Skip to content
 

Type S error rates for classical and Bayesian single and multiple comparison procedures

Type S error: When your estimate is the wrong sign, compared to the true value of the parameter

Type M error: When the magnitude of your estimate is far off, compared to the true value of the parameter

More here.

2 Comments

  1. http://models.street says:

    Page 5 you refer to some kind of parameter Beta but there is no Beta in the equations.

  2. K? O'Rourke says:

    Very clear crisp start and very scary Fig 3.

    When I was reading it, I was a bit uncomfortable with the prior assumptions matching the true distribution of parameter values – simulations under mis-matches as well seemed as important if not more important.

    And then I was sent this Gustafson and Greenland paper on an unrelated manner http://arxiv.org/pdf/1010.0306

    Different setting but similar findings – Bayes usually beats frequentist on "thoughtful" evaluations of coverage.

    Also a bit concerned about taking sigma as known (other than as a start). I once almost got burned very badly when assuming tau was 0 and thinking it did not matter much if I took sigma as unknown versus known (and set equal to its reported estimate).

    Believe it will be a problem for tau being taken as equal to something less than the true tau leading to what Meng called the non-negligibility of R http://arxiv.org/pdf/1010.0810

    K?