[R] Chi-square test

[David Winsemius]

On Feb 20, 2015, at 10:05 AM, pari hesabi wrote:

> Hello,
> If the vector of observed frequencies is:  f<-c(0,0,0,2,3,6,17,15,21,21,14,10,5,1,5)
> and the vector of probability :p11<-c(7.577864e-06, 1.999541e-04  ,1.833510e-03,  9.059845e-03, 2.886977e-02, 6.546229e-02 ,1.124083e-01, 1.525880e-01, 1.689712e-01, 1.563522e-01,   1.232031e-01, 8.395000e-02, 5.009534e-02, 2.645857e-02,0.0205403)
> The sum of the probabilities is equal to one.

Well, the sum is close to 1.0 but not exact. There's a simple fix:

> sum(p11)==1
[1] FALSE
> sum( p11/sum(p11) )==1
[1] TRUE

>  But when I want to do the the Chi-square test, I get this error: probabilities must sum to one.
> Does anybody know the reason?

Numerical accuracy. See R-FAQ 7.31

-- 

David Winsemius
Alameda, CA, USA