Modelling Generalized Linear (Loglinear) Models for Raters Agreement Measure

Raters have become an important source of measuring errors in recent time researches in nearly all the fields of human endeavour. Kappa-like statistics have been used to great advantages in almost every area of life and social sciences. However, these Kappa-like statistics have some shortcomings. In the book, some selected loglinear models are considered for raters agreement using generalized linear models (GLMs) approaches for complete and missing ratings. We observed that modelling the structure of agreement with some selected loglinear models is more appropriate for raters agreement than any of the Kappa-like statistics. Two methods (Likelihood and Data-based approaches) proposed for estimation of parameter for missing ratings in this book can also be viewed as means of multiple imputations.