Psychometric model for safety culture assessment in nuclear research facilities

•A psychometric model to evaluate ‘safety climate’ at nuclear research facilities.•The model presented evidences of good psychometric qualities.•The model was applied to nuclear research facilities in Brazil.•Some ‘safety culture’ weaknesses were detected in the assessed organization.•A potential tool to develop safety management programs in nuclear facilities. A safe and reliable operation of nuclear power plants depends not only on technical performance, but also on the people and on the organization. Organizational factors have been recognized as the main causal mechanisms of accidents by research organizations through USA, Europe and Japan. Deficiencies related with these factors reveal weaknesses in the organization’s safety culture. A significant number of instruments to assess the safety culture based on psychometric models that evaluate safety climate through questionnaires, and which are based on reliability and validity evidences, have been published in health and ‘safety at work’ areas. However, there are few safety culture assessment instruments with these characteristics (reliability and validity) available on nuclear literature. Therefore, this work proposes an instrument to evaluate, with valid and reliable measures, the safety climate of nuclear research facilities. The instrument was developed based on methodological principles applied to research modeling and its psychometric properties were evaluated by a reliability analysis and validation of content, face and construct. The instrument was applied to an important nuclear research organization in Brazil. This organization comprises 4 research reactors and many nuclear laboratories. The survey results made possible a demographic characterization and the identification of some possible safety culture weaknesses and pointing out potential areas to be improved in the assessed organization. Good evidence of reliability with Cronbach's alpha coefficient of 0.951 was obtained. Validation method was based on Exploratory Factor Analysis (EFA), using Principal Components Analysis (PCA) and Varimax orthogonal factor rotation. The results confirmed the unidimensionality of the items and, almost entirely, the conceptual framework of the safety culture proposed for the instrument. However, the results also suggested that some adjustments to the conceptual framework of the instrument must be performed in case of a new application.