Kajian metode pendugaan pada model regresi dengan peubah penjelas bersifat acak

Abstract

Regression analysis is a statistical method to evaluate the relationship between a variable with other variables. Model I regression analysis describes the one-way relationship between variable X and variable Y with the value of X as fixed variable or measured without error. Model I regression analysis uses Ordinary Least Square (OLS) as an estimation method. Model II regression analysis is a regression model with predictors variable that becomes a random variable. There are two estimation methods of model II regression analysis, ordinary least product regression (Model IIA) and major axis regression (Model IIB). The convenience of using model I regression analysis with OLS estimation method causes many researchers to use this model as a model regression analysis with random predictor variables. The purpose of this study is to compare model I regression analysis (OLS estimation method) and model II regression analysis (ordinary least product estimation method and major axis regression estimation method). The data used in this research are simulation data where response variable (Y) and predictor variable (X) are random. The result of this study showed for predictor variable is random, ordinary least product regression is the best estimation method compared to the other methods because it produces the smallest value of bias and the smallest value of mean square error. In this condition, OLS can be used for the estimate parameter of simple linier regression model if the correlation value between predictor variable (X) and response variable (Y) is high (r ≥ 0.9).