Assessing the inter-analyst reliability of the RIAAT process

This paper presents a preliminary assessment of the inter-analyst reliability of the RIAAT procedure, which is a recent tool developed for the Recording, Investigation and Analysis of Accidents at Work. The study involved 5 analysts who applied the RIAAT procedure to a set of 11 accidents at work occurred in a Portuguese company. The study focused on 8 nominal variables considered as key-variables within RIAAT. Reliability was measured with three coefficients for calculating the level of agreement between analysts. Overall, the results showed a low inter-analysts reliability for all variables tested; the first 4 (Eurostat variables included in RIAAT) held a reliability level almost acceptable. The remaining 4 variables (RIAAT-specific) led to lower agreements; this may be explained by their use in the analysis and coding of more "distant" causal factors. The authors argue for the benefit of having a well-trained team of investigators rather than one person, as it enhances reliability of information.