Roberts, E.A.Raison, J.M.2014-10-012014-10-011983Biometrics, 1983; 39(4):1097-11050006-341X1541-0420http://hdl.handle.net/2440/85735An analysis of the moisture content of 10 segments of soil cores in a designed experiment is presented. Difficulties arise in the analysis of such data, as the observations at successive depths are usually correlated and have different variances. From the data, treatment contrasts at each depth were calculated, and a polynomial regression analysis of these contrasts was carried out. Polynomials up to third degree adequately describe the variation, with depth, in the treatment contrasts. The variance of all polynomials is estimated more efficiently by including higher-order coefficients as covariates and this is particularly so for zero-degree polynomials for which the residual variance is reduced from 3.86 to 1.57 by using the first-order coefficient as a covariate, and to 1.13 by using the first- and ninth-order coefficients.enCopyright status unknownJournal article003000900010.2307/2531345A1983SF544000292-s2.0-002091866284558