Date:         Thu, 1 Apr 1999 17:51:34 -0600
Reply-To:     "Nichols, David" <nichols@SPSS.COM>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@UGA.CC.UGA.EDU>
From:         "Nichols, David" <nichols@SPSS.COM>
Subject:      Re: Curve Estimation "S-Curve"

Correction: the natural log of Y is regressed on the reciprocal of the
natural log of X.
 
David Nichols
Principal Support Statistician and
Manager of Statistical Support
SPSS Inc.
 
> -----Original Message-----
> From: Nichols, David [SMTP:nichols@spss.com]
> Sent: Thursday, April 01, 1999 5:23 PM
> To:   SPSSX-L@UGA.CC.UGA.EDU
> Subject:      Re: Curve Estimation "S-Curve"
>
> All of the models fitted in CURVEFIT are actually linear models fitted to
> the original or transformed data. The S model is Y=exp(B0+B1/X), so the
> natural log of Y is regressed on the natural log of X.
>
> David Nichols
> Principal Support Statistician and
> Manager of Statistical Support
> SPSS Inc.
>
> > -----Original Message-----
> > From: Chris Wirtalla [SMTP:ronchito@WORLDNET.ATT.NET]
> > Sent: Wednesday, March 17, 1999 4:48 PM
> > To:   SPSSX-L@UGA.CC.UGA.EDU
> > Subject:      Curve Estimation "S-Curve"
> >
> > Hello -- I am using Curve Estimation in 9.0 to compare the relationships
> > between each of my independent variables against the dependent.  For one
> > particular ind. variable, the "S-Curve" model showed the greatest
> strength
> > in
> > terms of R-squared.  Is there a way that I can "insert" this S-curve
> logic
> > into a linear regression to check it out?  For example, some of my other
> > variables have been shown to have strong quadratic relationships.  To
> > account
> > for this, I add the square of that variable into the linear regression.
> > Is
> > there a similar way that I can do this for the S-curve?
> >
> > Thanks,
> >
> > Chris Wirtalla
> >
> > -----------== Posted via Deja News, The Discussion Network ==----------
> > http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own