Biometrika 1976 63(2):412; doi:10.1093/biomet/63.2.412-a
© 1976 by Biometrika Trust
CORRECTIONS AND AMENDMENTS |
D-optimum and quasi-D-optimum second-order designs on a cube
I should like to point out the error in construction of minimal second-order symmetric set I4 42 and absolutely minimal design
4 in my paper.
The proper set I4 41 may be constructed as a combination of four half-replicates of the 24 factorials with the defining relations, x2x3x4 = I, x1x3x4 = I, x1x2x4 = I and x1x2x3 = I correspondingly for the subset B4 and of four duplicated half-replicates of the 23 factorials with the same but negative defining relations for the subsets B41(i); the number of different points in B4 is 15 and in I4, 4l is 31, although the number of trials is 64, against corresponding numbers 36 and 40 in Table 1 of the paper.
For the construction of absolutely minimal design
4 we need to take some modification of set I4 41 with 34 points and we obtain in the same way as in the paper a D-optimum design
4 with the number of points 37 instead of 45 presented in Table 2.