I have 6 series of observed values
(y,y1,y2,y3,y4,x1,x2) that can be estimated as:
y=y1+y2
y1=a1+b1x1+c1x2
y2=a2+b2x1+c2x2
or, alternatively
y=y3+y4
y3=a3+b3x1+c3x2
y4=a4+b4x1+c4x2
I'm particularly interested in the relationships
b1/(b1+b2) and c1/(c1+c2), and its comparision with
b3/(b3+b4) and c3/(c3+c4), respectively.
How can I solve my problem? is it two independent
regressions, a two stage regression, a seemingly
[un]related regression or a three stage regression?.
By the way, my sample is very small.
Any -easy to read- reference would be also
appreciated.
Cave.
__________________________________
Do you Yahoo!?
Yahoo! Tax Center - File online by April 15th
http://taxes.yahoo.com/filing.html
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
