This paper presents a generalization of the kappa coefficient for multiple observers and incomplete designs. This generalization involves ordinal categorical data and includes weights which permit pondering the severity of disagreement. A generalization for incomplete designs of the kappa coefficient based on explicit definitions of agreement is also proposed. Both generalizations are illustrated with data from a medical diagnosis pilot study.

In the framework of standard model of asymptotic statistics we introduce a global information in the statistical experiment about the occurrence of the true parameter in a given set. Basic properties of this information are established, including relations to the Kullback and Fisher information. Its applicability in point estimation and testing statistical hypotheses is demonstrated.

The concept of global statistical information in the classical statistical experiment with independent exponentially distributed samples is investigated. Explicit formulas are evaluated for common exponential families. It is shown that the generalized likelihood ratio test procedure of model selection can be replaced by a generalized information procedure. Simulations in a classical regression model are used to compare this procedure with that based on the Akaike criterion.