chitest and interpretation of the result
 

When I am using the statistical function "chitest" (the word is in Danish,
since i dont have the english version, but i presume the word is more or less
the same in the english version), I have difficulties interpretating the
result.

If the result is 0.01, does is mean that there is 1% probability of my
samples originating from the same underlying distribution? Or does it mean,
that my samples are from the underlying distribution with a CI of 1 %???

I sincerely hope someone in this forum can help me.

Christine -

Christine -

When I am using the statistical function "chitest" (the word is in Danish,
since i dont have the english version, but i presume the word is more or
less the same in the english version), I have difficulties interpretating
the result. If the result is 0.01, does is mean that there is 1%
probability of my samples originating from the same underlying
distribution? Or does it mean, that my samples are from the underlying
distribution with a CI of 1 %??? <

Because of the way the CHITEST function computes degrees of freedom, it is
most appropriate only for tests of the independence of classifications
arranged in a contingency table.

If you are doing a chi-square test for goodness of fit (comparing sample
data with a hypothesized underlying distribution), I recommend using the
CHIDIST function to obtain the p-value. To use CHIDIST, you must first
compute the chi-square statistic yourself. But CHIDIST allows you to specify
the appropriate degrees of freedom for your situation.

The p-value returned by both CHITEST and CHIDIST is a standard way of
reporting the result of a hypothesis test. In general, a p-value reports how
likely it is that the observed sample result, or a sample result more
extreme, could be obtained if the null hypothesis is true.

For a test of the independence of classifications arranged in a contingency
table, the CHITEST function returns the probability that the actual
frequencies (or more extreme frequencies) could be obtained in a random
sample if the classifications are independent.

For a test of goodness of fit, the CHIDIST function returns the probability
that the actual frequencies (or more extreme frequencies) could be obtained
in a random sample from the hypothesized distribution.

In general, a small p-value indicates a very unlikely result under the null
hypothesis, so the null hypothesis may be rejected. A large p-value
indicates the observed sample is quite likely to occur under the null
hypothesis, so the null hypothesis may not be rejected.

- Mike
www.mikemiddleton.com