Re: Chi Square Test

["Fritz A. Seiler" <faseiler@nmia.com>]

RON LAVERA wrote:

> We've used the CHI Square test for many years as a method of checking
> the performance of gas proportional detector, scaler equipped
> instruments.  We've always assumed that when the tested indicated that
> there was not enough variation in the data, that the cause was most
> likely due to constant frequency electronic noise reducing the variability
> in the count rate.
>
> Would any body out there be willing to share their knowledge /
> experience on this one ?
>
> Thank you in advance for your time and consideration.
>
> Ron LaVera
> LAVERA.R@NYPA.GOV

Dear Ron,

And an appropriate measure it is!  We used to predict the come-uppance of high
energy solid state detectors by a chi-square test.  Once it got too narrow, we
knew that the detector would die shortly.  There is something very suspicious
about a stochastic quantity that does not use its full measure of variability
--  something must be wrong!  There is just no way that a Poisson process such
as counting decay events can be more narrow than sqrt(n).  A non-Poisson process
must be involved.  In beginning physics lab I introduced a Poisson statistic
experiment which - by the flick of a switch - introduced some additional 50
cycle spikes.  The student had to find out why the counting statistic all of a
sudden did not work anymore.  That was an off-shoot of our experience at the
meson factory, and is obviously in line with your finding.

Best regards

Fritz

*************************

Fritz A. Seiler, Ph.D.
Principal
Sigma Five Associates
P.O. Box 14006
Albuquerque, NM 87191-4006
Tel.     505-323-7848
Fax.    505-293-3911
e-mail: faseiler@nmia.com

**************************