Designing Warnings to be Read and Remembered

Three kinds of memory for on-product warnings were tested as a function of four variables in an applied laboratory setting. The three kinds of memory tested were recall of hazards, recall of the means to avoid an accident, and recall of the action to take in the event of an accident. The four variables investigated were message length, serial position, message format, and pre-questionning. Basic research findings were substantiated in that message length, message format, and pre-questionning did have an effect on subsequent recall. However, the findings were not always in the expected direction and the effects were far from robust. Suprisingly, no significant effects of serial position were found for any of the three recall measures. Encoding problems due to the nature of the experiment, prior knowledge of warning information, and interference are discussed as explanations for unexpected effects.