Re: st: specifying difference in Ho of ttest

[Richard Williams <Richard.A.Williams.5@nd.edu>]

At 11:01 PM 1/14/2004 -0500, Michel Camus wrote:
> I am not sure that I get the issue correctly, having missed the original 
> thread.
> However, if a t-test were appropriate and if the samples are independent, 
> would not a linear regression of the continuous variable against the group 
> variable, followed by testing the parameter estimated for the group 
> variable address the issue in a more straightforward way?
> Ex. : suppose you want to test if the difference if significantly 
> different from 10 between two groups.
>     . regress continuous dichotomous
>     . test dichotomous = 10
> Michel
Yet another more straightforward solution than I initially proposed; about 
the only downside I see would be if you wanted to specify unequal variances 
on the t-test.  With yours and Kit's and Scott's solutions, the trick is to 
take the original problem, which is awkward to address with a matched pairs 
or independent samples t-test, and recast it into an equivalent problem 
where it is easy to test the desired constraints.  It might be nice if some 
future version of the t-test command let you specify differences other than 0.

Kit mentions Anova may be able to do this too.  Yes, it would; commands 
equivalent to what you have above are

. anova  continuous dichotomy
. test _b[dichotomy[1]]=-10

In the case of Kit's example, the corresponding Anova commands would be

. anova  price foreign
. test _b[ foreign[1]] = -200

The sign reversal is necessary because Anova, in effect, reverses the 
coding of a 0-1 dichotomy.  I find this syntax more awkward than the 
corresponding regression command, but it works fine.


-------------------------------------------
Richard Williams, Associate Professor
OFFICE: (574)631-6668, (574)631-6463
FAX:    (574)288-4373
HOME:   (574)289-5227
EMAIL:  Richard.A.Williams.5@ND.Edu
WWW (personal):    http://www.nd.edu/~rwilliam
WWW (department):    http://www.nd.edu/~soc

*
*   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/