A popular method for analyzing unreplicated factorials was proposed in 1989 by Russell Lenth. This paper proposes and illustrates the use of p-values, estimated by simulation, for Lenth t-statistics. This approach exhibits versatility in its use of p-values for both orthogonal and non-orthogonal designs, whereas tabulated critical values are restricted to uncorrelated estimates. For cases where there is limited replication, computing t-statistics and corresponding p-values should be done using an estimator that combines pure error mean square with modified Lenth pseudo-standard error.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados