Re: st: T-test and F-test

[natasha agarwal <agarwana2@googlemail.com>]

On Mon, Jul 5, 2010 at 3:25 PM, Airey, David C
<david.airey@vanderbilt.edu> wrote:
> .
>
> I was trying to remind myself about collinearity and suppression in regression recently:
>
> Graphical Views of Suppression and Multicollinearity in Multiple Linear Regre...
> Lynn Friedman; Melanie Wall The American Statistician; May 2005; 59, 2; pg. 127
>
> Maybe, besides what Roger says, your regression has some issues with collinearity? Not that this is always bad, as the the above article says.
>
> You might look at the Stata command -vif-, but also the user written command -collin-, to understand more about your predictors as a set, and why each individually predicts Y, whereas as a set you have a significant model but lose significance on each individual coefficient.

I tried using the correlation matrix to look at the same. The
correlation matrix gives me very low correlation between my variables
(atleast according to the matrix is quite low).

However, I also tried collin and coldiag2 but there is no information
available on the net to interpret the same. But the time dummies show
1's in the variance decomposition proportions which I found it very
strange.

These two conflicting results have led me to a confusion. Could you
tell me some more reading literature on this if you are aware of?

Thanks
Natasha

> -Dave
>
>
>> Dear everyone,
>>
>> Can anyone explain me why the t-test statistics for variables in the
>> model are statistically insignificant but when tested for joint
>> significance (F-test) they are jointly significant?
>>
>> Thanks
>> Natasha
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
>

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/