Re: st: Comparison of coefficients across groups

[Richard Williams <richardwilliams.ndu@gmail.com>]

At 10:48 AM 3/9/2012, Alina Rusakova wrote:
> Dear all,
>
> I use -suest- in order to compare probit coefficients across groups 
> and the results suggest that there are no significant differences in 
> a variable of interest across groups. However, a coefficient is 
> significant in one group and it is not significant in another group. 
> Given that  this coefficient does not vary across groups, how can I 
> iterpret the results? Can I infer that the variable of interest is 
> important only for one, but not for another group?

There are any number of issues here.

First, group comparisons involving probit and logit coefficients are 
potentially problematic in the first place. See

http://www.nd.edu/~rwilliam/stats/Oglm.pdf

and the sources it cites.

Second, to say that an effect is statistically insignificant is not 
the same as saying it is 0. It just means that you can't rule out 0 
as a possible value. You also can't rule out the variable having even 
larger effects than those estimated.

Third, differences in statistical significance might reflect, say, 
differences in sample size. If you have 1,000 people in the first 
group and only 100 people in the 2nd group, it will be easier to find 
significant effects in the first group.

If this were an OLS regression, I would NOT conclude "that the 
variable of interest is important only for one, but not for another 
group." The effects do not significantly differ across groups. Given 
that it is a probit model, I'd be careful about making the comparison 
in the first place, unless maybe I was trying one or more of the 
approaches listed in the above pdf file.


-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
HOME:   (574)289-5227
EMAIL:  Richard.A.Williams.5@ND.Edu
WWW:    http://www.nd.edu/~rwilliam

*
*   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/