Performance of balanced two-stage empirical predictors of realized cluster latent values from finite populations: A simulation study,” Computational Statistics and Data Analysis (2007), doi: 10.1016/j.csda.2007.07.013

Prediction of random effects in clustered finite populations is important in many practical problems. Predictors are usually based on the classical mixed effects model (assuming a conceptual infinite population), on the finite super-population model proposed by Scott and Smith (1969, JASA, 64: 830-840) or on the random permutation model recently proposed by Stanek and Singer (2004, JASA, 99:1119-1130). When variances are known and within cluster variances are equal, the random permutation model based predictor has smaller mean squared error than corresponding predictors obtained under either the mixed effects or Scott and Smith’s models. As population variances are rarely known, we propose method of moment estimators to obtain empirical predictors an