You are doing some kind of exploration, trying to find significant
factors (ind. vars.), right?  Then you can drop the 'not significant'
ones, at an SL you have previously selected, and see what happens.

I suggest you do the stepwise regression in 'proper' form - run them all,
drop the weakest factor, run again, drop the next weakest one, etc.

Note:  You should not drop low order variables when a high order is
significant.

If you are looking to develop a prediction equation, dropping weak ones
is problematical - I do it a lot, but I get criticized for it.  I like my
prediction equations to be as simple as possible, but not simpler.

If you carefully set up the experimental conditions to test for all 15
variables (first question - why 15?), so that the design array is
orthogonal, then some people say you should keep the weak ones, on the
grounds that you had good technical reason for putting them in in the
first place.  I say, let's go back to that 'simple' part, OK?

If your design array is not near orthogonal, all your interpretations are
suspect anyway, and dropping variables may have dramatic effects on the
coefficients.  How you interpret these changes, and the coefficients, is
not for the faint of heart.

Cheers,
Jay

Bastian wrote:

> Hello,
>
> I did a regression analysis with 15 variables and 4 of them were not
> significant. I'm not quite sure what's the best solution for this
> problem:
>
> - leaving the regression equation like it is with all variables and
> just don't interprete the not signifikant variables
>
> or
>
> - making a new regression analysis without the not significant
> variables, i.e. with the method "stepwise".
>
> Any comments on this or literature how to solve this problem right? I
> really appreciate every answer and have to admit I'm quite a newbie in
> statistics...
>
> Thanks a lot,
>
> Bastian
> .
> .
> =================================================================
> Instructions for joining and leaving this list, remarks about the
> problem of INAPPROPRIATE MESSAGES, and archives are available at:
> .                  http://jse.stat.ncsu.edu/                    .
> =================================================================

--
Jay Warner
Principal Scientist
Warner Consulting, Inc.
4444 North Green Bay Road
Racine, WI 53404-1216
USA

Ph: (262) 634-9100
FAX: (262) 681-1133
email: [EMAIL PROTECTED]
web: http://www.a2q.com

The A2Q Method (tm) -- What do you want to improve today?




.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to