Thanks everyone for your suggestions.  I think that I'll be setting up
the experiment such that the treatments are fully randomized, and that
should prevent any systematic error and can be controlled as a
residual variance, as Donald and Jos pointed out.  I've also decided
to take one treatment and repeat it (randomly, of course) throughout
the experiment, and use that as a control.  I can do a run test on
those results to see if they're changing much, and if it can be
rearranged with time and modelled, while I wouldn't use it to correct
the other results because it may be specific to that treatment, I can
note it in the discussion.  That's the best I think I can do though,
as far as measuring these changes in the water samples with time,
since there is no specific measurement that I can take to see how
they're changing and how it will affect the results of the experiment.
Again, thanks for your help.  If you have any other comments on the
matter, I'd like to hear them.
- Dave



[EMAIL PROTECTED] (Joe Ward) wrote in message news:<[EMAIL PROTECTED]>...
> Dave --
> 
> If you "can't control it, then measure it" and include it in your model.
> That's the VALUE of using RERESSION/LINEAR MODELS for your analyses.
> 
> -- Joe
> 
> **********************************
> Joe H. Ward, Jr.
> 167 East Arrowhead Dr.
> San Antonio, TX 78228-2402
> Phone: 210-433-6575
> Fax:     210-433-2828
> Email:  [EMAIL PROTECTED]
> http://www.northside.isd.tenet.edu/healthww/biostatistics/wardindex
> ==============================
> Health Careers High School
> 4646 Hamilton Wolfe Road
> San Antonio, TX 78229
> **********************************
> 
> 
> 
>
.
.
=================================================================
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