يك خطمشي در مدل بندي توافق داده هاي ترتيبي

Background & Objective: For determining agreement between ordered scale data, there are several methods in statistics, like computing indexes of agreement and modeling agreement. Method: In this article we focus on modeling agreement by introducing new square scores association model and reviewing three other available models (diagonal parameter model, uniform association model and agreement plus uniform association model) in detail. Then we fit these models to the ordinal data of IOTN (Index of Orthodontic Treatment Need) and compare the results of the models. 2 dentists have measured this index. Results: We saw that uniform association model do not fit the data (p=0.049) but diagonal parameter model (p=0.133), square scores association model (p=0.163) and agreement plus uniform association model (p=0.684) fit the data well. Discussion: Between models which had only one parameter more than independence model, square scores association model was the best but because agreement plus uniform association model had one parameter more than all other models it fitted better than all of them. Finally we saw that according these two models dentist 1 tended to rate the samples higher than dentist 2.