Thank you for your response. Let me describe my problem in more detail
My response variable is continuous and ranges betwen 0.77 to 1.2. I
have 6 predictor variables each at 3 levels. I have 31 observations.
My full model contains all the main effects (which would be 6 terms),
the pair wise second order interaction terms(which would be 15 terms)
and the square terms , i.e. x1*x1 x2*X2 X3*X3 X4*X4 X5*X5 and X6*X6.
So, I have a total of 28 terms including the intercept
When I did the stepwise regression I was able to get rid of X1*X2 ,
X1*X3, X2*X5 , X4*X4 and X5*X5(I apologize the error in my post, which
says I was able to get rid of only 3 variables) So, I basically have 8
df for my error term.
Thanks for your help!
In article <[EMAIL PROTECTED]>,
[EMAIL PROTECTED] (Donald Burrill) wrote:
> I don't usually respond to anonymous querents, but the problem is
> intriguing.
>
> On Sat, 30 Sep 2000 [EMAIL PROTECTED] wrote:
>
> > I constructed a D-optimal design for 6 continuous variables, each at
> > three levels.
>
> Is that 6 predictors, or 5 predictors and a response variable?
>
> > I have 31 runs.
>
> By this I understand you to mean that there are 31 observations in
> the data set. If this is in error, perhaps you could describe things
> a tad more explicitly.
>
> > My initial model includes, all the main effects, all interactions
and
> > polynomial terms.
>
> Then you must be running out of d.f. for error. You have only 30
d.f.
> in total, and you've described a model requiring 31 d.f. for 5
> predictors, even without considering the polynomial terms.
>
> > I was only able to remove 3 of the terms using step wise
regression.
> > My final model has an R square of 0.9954, which looks very
artificial.
>
> A consequence of the variety required in the model vs. the variety
> supplied in the data.
>
> > My adjusted R square is also very close to R Square.
>
> As is natural for R very close to 1.
>
> > Does anyone have any suggestion on what could have gone wrong or if
> > there is a different analysis technique that I can use?
>
> You should be able to deal with it from here.
>
> > I even did a lack of fit test, and the interaction terms and square
> > terms were significant.
>
> If as I suspect your error SS was virtually zero, this is to be
expected.
>
> One still wonders what you were using for error d.f. Perhaps your
claim
> of modelling "all interactions", which I interpreted as meaning "up
to
> and including 5-way interactions" (or 6-way interactions, if you
meant
> that your model began with 6 predictors), was hyperbolic?
>
-- DFB.
> ---------------------------------------------------------------------
-
> Donald F. Burrill
[EMAIL PROTECTED]
> 348 Hyde Hall, Plymouth State College,
[EMAIL PROTECTED]
> MSC #29, Plymouth, NH 03264 (603) 535-
2597
> Department of Mathematics, Boston University
[EMAIL PROTECTED]
> 111 Cummington Street, room 261, Boston, MA 02215 (617) 353-
5288
> 184 Nashua Road, Bedford, NH 03110 (603) 471-
7128
>
> =================================================================
> Instructions for joining and leaving this list and remarks about
> the problem of INAPPROPRIATE MESSAGES are available at
> http://jse.stat.ncsu.edu/
> =================================================================
>
Sent via Deja.com http://www.deja.com/
Before you buy.
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
