|
Abstract:
|
In many sports competitions, participants are evaluated based on the counts of their successes; often, a single-count variable does not suffice. Thus, the assumptions of traditional approaches based on linear combinations of counts by category often are not justified on theoretical grounds. Establishing concept validity through empirical validation, as often is suggested, is problematic, in part because the choice of the criterion to validate against can shift the focus from assessing an athlete's ability to assessing context-dependent performance. This presentation proposes the use of u-statistics for scoring objects (e.g., players, teams, countries) based on several count variables. In particular, these variables are allowed to be graded by importance or relevance.
|
