16th World Congress on Ergonomics - 2006 Safety Signs Comprehension and the Context Effect M. Emília C. Duarte a , F. Rebelo b , a IADE / UNIDCOM – Research Unit of Design and Communication. Escola Superior de Design. Av. D. Carlos I, nº4. 1200-649 Lisboa, Portugal. mduarte@iade.pt b Ergonomics Laboratory, Technical University of Lisbon. Estrada da Costa. 1499-002 Cruz Quebrada - Dafundo Portugal. frebelo@fmh.utl.pt Abstract This paper focuses on a major factor in the evaluation of safety signs comprehension: context. Commonly the warning’s comprehension is tested using open-ended questionnaires with the signs presented individually, printed out in colour together with a plausible context of use. The presence of context can provide the test with ecological validity, but can also, inappropriately, inflate comprehension scores, as well as provide false cues. In this study, a sample of 90 individuals evaluated 5 safety signs. The obtained verbalizations were now thoroughly re-analyzed. A composite variable named “context effect” was produced from the average rating of context references. The results shows that of all answers obtained 56% were context positive, ranging between a maximum of 70% and a minimum of 44,44%.The obtained “context effect” average was 70,4 references, or 0,78 references/answer. The statements content was also semantically analysed to understand what the subjects had seen and to what extent that had influenced their own conclusions. The obtained results showed that the context influenced the safety signs comprehension. This methodological problem can be overcome by using interactive procedures, such as, the creation of a virtual reality context in a computerized simulation. Methodological implications for safety signs analysis and design are discussed. Keywords: experimental project; warnings; information design; cognition; design for all; ergonomics at safety & health. 1. Introduction Warnings are an important research subject in ergonomics and safety due to its role inside the safety systems. But a warning utility in conveying safety information is proportional to the extent to which it is comprehensible in the population to which is directed. Therefore, comprehension testing is considered a useful tool in developing effective warnings and determining which warnings must be revised. Several authors have discussed comprehension tests (Silver et al. [1]; Wolff & Wogalter [2]; Zwaga [3]) and showed that obtained scores are affected by the context provided and test method. There are comprehension test procedures that employ multiple-choice techniques. But, Wolff and Wogalter [4] concluded that this procedure lacks ecological validity and inappropriately inflates