Tests for Randomness?

Theoretical difficulties in the assessment of randomness are discussed here in the context of a specific coin tossing problem presented to 2930 schoolchildren by Green.1 It is argued that the problem, and indeed any task dependent on the detection of randomness, cannot be solved on any purely logical grounds and some methods offered for obtaining a solution are criticised. Finally, the data presented by Green, which show that significantly more children in each age range tested prefer the putatively incorrect answer to the putatively correct answer, are explained by reference to the representativeness heuristic described by Kahneman and Tversky. © 1987 Oxford University Press. All rights reserved.