Workers' Mental Models of Chemical Exposure in the Workplace

The objective of this study was to examine workers' mental interpretation models developed in response to occupational chemical exposure. The study was performed in six companies within the reinforced plastics industry in northern Sweden, in which styrene was used; 32 workers participated in the study. Each worker performed between four and seven exposure measurements. Before receiving each result of the second to seventh measurements, the workers were asked to predict the level of their next exposure measurement. Their predictions were evaluated with respect to two judgmental principles: coherence (that the predictions are based on logical decision rules, that is, the mean value of the prior exposure levels); and correspondence (the predictions have high empirical accuracy) by calculating the mean absolute percent forcast error (MAPE). The coherence principle was tested by comparing each of the workers' predictions with the mean, median, and last exposure level (last value) of the prior measurements. The correspondence principle was tested by comparing the worker's prediction with the outcome of the measurement. The coherence principle was found to be the best descriptor of the workers' predictions and the median model had the best fit. The mean model had a similar but significantly poorer fit (MAPE values of 29 and 31, respectively). The correspondence model had a poor fit with a MAPE of 54. The workers' predictions were generally lower than their average exposures. We conclude that the workers' interpretation model can be best described by a coherence model rather than by a correspondence model.