For example Beus and co-workers (Beus, et al., 2010) found that injury rate was a stronger
predictor of safety climate than the reverse. To better understand the causal relationships
between safety climate and safety outcomes, longitudinal studies based on multiple
measurement points are needed. The first aim of this study was therefore to investigate the
causal relationships between safety climate and safety behaviour by means of a four wave
longitudinal design. We also applied a multi-level approach to further investigate the causal
relations at both the group and the individual level