[R] OFF topic testing for positive coeffs
vmuggeo at dssm.unipa.it
Wed Dec 17 11:13:41 CET 2008
This is off-topic,
however I hope someone can give me useful suggestion..
Given the regression model
y = b0 + b1*x + e
I am interested in testing for positive coeffs, namely
H0: b0>0 AND b1>0
H1: b0,b1 unconstrained
It is simple to estimate the model under H0 and H1 (there are several
suggestions on the Rlist about estimation but nothing about testing..)
perform a likelihood ratio test by comparing the logLik under the
constrained and the unconstrained models, however I do not know how many
degrees of freedom..
Model under H0 uses two df, however it reasonable to believe that the
real dimension is <=2..
Is there anyone which can give me any advices or suggest me references?
Vito M.R. Muggeo
Dip.to Sc Statist e Matem `Vianelli'
Università di Palermo
viale delle Scienze, edificio 13
90128 Palermo - ITALY
tel: 091 6626240
fax: 091 485726/485612
More information about the R-help