Özet:

Studies on the analysis and interpretation of data in the repeated measurement structure have shown great progress in the past years, and strong methods have been developed in this sense. In this study, three models were created in which the time variable was incorporated differently, taking advantage of the specific conditions of the linear mixed model. These models are created in the form of the random cutting point and curve model (Model 1), the random cutting point model (Model 2) and the random cutting point and curve model (Model 3) in which the time variable is included in the model both continuously and categorically. In the determination of repeated measurement-intercovary structure, the structures of Compound Symetry (CS), Unstructured (UN)) and First Order Autoregressive (AR(1)) were applied, and with these structures the most likely (ML), the limited most likely (REML) and the Minimum Variant Quadratic Non-Variant Predictor (MIVQUE) methods were used. The best method of prediction for the data set and the choice of the quarry structure are assessed by the conformity criteria of AIC and BIC. The study material was created by the Serum testosterone concentration values of the 33 main Norduz male cousins. As a result, the best conformity to the data set has been determined by the United Nations quarrying structure, which takes into account a heterogeneous structure along with the ML prediction method in both three models.  Keywords: Repeated measurement, methods of prediction, quarry structure

Anahtar Kelimeler: