CHITEST function

[romunov]

I'm having trouble using the aforementioned function. When I calculate the chi test by hand I get a "good result", however, the function keeps returning minute "incorrect" values. Has anyone toyed around with this function and can chip in? Is the function functional at all?

[Villeroy]

Works for me.

[romunov]

Can you test these values?

Experimental: 5 4 1 1 0 0 0 0 0 0
Expected:2,63 2,87 1,91 1,91 0,96 0 0,24 0 0,48 0

When I perform CHITEST, I get a return value of zero (0,00).

[jrkrideau]

Can you test these values?

Experimental: 5 4 1 1 0 0 0 0 0 0
Expected:2,63 2,87 1,91 1,91 0,96 0 0,24 0 0,48 0

When I perform CHITEST, I get a return value of zero (0,00).

For some reason Calc gives 0 (zero) as the answer but unless I missing something it shouldn't give anything except an error message since expected cell sizes of zero would mean dividing by zero.

Is this a bug?

[Villeroy]

If we accept Gnumeric as pseudo-standard for statistics in spreadsheets then yes, it's a bug. Gnumeric returns #NUM if there are zero measurements.

[jrkrideau]

If we accept Gnumeric as pseudo-standard for statistics in spreadsheets then yes, it's a bug. Gnumeric returns #NUM if there are zero measurements.

Gnumeric is much closer to real stats than either Excel or Calc.

A couple of refs gives the formula for a X^2 test as X^2 = Σ(O − E)^2/E which definately implies division by zero with this dataset. so I think we have bug.

Just tried it in Excel and I get a div/0 error

[romunov]

Some datasets do have zero, but the chi test is a SUM of devisions, so an invalid (div/0 error) should not render a function useless.

[Villeroy]

Just for the records: http://www.openoffice.org/issues/show_bug.cgi?id=89691
Thank you for filing, jrkrideau.

[jrkrideau]

Just for the records: http://www.openoffice.org/issues/show_bug.cgi?id=89691
Thank you for filing, jrkrideau.

Thanks for posting this. I totally forgot it.