R help - Apr 2005 - insignificant factors in regression model

Hi list,

I am building a regression model with categorical predictor variable coded 
by treatment contrasts. The summary of the regression model shows that some 
levels are significant while others are not. The significant ones show that 
they are statistically significant from the basis factor (at level 0). How 
do we generally deal with those insignificant levels? If they are used for 
prediction, sometimes they will produce strange results because their 
coefficient estimates have large variances. Do we just simply ignore them 
assuming they are not different from level 0? Or do we exclude them in 
factors (by treating them as zero's) and refit the model?

any suggestions?

Weijie Cai wrote:
> Hi list,
> 
> I am building a regression model with categorical predictor variable 
> coded by treatment contrasts. The summary of the regression model shows 
> that some levels are significant while others are not. The significant 
> ones show that they are statistically significant from the basis factor 
> (at level 0). How do we generally deal with those insignificant levels? 
> If they are used for prediction, sometimes they will produce strange 
> results because their coefficient estimates have large variances. Do we 
> just simply ignore them assuming they are not different from level 0? Or 
> do we exclude them in factors (by treating them as zero's) and refit
the
> model?
> 
> any suggestions?
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide! 
> http://www.R-project.org/posting-guide.html
> 
The fact that an estimated coefficient is not significant does not mean 
that it is negligible. The lower tail of the F-distribution can be used 
to help you decide whether or not to omit a term.


-- 
Bob Wheeler --- http://www.bobwheeler.com/
         ECHIP, Inc. ---
Randomness comes in bunches.