“…Probably more often, the transfer in the within-subjects design simply reduces the sizes of the differences (Chase and Calfee 1969, Nilsson et al 1975, Poulton 1975, figure 4, Poulton and Brown 1968, Stone and Peeke 1974. This may be offset by the greater power of the statistical comparisons, because the average differences between the people are moved from the error variance.…”