A note on the interpretation of Springer's simulation studies of the Kiefer-Wolfowitz process
MARK A. JOHNSON · Biometrika · 1970
It is shown that the pronounced small sample bias that Springer (1969) observes in his simulationstudies of a modification of the Kiefer-Wolfowitz process can be attributed to his choice of the sequence{am}. His sequence, defined by a m+1 = 1/2am, violates the usual assumption that Σm = ∞