Analysis of variance for strip plot design with missing values: bias correction of the mean squares

Abstract

Missing values in strip plot design cause an unbalanced design. As a result, several problems occur in analyzing the data. This paper presents the estimation of two and three missing values from strip-plot design with the fixed model using the Yates approach. However, the estimation toward the missing values in strip-plot design causes bias on the treatment mean squares which is shown by its expected value. To omit the bias caused by the estimated values, a procedure of Satterthwaite-Cochran approximation is applied. The procedure corrects the bias and gives an adjusted analysis of variance for the hypothesis test.