Multilevel regression models and application of multilevel regression analysis

Multilevel regression models, discovered over the past decade, have widely accepted in many different fields since it is considerably useful way to analyze nested or hierarchical data sets. This method can easily deal with the levels of the data and the possible correlation problems which usually occur as a result of the hierarchical data sets. Ignoring nested structured and correlation in traditional statistical analysis lead to biased parameter estimates and incorrect standard error estimation. Thus, using multilevel regression models for hierarchical data structure is kind of obligation for researchers. In agriculture field, the using of the multilevel regression model is not common; therefore, the aim of this study is to introduce this method to this area with applying multilevel modeling technique to milk yield data which obtained from England Holstein Friesian cattle. The average milk yield is considered as a dependent variable, and ordinal least square regression models (single level model) and multilevel regression models were used as a modeling method. Besides test day milk yield records is considered and a multilevel approach for repeated measurements were used. As a result of this study, the level structure of the data had to be considered and multilevel regression model is better fit the data.KEY WORDS: Multilevel regression models, intra-class corelation, milk yield, test day records