Empirical Likelihood Inference Of Error Variance In Linear Regression Models With Fixed Design Case

In this paper,the complete data method is adopted to fill the missing data,using the empir-ical likelihood method,the response variables to the fixed design cases for lack of random error variance of the linear regression model to estimate is studied,has been based on the empirical likelihood method and the traditional method of residual sum of squares estimator,and presents the asymptotic variance of these two estimators,and proves that these two kinds of distribution of the estimator is asymptotically normal distribution.Through simulation,using the empirical likelihood method the asymptotic variance of the estimator is less than using traditional residual sum of squares method the asymptotic variance of the estimator.In addition,this paper discusses the construction of the confidence interval of error variance in the linear regression model without and without additional information,and proves that the limit distribution of the empirical likelihood ratio statistics based on the empirical likelihood method is chi-square distribution.