John N. Towse
Derek Neil
Behavior Research Methods, Instruments & Computers, 1998, 30(4), 583-591.
This paper considers the different methods that have been developed
to quantify random generation behavior, and incorporates these measurement
scales into a Windows 95 computer program called RgCalc. RgCalc analyzes
the quality of human attempts at random generation, and can provide computer-generated,
pseudo-random sequences for comparison. The different algorithms for the
evaluation of a dataset are detailed, and an outline of the program is
described. Performance measures are available for assessing various aspects
of the response distribution, the sequencing of pairs, the ordinal relationships
between sets of items, and the tendency to repeat alternatives over different
lengths. The program is designed to be appropriate for the analysis of
most types of random generation situation employed in the psychological
literature, and a factor analysis is used to illustrate the multiple dimensions
underlying human randomization processes.