Prediction intervals in the simple linear regression model with balanced nested error structure

Abstract

Prediction intervals are proposed for a new observation corresponding to a given level of the independent variable in the simple linear regression model with nested error structure. Ordinary least square estimators and best linear unbiased estimators of the regression coefficients in the model are used to predict a response variable. The prediction intervals are derived by using distributional properties of regression coefficient estimators and variance components in the model. The prediction intervals are compared using computer simulation and a numerical example is provided.