Re: Set Repeatability

ravenous.wolves_at_gmail.com
Date: 02/02/05

  • Next message: Ray Koopman: "Re: random numbers from a lognormal distribution" 
    Date: 1 Feb 2005 18:28:16 -0800

    Thanks,

    A few questions..
    1. FF' == F x INVERSE(F) ?
    2. "average off-diagonal covariance" = Average of all FF'[i,j] elements
    in FF' where i != j ?
    3. Where does the "average variance" come from? Is this
    SIGMA(i,j)[ ABS(F[i,j] - MEAN(F)) ] ?

    Thanks,
    -Ken

    Ray Koopman wrote:
    > Organize the data into a matrix F in which f[i,j] = 1 or 0 according
    > as run i did or did not flag region j as defective. Then think of F
    as
    > containing "factor loadings", so that FF' gives a "covariance"
    matrix.
    > Then the ratio of the average off-diagonal "covariance" over the
    > average "variance" is an index of repeatability. (I'm sure there are
    > other rationales that will generate the same index. I'm just letting
    > my psychometric roots show.)
    >
    > For the 6 examples, the repeatability indices are
    >
    > [1 0 .33 .90 .33 .24]
    >

  • Next message: Ray Koopman: "Re: random numbers from a lognormal distribution"
    • Quantcast