st: Re: stata command for joint significant in large sample test

[Kit Baum <baum@bc.edu>]

As Joanne is interested in testing for joint significance, she will  
no doubt find that one of these tests has three degrees of freedom  
(representing constraints placed on the unrestricted parameter set)  
and the other one has one degree of freedom. If Joanne had taken the  
time to read through any elementary statistics/econometrics text on  
the subject of 'joint significance' she would have been too  
embarrassed to ask this question of Statalist.

In terms of a 'large sample' test, whatever Joanne is estimating via  
regress, e.g.

regress y educ wage married

If she would use ivreg2 (from ssc):

ivreg2 y educ wage married

By default ivreg2 produces 'large-sample' statistics, i.e. z rather  
than t, and -test- used after ivreg2 will produce chi^2 (large- 
sample) rather than F (finite-sample) tests.


Kit Baum, Boston College Economics
http://ideas.repec.org/e/pba1.html
An Introduction to Modern Econometrics Using Stata:
http://www.stata-press.com/books/imeus.html


On Jan 21, 2007, at 2:33 AM, Joanne wrote:


> if i was to test the joint significance test using a large sample  
> test in
> stata for educ, wage and married
>
> which of the following command do I use in Stata?
>
> test educ wage married
>
> or
>
> test educ+wage+married=0
>
> they generate different results, actually opposite results, one being
> significant and one insignificant. Note I will want to use a large  
> sample
> test here.
*
*   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/