Re: median regression vs linear regression




It depends on the nature of the skew. For many skewed data sets a
transformation of the DV is a sensible approach. For example you could
model the logarithm of the data. this changes the model, but often the
resulting model is more plausible than that for the untransformed data.

Thom

.



Relevant Pages

  • Re: Natural selection is a false term - says Darwin in Origin Species
    ... And please stop this common ancestor nonsense - your intent is the ... transformation of a banana tree into a monkey into a human. ... IOW there are no laws of nature. ...
    (talk.origins)
  • Re: FTL and references to works
    ... The above argument depends only on the validity of the Lorentz ... A Lorentz transformation is a transformation, ... What you can, possibly, falsify is the claim that all laws of nature ... If the laws are nature are Lorentz covariant, ...
    (sci.physics.relativity)
  • Re: Obama and abortion
    ... "Nature does not know extinction, all it knows is transformation." ... -- Werhner von Braun ...
    (rec.gambling.poker)