Good test--retest reliability for standard and advanced false-belief tasks across a wide range of abilities

Although tests of young children's understanding of mind have had a remarkable impact upon developmental and clinical psychological research over the past 20 years, very little is known about their reliability. Indeed, the only existing study of test-retest reliability suggests unacceptably poor results for first-order false-belief tasks (Mayes, Klin, Tercyak, Cicchetti, & Cohen, 1996), although this may in part reflect the nonstandard (video-based) procedures adopted by these authors. The present study had four major aims. The first was to re-examine the reliability of false-belief tasks, using more standard (puppet and storybook) procedures. The second was to assess whether the test-retest reliability of false-belief task performance is equivalent for children of contrasting ability levels. The third aim was to explore whether adopting an aggregate approach improves the reliability with which children's early mental-state awareness can be measured. The fourth aim was to examine for the first time the test-retest reliability of children's performances on more advanced theory-of-mind tasks. Our results suggest that most standard and advanced false-belief tasks do in fact show good test-retest reliability and internal consistency, with very strong test-retest correlations between aggregate scores for children of all levels of ability.