Hi Rafa. I ran into a similar problem just the other day. I believe a solution is to use the following slightly awkward syntax.
mat A=e(b)
scalar s = A[1,colnumb(A,"x2")]
The colnumb function returns the column number of a matrix that has a label matching the name in quotes.
Cheers,
Steve
-----Original Message-----
From: [email protected]
[mailto:[email protected]]On Behalf Of R.E. De Hoyos
Sent: Wednesday, July 13, 2005 11:30 AM
To: [email protected]
Subject: st: Re: Re: replacing a scalar from mat
Michael,
Thanks for your answer.
Your suggestion works fine. However I still face the row and column name
problem when I try to replace a the scalar (or 1x1 matrix) into another
vector. I should've explained myself better. The problem is the following.
Consider:
reg z1 x1 x2
mat A=e(b)
reg z2 x1 x2
mat B=e(b)
I want to know how much of the difference between z1 and z2 is explained by
the difference in parameter x2 (a similar exercise like the Oaxaca
decomposition). Therefore I get the parameter of x2 from mat A:
mat X2 = A["y1","x2"]
The second step is to replace this scalar (or 1x1 matrix) into mat B. To do
so, I have to let Stata know the name of the variable--column--in mat B that
has to be replaced with mat X2. My first though would be:
mat B["y1","x2"] = X2["y1","x2"]
However this syntax is not valid. The second thing I tried--following your
advice--was to create a scalar from X2:
scalar x2 = X2[1,1]
Again I tried to replace mat B with scalar x2:
mat B["y1","x2"] = x2
And it didn't work either. To replace an element of a matrix you have to use
numbers of rows and columns a not the names of them (mat B[1,2] = x2 works
fine). Perhaps there is a strong reason for that but I cannot see it. In any
case the problem was solved by assigning a number to each element of the
independent variables:
local indv x1 x2
local i=1
foreach k in `indv' {
local `k' = `i'
local i=`i'+1
}
mat X`x1' = A[1,`x1']
mat B[1,`x1'] = X`x1'
I still wonder why Stata does not accept rows and columns names?
Best regards,
Rafa
----- Original Message -----
From: "Michael Blasnik" <[email protected]>
To: <[email protected]>
Sent: Tuesday, July 12, 2005 9:33 PM
Subject: st: Re: replacing a scalar from mat
> If you know that you are only extract one row and column, why not
> mat B = e(b)
> mat B1 = B["y1","eq1:x1"]
> scalar s=B1[1,1]
>
> Michael Blasnik
> [email protected]
>
> ----- Original Message -----
> From: "R.E. De Hoyos" <[email protected]>
> To: <[email protected]>
> Sent: Tuesday, July 12, 2005 12:48 PM
> Subject: st: replacing a scalar from mat
>
>
>>I am trying to get a scalar from a matrix (vector). Within a multiple
>>equations framework (-mlogit-) I want to get one of the parameters
>>estimated. After running -mlogit- I tried the following:
>>
>> mat B = e(b)
>> scalar s = B["y1","eq1:x1"]
>>
>> The following error appears:
>>
>> matrix operators that return matrices not allowed in this context
>> r(509);
>>
>> However if I try one of the following, there is no error reported:
>>
>> mat A = B["y1","eq1:x1"]
>> scalar s = B[1,1]
>>
>> Using the rows and columns names of the matrix is important since these
>> scalars will be replaced into a second vector of estimated parameters.
>>
>> Thanks,
>>
>> Rafa
>
> *
> * For searches and help try:
> * http://www.stata.com/support/faqs/res/findit.html
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
>
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
