C. Hyten et Gj. Madden, THE SCALLOP IN HUMAN FIXED-INTERVAL RESEARCH - A REVIEW OF PROBLEMS WITH DATA DESCRIPTION, The Psychological record, 43(3), 1993, pp. 471-500
Early research studying human performance on schedules of reinforcemen
t indicated that scalloped response patterns (initial pausing followed
by accelerating responding) were prevalent in fixed-interval schedule
s, but later research has shown that scalloping is rare in humans, or
it occurs only under limited conditions. A review of this research rev
ealed several problems with the description of response patterns. Firs
t, there was considerable variability in what was identified as a scal
lop. Second, the prevalence of scalloping was often exaggerated by the
manner in which authors summarized cumulative record data. Methods of
response pattern analysis including the Mathematical Index of Curvatu
re, interresponse time microanalysis, and visual inspection of cumulat
ive records are discussed. A method of interval-by-interval classifica
tion of response patterns in cumulative records is suggested as a tech
nique to improve descriptions of fixed-interval patterning. Cumulative
record data from a published study on schedule performance with infan
ts are reanalyzed using the method to illustrate its application.