Re: Extinction prior bias

From: Robert A. Knop Jr. (robert.a.knop@vanderbilt.edu)
Date: Fri May 23 2003 - 06:41:22 PDT

  • Next message: Reynald Pain: "Re: Bias of our low-extinction method"

    On Thu, May 22, 2003 at 04:36:14PM -0700, Greg Aldering wrote:
    >
    > Hi Rob,
    >
    > With the current fits, can you tell us how much of a bias was
    > introduced by using the Riess prior? You say in the figure
    > caption that a bias is apparent, so you must know what it is.

    It's not very big for the current fits-- it's huge for the P99 fit. The
    offset between our fit 3 and the version which has been run through the
    prior is 0.03 for OM in a flat universe. For Fit 1 (using only P99
    SNe), it's qualitatively similar to what is shown in the P99 figure,
    0.07 in OM for a flat universe.

    The other thing about this prior is that the way I've done it is
    completely wrong, although I've been assured that what I'm doing is
    exactly what the other team is doing. The E(B-V) error bars are *very*
    non-Gaussian after the prior has been applied, yet we still use a
    chisquare procedure with a gaussian that has the same mean and first
    moment as the probability distribution with the prior applied. Those
    are very different probability distributions. I have no idea how much
    doing it this way affects the results, but it's an outstanding question
    in my mind.

    -Rob

    -- 
    --Prof. Robert Knop
      Department of Physics & Astronomy, Vanderbilt University
      robert.a.knop@vanderbilt.edu
    


    This archive was generated by hypermail 2.1.4 : Fri May 23 2003 - 06:41:24 PDT