0
$\begingroup$

Consider the simple linear regression model $ y_i = \beta_0 + \beta_1{x_i} + \epsilon_i$

Suppose in an OLS, the $t$-statistic for the null hypothesis $\beta_1 = 0$ is $1.92$, what is the $F$-statistic for the overall significance of the model?

I am aware that $F$ statistic for testing exclusion of a single variable is equal to the square of the corresponding $t$ statistic: however, I can't seem to find how to link that information to this question. Am I missing information?

$\endgroup$
5
  • 1
    $\begingroup$ Consider what the null hypothesis would be for the “significance of the entire model”. $\endgroup$
    – Dave
    Nov 28 '21 at 22:45
  • 1
    $\begingroup$ Oh it would just be $\beta_1$ = 0, wouldn't it? $\endgroup$ Nov 28 '21 at 23:11
  • 1
    $\begingroup$ That’s what I would say, and that’s what R gives you in a “summary” of a linear model. And then the alternative hypothesis would be $\ne$. // This is a subtle point, but good for you for not writing $\hat\beta_1=0$. $\endgroup$
    – Dave
    Nov 28 '21 at 23:16
  • $\begingroup$ Thank you! I spent way too much time looking for an answer when it was sitting right in front me $\endgroup$ Nov 28 '21 at 23:19
  • 1
    $\begingroup$ Since you pretty much solved the problem yourself, you might consider posting a self-answer. (Cross Validated does not consider it arrogant or poor etiquette to do so. In fact, I have posted several self-answers.) $\endgroup$
    – Dave
    Nov 28 '21 at 23:37
1
$\begingroup$

In the simple linear regression model, testing the significance of the model requires testing the hypothesis: $$H_0 \colon \beta_1 = 0 \\ H_0 \colon \beta_1 \neq 0 $$

Given that we are whether a single parameter is equal to 0, the $F$ statistic will simply equal the square of the corresponding $t$ statistic. This means that if the $t$-statistic for the null is $1.92$, the $F$-statistic will simply be $1.92^2$.

$\endgroup$
2
  • 1
    $\begingroup$ Not sure how this answers your question. $\endgroup$ Nov 29 '21 at 15:36
  • $\begingroup$ The edit might should have made it clearer $\endgroup$ Nov 29 '21 at 17:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.