Random Number Generation by Normal Subjects

Abstract

Human facility for generating random numbers has been proposed as a brief simple test of psychological functioning (5). Evans (2) reported a standardised procedure and formulae for obtaining an index.of random number generation. A sample of 50 male and 71 female Christchurch residents was interviewed at home. In terms of age and socioeconomic status (1) the sample was representative of the general population. Subjects were instructed in the concept of randomness and then attempted to generate random numbers, each subject generating 100 numbers between 1 and 10 at a paced rate of l/sec. An index, based on the frequency with which each number was succeeded by another, was calculated for each subject. Index values can range from zero (indicating perfect randomness) to one (when each number is completely predictable from its predecessor, e.g., 7 might be always followed by 3, etc.). The mean index was 0.335 (SD = 0.056). This is somewhat higher than that found with volunteer college students (0.317) but lower than that found with some psychiatric inpatient groups (3). No significant (# < 0.05) effect of age, age on leaving school, history of concussion, or of serious concussion (defined by hospitalization for 24 hr. or more for concussion) was found on the index. There was a significant correlation (0.33) between the index and number of errors on a word-reading test known to correlate highly with intelligence (4). Facilicy at generating random numbers is then related .to general intelligence, a feature which should be considered when the index is used clinically. There was also a suggestive (P < 0.10) positive correlation bemeen the index and socioeconomic status which, however, disappeared when perform