iSixSigma

Binary Regression – Concordant

Six Sigma – iSixSigma › Forums › Old Forums › General › Binary Regression – Concordant

Viewing 3 posts - 1 through 3 (of 3 total)
  • Author
    Posts
  • December 20, 2007 at 2:10 pm #48963

    AB
    ★ 10 Years ★
    Participant

    I recently completed a binary regression with 5 significant variables and a concordant value of 79.6%.  I’m looking for assistance on how to interpret the concordant and how to describe it to a non-statistical audience.  Thanks for any assistance.

    0
    December 20, 2007 at 3:09 pm #166389

    rolly
    ★ 10 Years ★
    Member

    I’ll answer your second question.  Presenting stats to a non-statistical audience would require you to use analogies.  Be creative.  For instance, if you want to describe the number 23, get a picture of Michael Jordan, and so on.  In the end, stick tough to the issue.  Presenting stats is presenting stats with corresponding human touch.

    0
    December 21, 2007 at 1:33 pm #166447

    Robert Butler
    ★ 20 Years ★
    Participant

     The concordance index is a measure of of the probability of agreement between what the model predicts and what was observed.  A concordance index of 50% means that the predictions are no better than flipping a coin.  A concordance index of 79.6% would say what you saw and what the model says you saw will agree about 80% of the time.
      So, for a non-technical audience you could say that 20% of the time your model won’t agree with actual measurements. The problem here is you don’t have anything for contrast.  If the current best mode has a 30% disagreement then you have an improvement but if the current model is 10% then you need to go back to the drawing board.

    0
  • Author
    Posts
Viewing 3 posts - 1 through 3 (of 3 total)

The forum ‘General’ is closed to new topics and replies.