Arbeitspapier
Conditional symmetry model as a better alternative to symmetry model for rater agreement measure
Abstract In almost all life or social science researches, subjects are classified into categories by raters, interviewers or observers. Many approaches have been proposed by various authors for analyzing the data or the results obtained from these raters. Symmetry and conditional symmetry models are models designed for square tables like the one arising from the raters results. Conditional symmetry model which possessed an extra parameter for the off-diagonal cells is a special case to symmetry. In this research work, we examined the effect of the extra parameter introduced by conditional symmetry model over that of symmetry on structure of agreement as well as their fittings. Generalized linear model (GLM) approach was used to model the loglinear model forms of these models with empirical examples. We observed that conditional symmetry based on it extra parameter gave a tremendous improvement to the significant level of the test statistics over that of its symmetry model counterpart, hence conditional symmetry model is better for raters agreement modelling which require symmetric table.
- Language
-
Englisch
- Bibliographic citation
-
Series: Discussion Paper ; No. 386
- Subject
-
Agreement
symmetry
conditional symmetry
raters
loglinear
- Event
-
Geistige Schöpfung
- (who)
-
Adejumo, A. O.
Heumann, Christian
Toutenburg, Helge
- Event
-
Veröffentlichung
- (who)
-
Ludwig-Maximilians-Universität München, Sonderforschungsbereich 386 - Statistische Analyse diskreter Strukturen
- (where)
-
München
- (when)
-
2004
- DOI
-
doi:10.5282/ubm/epub.1756
- Handle
- URN
-
urn:nbn:de:bvb:19-epub-1756-3
- Last update
-
10.03.2025, 11:46 AM CET
Data provider
ZBW - Deutsche Zentralbibliothek für Wirtschaftswissenschaften - Leibniz-Informationszentrum Wirtschaft. If you have any questions about the object, please contact the data provider.
Object type
- Arbeitspapier
Associated
- Adejumo, A. O.
- Heumann, Christian
- Toutenburg, Helge
- Ludwig-Maximilians-Universität München, Sonderforschungsbereich 386 - Statistische Analyse diskreter Strukturen
Time of origin
- 2004