This study examined characteristics of 29 pupils (selected from an initial sample of 69 primary school children) whose performance on an arithmetic task deteriorated after failure. On the basis of their responses to a "mitigating circumstance" which could explain failure without implicating low ability as the cause (a description of the task as "very difficult"), subjects were classified as either learned helpless or motivated to protect self-worth. Both of these groups had lower self-concepts than the rest of the initial sample. Attribution retraining resulted in increased effort attributions and decreased inability attributions in the "learned helpless" group, and inoculated these subjects to the experience of failure. In the "self-worth group" there was an increase in effort attributions but no change in ability ratings or performance after failure, following training. Effort Ability Composite Scores' Internality Stability ~ M SD M SD M SD M SD M SD M S D M SD LH sw Boys Girls (N = 8) (N=IO) -1.50 1.00 AcKNowLmcMENTs.--The author is grateful for the assistance of Iain Montgomery and John Davidson. Correspondence and requests for reprints should be sent to Dr. M. L. Craske,

Summary. This study aimed to improve the persistence of learned helpless children through reattribution training accomplished via observational learning. In addition, the relationships of effort attributions and intelligence to persistence behaviour were examined. In the initial phase, 65 Grade 5 and 6 subjects were presented with unsolvable puzzle‐completion problems to gauge persistence levels; they also completed the Intellectual Achievement Responsibility Scale (IARS), and Raven's Progressive Matrices. IARS scores which reflected neglect of the effort factor in academic outcome correlated highly with persistence; intelligence scores were not related to persistence. In the second phase, the 32 least persistent subjects were assigned to control or vicarious reattribution training groups. The treatment was associated with increased persistence in female, but not male, subjects. The results are discussed in terms of exchange theory analysis and self‐worth theory.