Quantile regression with some parameters fixed across tau..
I would like to fit the following model with quantile regression:
y ~ alpha + beta
where both alpha and beta are factors. The conceptual model I have in my
head is that alpha is a constant set of values, that should be independent
of the quantile, tau and that all of the variability arises due to beta. If
I just fit the model using the quantreg package like so:
mdl <- rq( y ~ alpha +beta, data, tau=c(0.25,0.5,0.75))
..the values of alpha that I get vary according to the tau. Is there anyway
to force alpha to be the same across all taus?