Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: AW: Question about "testparm"


From   Constantine Daskalakis <c_daskalakis@mail.jci.tju.edu>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: AW: Question about "testparm"
Date   Sun, 01 Feb 2009 11:41:27 -0500

There is no redundancy here.

What you asked Stata for was a "global" test

H0: Var1 = Var2 = Var3
HA: any difference

In other words, is any of these 3 coefficients different from any one else. Hence, a global test w/ 2 df.

I think you are looking for the pairwise tests, ie, 3 tests

H0: Var1=Var2
HA: Var1 NOT= Var2

H0: Var1=Var3
HA: not

H0: Var2=Var3
HA: not

Then, you would need to run your -testparm- 3 separate times: one for Var1/Var2, another for Var1/Var3, and a third for Var2/Var3.

*** Comment:
If these are 3 dummy variables for a 4-level nominal variable, why are you doing this test?

1. Are you interested in whether these 3 levels are the same and different from the baseline? If so, a global test of var1=var2=var3 (and if significant, followed by the 3 pairwise ones) would seem to be reasonable.

2. If you are interested in whether the nominal variable as a whole is significant, then this is NOT the right test. You should be testing the global

H0: Var1=Var2=Var3=0
HA: not

In other words, a 3 df test of

H0: Var1=0 & Var2=0 & Var3=0
HA: not

Best,
CD





On 1/31/2009 4:34 PM, Ted Fuller wrote:
Martin,
Thanks for your quick response!
I confess I am still a little puzzled.
Why is one constraint redundant?
My three dummy variables are three categories of a nominal variable,
but my nominal variable has four categories. So, knowing the value of
two variables does not necessarily indicate the value of the third.

If the single tests are accumulated into "one big test", that does not seem to answer my question, which was to find out if the three coefficients vary among themselves. With three coefficients, isn't that three questions (does a=b, does b=c, does a=c)? How can "one big test" answer all
three questions?

In answer to your question, "Why open the dialog box for -testparm- before "submitting your analysis"?" it is because I am using menus rather than lines of code, and I could not see a way to ask for the what I wanted in the menu for "Statistics - Survey data analysis - Distribution-specific models - Linear regression".

Thanks.

Ted

At 03:59 PM 1/31/2009, you wrote:

<>

Stata realizes that one constraint in your specification is redundant and
disregards it, just as it should... See [R] test, page 456. You can of
course accumulate the single -test-s into one big test one after the
other...

*************
webuse highschool, clear
svy: regress weight height state county school
test he=sch
test he=state,accum
test scho=state,accum
*************

As the example shows, even then Stata recognizes the redundancy in the last
line.

BTW, why open the dialog box for -testparm- before "submitting your
analysis"?

HTH
Martin


-----Ursprüngliche Nachricht-----
Von: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] Im Auftrag von Ted Fuller
Gesendet: Samstag, 31. Januar 2009 21:46
An: statalist@hsphsun2.harvard.edu
Betreff: st: Question about "testparm"

This is probably a pretty basic question.
I am using svyregress.
Before submitting my analysis, I opened the dialog box for "testparm"
and specified three variables and indicated that I wanted to test
whether the coefficients for these three variables were equal.
The output said:
(1) - var1 + var2 =0
(2) - var1 + var3 =0
    F(2, 338) = 13.10
        Prob > F = 0.0000

I thought I was going to get three tests:
(1) to see if the coefficient for var1 = coefficient for var2;
(2) to see if the coefficient for var1 = coefficient for var3;
(3) to see if the coefficient for var2 = coefficient for var3.

But I only got one test.
How can I get the three tests that I mentioned?
Thanks!
Ted Fuller

Theodore Fuller
Department of Sociology (0137)
Virginia Tech
Blacksburg, VA 24061
Phone: 540/231-8969

*
*   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/

Theodore Fuller
Department of Sociology (0137)
Virginia Tech
Blacksburg, VA 24061
Phone: 540/231-8969

*
*   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/


--


The documents accompanying this transmission may contain confidential
health or business information. This information is intended for the
use of the individual or entity named above. If you have received
this information in error, please notify the sender immediately and
arrange for the return or destruction of these documents.

Constantine Daskalakis, ScD
Associate Professor,
Thomas Jefferson University, Division of Biostatistics
   1015 Chestnut St., Suite M100, Philadelphia, PA 19107
   Tel: 215-955-5695
   Fax: 215-503-3804
   Email: c_daskalakis@mail.jci.tju.edu
   Webpage: http://www.jefferson.edu/clinpharm/biostatistics/


*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index