Hierarchical Task Analysis (HTA) is used particularly in the context of instructional development. This paper involves two exploratory studies concerning the difficulties of those learning to perform HTA (Study 1) and how these might be overcome (Study 2). In Study 1 seventeen students were provided with declarative training in the major features of HTA and were then asked to analyse the task of making a cup of tea (task 1) or of painting a door (task 2). HTAs were analysed in terms of five HTA criteria (hierarchical representation, logical decomposition rule, logical equivalence, specification of plans and the P x C rule) and four other error categories (task boundaries incorrect, cognitive goals omitted, operations described as activities rather than goals, and lack of versatility of the analysis in terms of encompassing task variation). Errors occurred with respect to all HTA criteria and other error categories suggesting that carrying out HTA is itself a complex cognitive task. This together with an analysis of questionnaire responses concerning self-reported difficulties and strategies suggested that the tendency to use an action-oriented representation of the task being analysed might be one cause of poor performance. Study 2 investigated the effectiveness of three instructional conditions at improving analysts' performance at HTA: procedure training which specified eight main goals in carrying out HTA, criteria training which involved understanding and practice at using or recognising the five HTA criteria and types of error, and combined criteria/procedure training. Performance at HTA improved in both conditions that involved criteria training.
