Given a Full and Reduced model, the F-test to see if the reduced model is significant is given by

$$ {{SSE_R-SSE_F}\over{(n-q)-(n-p)}}/{{SSE_F}\over{n-p}} $$

I'm trying to understand how this is equal to ${MSR}\over{MSE}$. It's trivial to observe that $MSE = {{SSE_F}\over{n-p}}$, but what are the steps needed to show that $$ {{SSE_R-SSE_F}\over{(n-q)-(n-p)}} = MSR $$ i.e. is this based on an assumption, or can they be shown to be equal algebraically?

  • $\begingroup$ Please define MSR. $\endgroup$ May 19 '18 at 14:39
  • $\begingroup$ Are you talking about the homoskedasticity only F-statistic? Where did you find that it is equal to $MSR/MSE$, could you include some references? Usually this statistic is expressed either in terms of $SSR$ or $R^2$. $\endgroup$
    – byouness
    May 19 '18 at 14:40

Your Answer

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

Browse other questions tagged or ask your own question.