[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Psychophysical measurement technique - PEST?



Having successfully used the up/down technique to get estimates of
psychometric functions for different % correct levels in various
psychoacoustic tasks (see Levitt, 1970, Journal of the Acoustical Sociery
of America, V49 #2, pp467), it was suggested I look into PEST (Parameter
Estimation by Sequentual Testing) as a more efficient adaptive procedure.
The original reference is Taylor & Creelman, 1967, JASA, V41 #4, pp782f.

My (current) application? I want to find difference limens at 75% correct
in psychiatric patient populations and so want to do the measurement with
the least number of trials possible.

While I think I understand the basic concepts of the 1967 PEST paper, I
do *not* understand how to determine the parameters of the algorithm for
a given "probability of positive response" (P sub t in the paper).  I
also don't fully understand the meaning of W.  The paper references a
book by Wald (1947, "Sequential Analysis") which I admit not to have seen
yet. Would this shed some illumination?

So, my questions: Can someone offer any of 1) a pointer to a more recent
paper describing PEST and its theoretic efficiencies (and/or
implementations), 2) code fragments illustrating an actual computer
implementation, or 3) an intuitive explanation of the variables and their
effects in the PEST procedure?

Many thanks in advance,
Peter "maybe I need RAID?" Marvit

: Peter Marvit <marvit@cattell.psych.upenn.edu>                            :
: Psychology Department,  University of Pennsylvania                       :
: 3815 Walnut St., Philadelphia, PA  19104  w:215/898-9208  h:215/387-6433 :