> Date: Fri, 22 Dec 2000 08:16:50 -0600
> From: Debraj <[EMAIL PROTECTED]>
> Subject: bias
>
> hi,
>
> Some data values in a particular experiment performed exhibits
> unexpected results. Looking into it, I found some explanations for the
> same. Can I add in some compensatory values/function to eliminate the
> bias caused in the data instead of redoing the experiment again (with
> corrections) ?
>
> regards,
> debraj
The 'advantage' of this approach is that your model will fit your 'data' better. (Of
course it will - some dubious points were removed,
replaced by ones that fit much more nicely.)
the disadvantage is that you are adding points to fit a model of the world that you
have, perhaps in your head. So the data will tend
to fit it better than the reality. At which point we must ask, of what use is the
model & data, since the data came from the same place
as the model - your head?
IN order to produce something supportable, you _must_ select and adjust your odd-ball
points by means which are independent of the model
& fit that you are calculating. Preferably before you initialize your model. It's a
bit tricky, and you may have trouble justifying
your choices to a review committee. But once you have the model in front of you, your
judgment is biased even to the
selection/adjustment of data points.
Jay
--
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 and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
