Re: comments on Oct 12 manuscript

From: Alexander Conley (AJConley@lbl.gov)
Date: Mon Nov 08 2004 - 14:15:15 PST

  • Next message: Alexander Conley: "Difference between Knop and Barris samples"

    Okay -- next item. Resids versus subsample.

    At low-z, the mean resid and error are:

       Hamuy: 0.02163 +- 0.0466 (14 SNe)
       Riess: 0.05203 +- 0.0654 (6 SNe)
       Jha: -0.06338 +- 0.1169 (6 SNe)
       Other: -0.01193 +- 0.1073 (6 SNe)

    The means are not significantly different, but the Jha and Other
    samples seem to
      have a lot more scatter (as evidenced by their larger scatter). These
    HAVE been
      divided by sqrt(N).

    By Kcorr type,
       R->B : 0.0027 +- 0.04844 (21 SNe)
       I->B : 0.04816 +- 0.115619 (3 SNe)

    As you can see, the problem is that there are so few I->B available
    that you can't
    say much.

    On Knop vs. Barris, I have done plenty of work. I refer you to the end
    of section 8.3
    (around page 77) in the methods document, as well as Fig 28 and 29. I
    went further
    here and actually did some MC simulations to try and figure out how
    significant the
    difference was. The difference is 0.31 mag ignoring extinction, and
    this shrinks
    to 0.16 if you extinction correct. If you RMS/sqrt(N) you see that
    this is
    about 2 sigma before and 1 sigma after. Monte-Carlo suggests that the
    difference
    is actually a bit more significant -- 3 sigma and 1.6 sigma,
    respectively.

    Alex



    This archive was generated by hypermail 2.1.4 : Mon Nov 08 2004 - 14:15:14 PST