[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: LARGE F-values in ordinary OLS
In a normal regression (includes intercept) the test is whether the fit
differs from the mean y value.
Without the intercept, the test is whether the fit differs from zero.
Thus the F is testing whether the mean differs from zero as well as whether
the fit differs from the mean. The size of the F will be directly related
to the absolute size of mean y in a no-constant regression. What your
results suggest is that your absolute mean y is large.
P.S. I've never heard of suppressing the constant for
----- Original Message -----
From: "carambas" <firstname.lastname@example.org>
Sent: Saturday, June 12, 2004 4:42 PM
Subject: Re: st: LARGE F-values in ordinary OLS
> My sample size is even lower than a hundred. I forgot to say that I am
> suppressing constant (i.e. homogeneity reasons) and I know that this is
> it gives high F values (because by normal regression, everything is fine)
> but I am not quite sure how this noconstant fn comes up with large F
> I would appreciate another hint. Thanks again.
> At 07:27 AM 6/11/2004 +0200, carambas wrote:
> Dear Statalist,
> I'm kinda alarmed to see F values that are so high (reaches 38,000). Is
> there anything wrong. I thought it's too good to be true. Thanks.
> If the sample size is 100, you may be on to something pretty cosmic. If it
> is 100,000 it may not mean much. It is hard to tell just based on the info
> you have given.
> * For searches and help try:
> * http://www.stata.com/support/faqs/res/findit.html
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
* For searches and help try: