An Alternative to Eisenhart's Model II and Mixed Model in the Case of Negative Variance Estimates

Abstract An alternative model useful for addressing the problem of negative variance component estimators is proposed. Specifically, it is possible to view certain of the parameters as covariances rather than as variances. Thus, the problems of the negative variance component estimator may be resolved by using a different model. Distributional results relating to both estimation and tests of hypotheses are also given.