October 1971 L E T T E R S T O T H E E D I T O R 1429
Staircase Simulations DAVIDA Y. TELLER, RICHARD M. ROSE, AND PAULA RENDLEMAN
Department of Psychology, University of Washington, Seattle, Washington 98105 (Received 14 June 1971)
INDEX HEADING: Vision.
We have read with interest the letter from Wales and Blake1
concerning bias in two-alternative forced-choice staircase techniques. We have carried out a similar but more extensive analysis of both forced-choice and yes-no staircase methods.2 Our results generally support Wales and Blake's conclusion that two-alternative forced-choice staircases result in estimates that are biased below the 0.75 point on the psychometric function. However, the extent of this bias depends upon the number of stimulus steps used to span the psychometric function. We have also completed a similar analysis3 of the PEST procedure proposed by Pollack4; contrary to Wales and Blake's findings, we find that it suffers from the same kind of bias as do two-alternative staircase procedures.
Simulations of additional staircase algorithms are in progress, and we would welcome descriptions of algorithms that are in use in other laboratories.
1 R. Wales and R. Blake, J. Opt. Soc. Am. 60, 284 (1970). 2 R. M. Rose, D. Y. Teller, and P. Rendleman, Percept. Psychophys. 8, 199 (1970). 3 P. Rendleman, R. M. Rose, and D. Y. Teller, Percept. Psychophys. 9 208 (1971). 4 I. Pollack, Percept. Psychophys. 3, 285 (1968).