Re: Chi Squared Test

[William V Lipton <liptonw@dteenergy.com>]

My apologies, 1 correction:  chi square is actually (n-1) times the ratio of the
actual variance to the theoretical variance, where n is the number of counts.

The opinions expressed are strictly mine.
It's not about dose, it's about trust.

Bill Lipton
liptonw@dteenergy.com


William V Lipton wrote:

> At Fermi 2, we require 15 sequential source counts when setting up a quality
> control chart for a laboratory instrument.  The count time must be
> sufficient to generate at least 40,000 counts.  The results are used to
> calculate a chi square value, which is the ratio of the observed variance to
> the theoretical variance based on Poisson statistics.  The upper and lower
> acceptable limits for this value are the 0.975 and 0.025 probability values
> for chi square, which, for 15 counts, would be 26.1 and 5.63.   Please
> e-mail me if you have any questions on this.
>
> The opinions expressed are strictly mine.
> It's not about dose, it's about trust.
>
> Bill Lipton
> liptonw@dteenergy.com
>
> Joel Cehn wrote:
>
> > We use the chi-squared test on 10 sequential readings to check the
> > stability of a wipe counter (Ludlum 2929.)  What do people consider
> > acceptable results from this test?
> >
> > Joel Cehn, CHP
> > JCehn@att.net
> >
> > ***********
>
> ************************************************************************
> The RADSAFE Frequently Asked Questions list, archives and subscription
> information can be accessed at http://www.ehs.uiuc.edu/~rad/radsafe.html

************************************************************************
The RADSAFE Frequently Asked Questions list, archives and subscription
information can be accessed at http://www.ehs.uiuc.edu/~rad/radsafe.html