Ana,
It's not really clear what you are trying to do. From your
description, you aren't estimating anything. The code
a <- c(1, 0.6, 0.8)
x <- rnorm(2)
y <- crossprod(c(1, x), a)
generates 'y'; but nothing is estimated. 'y' is going to have
the
same variance regardless of how many observations you generate, since
the variance is coming from x1 and x2, not 'a'.
Your question sounds a little like a homework problem. If it's not,
please explain what you're trying to do.
--Gray
2010/7/27 Ana De Barros <belindadebarros at
gmail.com>:> No, but thanks anyway...
>
>
> On 27/07/10 16:38, "Gray Calhoun" <gray.calhoun at
gmail.com> wrote:
>
>> Hi Ana,
>> ? Does the "predict" function do what you want? ?Type in
?predict.lm
>> --Gray
>>
>> On 7/27/10, Ana De Barros <belindadebarros at gmail.com> wrote:
>>> Hi,
>>>
>>> Is there any way to estimate a DEPENDENT variable through a GLM/LM
model?
>>>
>>> Suppose I have the linear model: y=a0+a1*x1+a2*x2 (a0=1, a1=0.6,
a2=0.8,
>>> x1~N(1,1), x2~N(0,1)).
>>> The alphas and the auxiliary variables are given and I have to
estimate y.
>>> The point is if I estimate it, let? say algebraically, I get high
variances
>>> that do not decrease as sample sizes increases... Is the any other
way to do
>>> this?... It is not compulsory to use these alphas but my Y is
unknown...
>>>
>>> Any ideas?...
>>> Thanks,
>>> Ana
>>>
>>> [[alternative HTML version deleted]]
>>>
>>>
>>
>
>
>
--
Gray Calhoun
Assistant Professor of Economics, Iowa State University
http://www.econ.iastate.edu/~gcalhoun/