Re: [R] Possible bug in rlm

Have just checked with R 3.2.0 and MASS 7.3-40 and there still appears to be a problem/strangeness at around k=2.5 On 23 April 2015 at 14:09, Francis Bursa francis.bu...@quantics.co.uk wrote: Dear all, I believe I have found a bug in rlm in the MASS package. Specifically, the scale

[R] Possible bug in rlm

Dear all, I believe I have found a bug in rlm in the MASS package. Specifically, the scale estimate can be wrong when there are no outliers. The following code snippet is an example: dose - c(0,1,2,0,1,2) response - c(0.659,1.633,3.621,1.803,3.093,4.424) line - c(1,1,1,2,2,2) k2 -