[Limdep Nlogit List] Comparing coefficients

[William Greene]

Harold.  I'm assuming you have three separate, independent regressions,
and you have three coefficients, b1, b2, b3.  I'm not sure what you mean by
"compare."  It sounds like you want to test the hypothesis that the three
coefficients are equal.  But, the alternative is that any one of the
coefficients
differs from the other two (which might be equal) or that all three are
different.
Formally, the null hypothesis is beta1 - beta3 = 0 and beta2 - beta3 = 0,
two
restrictions.  As long as we're agreed on the ambiguity of the alternative,
which is "not the null," you can use a Wald test. (Asymptotically
chi-squared
with 2 degrees of freedom.)  The test statistic is d' * inv(V) * d, where d
is a
2x1 vector containing (b1 - b3) and (b2 - b3) and V is a 2x2 covariance
matrix
containing s1^2 + s3^2 then s3^2 in the first row and s3^2 then s2^2 + s3^2
in the second where sj^2 is the estimated asymptotic variance for
coefficient bj.
Regards
Bill

On Sun, Dec 25, 2022 at 1:02 PM Harold Mayaba via Limdep <
limdep at mailman.sydney.edu.au> wrote:

>  Dear Professor Greene,
> In line with the reply from Zein, the results are from three treatments
> analyzed separately.  I would like to establish if there is a statistically
> significant difference between the coefficients from three different
> treatments.  The method I found compares coefficients from each treatment
> as a group not individual coefficients. For example,  I have 'access to
> outdoor' attribute in each treatment.  I would like to compare the
> coefficients of this attribute i got from each treatment.  I hope this
> helps.
> Regards
> Harold
>     On Friday, December 23, 2022 at 08:20:20 AM GMT+13, Harold Mayaba <
> mayabach2009 at yahoo.com> wrote:
>
>  Dear Professor Greene and other nlogit user,
> How can I compare two or three coefficients from different treatments to
> establish if there is a significant difference? I have seen literature
> where coefficients of different treatments are compared but not
> coefficients of the same attribute in different treatments.
>
> Thanks
> Harold
>
> _______________________________________________
> Limdep site list
> Limdep at mailman.sydney.edu.au
> https://protect-au.mimecast.com/s/ZVTTCXLW2mUnR9rPKh6i8ND?domain=limdep.itls.usyd.edu.au
>


-- 
William Greene
Department of Economics, emeritus
Stern School of Business, New York University
44 West 4 St.
New York, NY, 10012
URL: https://protect-au.mimecast.com/s/V2qhCYW8Noc3q9BWlFGHyVp?domain=people.stern.nyu.edu
Email: wgreene at stern.nyu.edu
Editor in Chief: Foundations and Trends in Econometrics
Associate Editor: Economics Letters
Associate Editor: Journal of Business and Economic Statistics