I suspect you are trying to use 'rules of thumb' without finding out how
much deviation from 'perfection' you are willing to tolerate. And how
much the originator of the rule was willing to tolerate.
1) 1 var for 10 obs - would it be 1 for 8 if we had 8 digits on our
hands? If you have an orthogonal array, you can deal with 7 variables in
8 observations. If you have happenstance data, 1 var in 100 observations
may be questioned. Yes, this is not an answer. Sorry 'bout that.
2) No comment - I can't discuss
3) If the (indep) variable is highly correlated with another (indep)
variable, such a s 0.8, then you should _not_ add it in. A waste of
time. Don't even go there.
If the (indep) var is highly correlated with the (dep) variable,
then the data is begging you to include it. Subject to the previous
sentence, of course.
And again, I'm not sure why a number like 100 (=10^2) should be
used as a lower limit. What you need to look at is how well your other -
hidden, nuisance, etc. - variables were controlled/monitored.
All of this, of course, is IMHO. :)
Good luck,
Jay
Bastian wrote:
> Hello,
>
> there are lots of suggestions for the minimum data for regression
> models, i.e.
>
> 1) 1 var. for every 10 observations.
> 2) Variables can be added until Adjusted R-square deviates
> substantially Unadjusted R-square.
> 3) With relatively large samples (n=100), a variable can be added
> if is correlation with other variables is no larger than about
> 0.80 or 0.85.
>
> My question is, whether the 1) suggestion should be condsidered with a
> stepwise regression, too. That means when I do a stepwise regression
> and get i.e. 20 variables are 200 obeserved values sufficient for
> prediction or is it important to have as much observations as there
> are "possible" variables before the use of the stepwise method.
>
> 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/ .
=================================================================
