CODEWITHSUNDEEP
statisticsp-value
tooBad
tooBad

Reputation: 25

f-statistics and p value suggest different results

In a multiple linear regression, if I would like to know is there a relationship between the response and predictors,

The book said how large does the F-value need to be before reject Ho depends on the p-value. I just wondered if there gonna be a case where p-value and f-statistics suggest different things. Then which one should we rely on?

Thanks!

Upvotes: 1

Views: 973

Answers (1)

soakley
soakley

Reputation: 233

The F-statistic and p-value will not conflict with each other. The p-value is a measure of how extreme the F-statistic is - it's a tail probability from the F distribution.

So, for example, suppose your criterion of choosing the alternative hypothesis is a p-value less than or equal to 0.05 (a commonly used value) and the F distribution critical point corresponding to this p-value is 2.2

Then if the F statistic is > 2.2 the null hypothesis is rejected. If this occurs, the p-value must be < 0.05

Correspondingly, if the p-value is < 0.05, then the F statistic must be > 2.2

Upvotes: 2

Related Questions