A Cross-Level Investigation of Factors Influencing Unsafe Behaviors and Accidents
Hofmann, D. A., & Stetzer, A. — Personnel Psychology, Vol. 49, pp. 307–339
Overview
This study is one of the first to apply multilevel (hierarchical linear) modeling to the study of workplace safety, examining how group-level factors shape individual unsafe behaviors and accident rates in a manufacturing context. At the time of publication, most safety research treated organizations as single-level systems — studying either individual attitudes or broad organizational policies. Hofmann and Stetzer argued that safety outcomes are the product of both individual behavior and the group environment in which that behavior occurs.
Key Questions
The research addressed two central questions: Do group-level safety climate and work group norms predict individual unsafe behaviors above and beyond individual-level factors? And do these group-level variables also predict accident rates at the work group level?
Methods
The study drew on survey data from employees across multiple work groups in a manufacturing organization. Safety climate was measured at the group level by aggregating individual perceptions, following established aggregation justification procedures. Hierarchical linear modeling (HLM) was used to partition variance in unsafe behaviors across individual and group levels, enabling a true cross-level test of the hypotheses.
Findings
Group-level safety climate significantly predicted individual unsafe behaviors even after controlling for individual attitudes and perceptions. Work group norms regarding safety — the informal expectations within a team about how seriously to take safety procedures — were among the strongest predictors. Groups with strong, positive safety climates had members who reported fewer unsafe behaviors, and those groups also had lower accident rates. The findings underscored that safety is a fundamentally social phenomenon, not simply a matter of individual compliance.
Significance
With over 900 citations, this paper is among the most influential in the workplace safety literature. It helped establish the theoretical and methodological foundation for multilevel safety research and reinforced the importance of organizational and group climate — not just engineering controls or individual training — in accident prevention. The paper is widely assigned in I-O Psychology doctoral programs and remains a foundational reference for safety climate scholarship.