[R] Bivariate linear regression

[Prof Brian Ripley]

On Thu, 11 Dec 2003, dauphasuser wrote:

> I have measurements with uncorrelated uncertainties on both axes. I 
> would like to get the uncertainties on the intercept and the slope of 
> the weighted linear regression model taking into account the 
> uncertainties of the measurements. Is these any way to do that in R?

Yes, but

- this is not bivariate linear regression and
- either weighted linear regression is biased.

Perhaps the best approach is to fit a functional relationship model. I
would need to know a bit more -- you say `weighted' so do the standard
errors (`uncertainties') vary by point, for example?

Perhaps rather than (mis)use terminology you could explain in 
non-technical language the problem you are trying to solve?

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595