Adejumo, A. O. Heumann, Christian Toutenburg, Helge
Year of Publication:
Discussion paper // Sonderforschungsbereich 386 der Ludwig-Maximilians-Universität München 385
Agreement can be regarded as a special case of association and not the other way round. Virtually in all life or social science researches, subjects are being classified into categories by raters, interviewers or observers and both association and agreement measures can be obtained from the results of this researchers. The distinction between association and agreement for a given data is that, for two responses to be perfectly associated we require that we can predict the category of one response from the category of the other response, while for two response to agree, they must fall into the identical category. Which hence mean, once there is agreement between the two responses, association has already exist, however, strong association may exist between the two responses without any strong agreement. Many approaches have been proposed by various authors for measuring each of these measures. In this work, we present some up till date development on these measures statistics.
Agreement association raters kappa loglinear latent-class