Re: Chi square test help needed

luca.pamparana@xxxxxxxxx wrote in news:6b7773e6-ece2-4a72-abe3-
d537d423c880@xxxxxxxxxxxxxxxxxxxxxxxxxxxx:

Hi everyone,

I have a confusion about the Chi square test as to what the
denominator in the formula means. I am trying to assess if my data
fits the Poisson distribution and I have the following set of values
for observed and expected values:

observed expected
6 7.44
5 4.97
9 5.47
2 5.15
13 11.34

Now the formula says that the chi-square is:

sum (observed - expected)^2 / (sigma)^2

Now, in case of Poisson distribution, what is this sigma? is it the
standard deviation or the error the measurement error.

So, is the sigma = standard deviation on observed data / sqrt(N) in
this case (N = 5).


For Poisson distributed variables, sigma^2 = expected. So your formula
could be recast as:
sum(O - E)^2 / sum(E)

A point of terminology. That is a test statistic that is compared to chi-
square critical values with an appropriate number of degrees of free dom
(in this case 4) rather than _being_ chi-square.

--
David Winsemius
.

Quantcast