Date:         Wed, 27 Jan 1999 12:29:28 -0500
Reply-To:     "David F. Greenberg" <dg4@IS3.NYU.EDU>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@UGA.CC.UGA.EDU>
From:         "David F. Greenberg" <dg4@IS3.NYU.EDU>
Subject:      Re: SPSS 8.0
Comments: To: shailendra@CANADA.COM
In-Reply-To:  <9901271148445C.28024@weba3.iname.net>
Content-Type: TEXT/PLAIN; charset=US-ASCII

Most people would treat your scale as an interval-level variable, though,
strictly speaking, an ordinal logit would be more appropriate. SPSS gives
several diagnostic indictors such as variance inflation factors that will
help you make a decision about multicollinearity among your predictors. -
David Greenberg, Sociology Department,New York University 
 On Wed, 27 Jan 1999
shailendra@CANADA.COM wrote:
 
> In a perfect regression analysis, we need interval or ratio scaled variables.
> 
> Well I don’t have these. My dependent variable has Likert Scaling (1-to-5, where 1 is poor and 5 is Excellent). My independent variables have nominal scale. Well, nominal scale can be converted into dummy variables (0,1) so it they can be treated as an interval-level scale.
> 
> What you say about my dependent variable?
> 
> My another concern was collinerarity effect of independent variables. I am expecting correlation between some independent variables. How do you treat them if one or two variables are correlated? Also is there any correlation benchmark for accepting variables, which seems to be correlated?
> 
> Thank you,
> 
> Shailendra
> 
> 
> --------------------------------------------------------------------
> Get free personalized email from Canada.com at http://www.ehmail.com
>