! What do SPSS experts have to say about this? I know there are tons of companies out there that are built on analysing ordinal data as if it were continuos (taking means, etc), and they have been successful in sellig their products. I am sure there are tons of cases when things do not work, when people do not get the signs they are looking for etc, and then people start trying all sorts of manipulations (until they get what they are looking for ~ "the right signs")!
>
> I am from the school of thought that emphasizes testing of assumptions, but what does one gain really by trying to make sure the data conforms to the assumptions when we have the "wrong type of data" to begin with?
>
> I know most of us use SPSS as a data management tool (thus do not delve too much into statistical methods), but I'd appreciate any thoughts on this subject!
>
> Sibusiso.
>
>
>