A general recursive procedure for analysis of variance

Characterization of the multivariate Gauss-Markoff model with singular covariance matrix and missing values

“…The following methods of missing data played general role: 1) Yates' [18] iteration method, 2) Wilkinson's one [17], 3) Biggers one [4] for experimental designs, 4) covariance analysis of Bartlett [3], 5) Hartley's and Hocking's [7] method of maximum likelihood and many others. Yates's technique is connected with minimalizing the error sum of squares.…”

Nonorthogonal Analysis of Variance Using Iterative Improvement and Balanced Residuals

“…Alternatively, if we applied orthogonalization methods as summarized in [l], such as GramSchmidt or Householder transformations, then we would also need p ( p + 1)/2 storage locations for the resultant orthogonal transformation of Xo. Wilkinson's recursive procedure [9] could be applied to nonorthogonal models. However, in this case, his "adaptive mode of analysis" would be required in determining the "apparent efficiency factors."…”

“…minimal amount of computer storage. The method bears some semblance t o Wilkinson's procedure [9] and Hartley's algorithm for the balanced case [2] in that it utilizes a residual operator. Iterative formulas are developed which involve the computation of balanced analysis of variance estimates and residuals in solving the relevant normal equations and conducting tests of hypotheses.…”

The non-orthogonal analysis of variance

“…Three of these, Wilkinson (1970), Fowlkes (1969) and Bradley (1968) will be examined in some detail.…”

“…A general computing algorithm must include answers to these problems. Of particular interest is the algorithm due to Wilkinson (1970).…”