Chemical tanker vessels are at risk for large-scale accidents due to the nature of their cargo and operating conditions, challenging environments as well as general maritime hazards. To counteract such hazards, the ship-owning company works on maintaining safety at the organizational level, and the captain instantiates safety regulations on the interpersonal level. The crew members are expected to maintain safety by having accurate situation awareness, and beneficial safety attitudes and behaviour. We pre-registered an analysis to test for associations between safety variables in a survey for chemical tanker vessel crews. A structural equation model revealed that the ship-owning company’s safety climate and the captain’s leadership style were associated with the vessel’s safety climate. Further, the vessel’s safety climate was associated with individual safety attitude, situation awareness and adherence to safety management systems. Safety attitude had a central role in the model and was associated with situation awareness, reporting attitude, safe behaviour and adherence to safety management systems. The results imply that it may be beneficial to monitor and improve safety attitudes among crew on chemical tanker vessels and in similar work-environments.

We need reliable knowledge about what causes and prevents accidents in order to manage safety. Such associations are often investigated with self-report measures of psychological factors. However, few studies in the maritime industry have investigated the extent to which self-report measures predict objectively registered accidents. The current pre-registered study used structural equation modelling to test whether self-reported individual safety-related factors predicted objective safety outcomes in the following year. The study was conducted among crew on chemical tanker vessels operating in Arctic and Baltic waters. The pre-registered model and the expected associations between self-reported safety factors and safety outcomes were not supported. However, an exploratory model based on the pre-registered hypotheses supported an association between self-reported safe behaviour and the overall number of safety outcomes. While much safety research builds on the assumption that self-reported behaviour, attitude or cognitions are causally related to actual accidents, the current study shows that such a relationship can be difficult to confirm. We recommend more caution in assuming a causal relationship between self-reported psychological factors and safety outcomes until a causal relationship can be established.