Re: st: bootstrap _b VS bootstrap _se

[Maarten buis <maartenbuis@yahoo.co.uk>]

--- On Tue, 20/7/10, Sirak wrote:
> I have run bootsraping for coefficient and standard
> deviation with some constraints as in the following:
> 
> constraint 1 _b[y3x1] = 1 - _b[y2:x1] - _b[y1:x1]///
> constraint 2 _b[y3x2] = 1 - _b[y2:x2] - _b[y1:x2]///
> constraint 3 _b[y3x3] = 1 - _b[y2:x3] - _b[y1:x3]///
> 
>     1)     bootstrap _b,
> reps(100) : sureg (y1 x1 x2 x3 x4) (y2 x1 x2
> ) (y3 x1 x2 x4 ), constraints (1 2 3)     
>    ///
>     2)     bootstrap _se,
> reps(100) : sureg (y1 x1 x2 x3 x4) (y2 x1 x2
> ) (y3 x1 x2 x4 ), constraints (1 2 3)    ///
> 
> In the first equation the constraint works properly and I
> found the sum of respective X's in each equation is one.
> However, in the second equation, i got the estimates.....
> no errors are observed..., but the constraint that the sum
> of each respective x equals to one fails.

In the second command you are bootstrapping the standard
errors, i.e. you are looking how uncertain your estimate
of uncertainty is. That is didactically not a bad idea, it
clearly illustrates that we should not take our p-value of
5% to strictly as there is uncertainty in that p-value as
well. However, I doubt whether that is what you want to do 
here. 

Anyhow, the constraint you specified says that your 
parameters add up to one, not that your standard errors add 
up to one. So Stata does exactly what you told it to do.

Hope this helps,
Maarten

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://www.maartenbuis.nl
--------------------------


      

*
*   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/