Statalist


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

Re: st: testing the difference between the estimated coefficients


From   John Antonakis <john.antonakis@unil.ch>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: testing the difference between the estimated coefficients
Date   Sat, 16 Jan 2010 12:55:21 +0100

***resending***
(My note was incomplete and I hit send too soon)

1. Yes; it holds for probit or logit models and for dummies. Refer to the postestimation tests for a specific command to see what tests you can do after different estimation procedures.

2. I would not compare standardized coefficients, it confounds difference in the relation with differences in the variances.

3. If the test is significant, it means that the coefficients are
significantly different. You can use the -coef- option to see what the constrained coefficient is.

J.
____________________________________________________

Prof. John Antonakis, Associate Dean
Faculty of Business and Economics
Department of Organizational Behavior
University of Lausanne
Internef #618
CH-1015 Lausanne-Dorigny
Switzerland

Tel ++41 (0)21 692-3438
Fax ++41 (0)21 692-3305

Faculty page:
http://www.hec.unil.ch/people/jantonakis

Personal page:
http://www.hec.unil.ch/jantonakis
____________________________________________________



On 16.01.2010 12:29, Fabio Zona wrote:
Does it hold also for probit or logit regression?? Even if the two
variables in question are dummy variables? If variables are expressed in
different scales, I do standardize them, is it correct?
Finally: if test x = z is significant, it means that I reject the
hypothesys of equality, and, hence, the two coefficients are
significanltly different (and the larger is significantly larger than
the smaller one). Is it correct? Thanks!



----- Messaggio originale -----
Da: "John Antonakis" <john.antonakis@unil.ch>
A: statalist@hsphsun2.harvard.edu
Inviato: Sabato, 16 gennaio 2010 12:01:58 GMT +01:00
Amsterdam/Berlino/Berna/Roma/Stoccolma/Vienna
Oggetto: Re: st: testing the difference between the estimated
coefficients

Hi:

It is simply the -test- command:

reg y x z q
test x = z

Just read about the -test- command. This test constrains the
coefficients to equality and uses a nested F-test to ascertain decrement
in fit (r-square change).  Of course, the usual assumptions hold
regarding whether the test can be done (e.g., that x and z are on the
same scale).

HTH,
J.

____________________________________________________

Prof. John Antonakis, Associate Dean
Faculty of Business and Economics
Department of Organizational Behavior
University of Lausanne
Internef #618
CH-1015 Lausanne-Dorigny
Switzerland

Tel ++41 (0)21 692-3438
Fax ++41 (0)21 692-3305

Faculty page:
http://www.hec.unil.ch/people/jantonakis

Personal page:
http://www.hec.unil.ch/jantonakis
____________________________________________________



On 16.01.2010 11:55, Fabio Zona wrote:
A related issue:
What test should I run, and what STATA commands should I type to
test whether one coefficient of a regression is larger or smaller than
another coefficient? (in the same sample, not comparison of the same
coefficient between groups, but two different coefficients on the same
sample/population and in the same regression)
Thanks!



----- Messaggio originale -----
Da: "John Antonakis" <john.antonakis@unil.ch>
A: statalist@hsphsun2.harvard.edu
Inviato: Sabato, 16 gennaio 2010 11:10:16 GMT +01:00
Amsterdam/Berlino/Berna/Roma/Stoccolma/Vienna
Oggetto: Re: st: testing the difference between the estimated
coefficients

Hi:

This is a job for suest.

regress a b if year==2005 est store five

regress a b if year==2000 est store zero

suest five zero
test [five_mean]a = [zero_mean]b

HTH,
John.

____________________________________________________

Prof. John Antonakis, Associate Dean
Faculty of Business and Economics
Department of Organizational Behavior
University of Lausanne
Internef #618
CH-1015 Lausanne-Dorigny
Switzerland

Tel ++41 (0)21 692-3438
Fax ++41 (0)21 692-3305

Faculty page:
http://www.hec.unil.ch/people/jantonakis

Personal page:
http://www.hec.unil.ch/jantonakis
____________________________________________________



On 16.01.2010 04:39, Sibel S. wrote:
If we run the following regressions,
regress a b if year==2005      *equation 1
regress a b if year==2000      *equation 2
how can one test the difference between the estimated coefficients (b
in equation 1 and b in equation 2) and get the t-statistic of
difference?Thanks.
*
*   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/
*
*   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/
*
*   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/



© Copyright 1996–2022 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index