par Denuit, Michel ;Mesfioui, Mhamed ;Trufin, Julien
Référence Methodology and Computing in Applied Probability, 21, 2, page (491-509)
Publication Publié, 2019-03-01
Article révisé par les pairs
Résumé : Association measures based on concordance, such as Kendall's tau, Somers' delta or Goodmanand Kruskal's gamma are often used to measure explained variations in regressionmodels for binary outcomes. As responses only assume values in {0,1}, these associationmeasures are constrained, which makes their interpretation more difficult as a relativelysmall value may in fact strongly support the fitted model. In this paper, we derive the setof attainable values for concordance-based association measures in this setting so that thecloseness to the best-possible fit can be properly assessed.