F - distribution trouble

[ptc-5071151]

Hello everyone,

I have some trouble with a function dF(x, d1, d2). I would like to aquire this way p value from F distribution for a known sample with identified F value.

Number of independent variables: 2

Number of dependent variables: 1

Sample's size: 20

F value: 16.8833

I implemented data to formula: dF(16.8833,2,17) and got a result 3.064 *10^-5. But this is not correct for my needs, because the desired value is 9.15*10^-5.

When I calculate it in reverse way through formula qF(1-9.15*10^-5,2,17) I got 16.8833. Can you explain me what I am doing wrong?

[Werner_E · Accepted solution]

It may help if you would attach your worksheet.

You may also additionally consider explaining why you think that Mathcad's result is incorrect.

[ptc-5071151]

Maybe is correct, but this is not what I would like to get.

Mentioned formula qF calculates critical value from F - distribution for known level of confidence, what I would like to do is to find level of confidence with known critical value, so do it in a reverse fashion for multiple regression analysis (do not ask me why I do it step after step ). Presumably I messed something up with degreess of freedom. With t student everything was working in comply with scheme, which I presented above.

[Werner_E]

Shouldn't you rather user the cumulative probability pF() instead of the probability density dF()?

[ptc-5071151]

Oo. Thanks