Safety Measurement5 min read

Safety Climate Measurement: Which Survey and Why

Strong Evidence2 meta-analyses · 210 studies · 81,213 participants

The Question

If you are commissioning or selecting a safety climate survey, you face an overwhelming landscape. Since Zohar published the first validated safety climate scale in 1980, over 150 distinct measurement tools have been developed. Some are designed for specific industries — construction, healthcare, oil and gas. Others claim to work across any sector. Consultants each have their preferred instrument. How do you choose? And does the choice actually matter for what you can predict and act on?

What the Research Says

Jiang, Yu, Li and Li (NIOSH, 2018) conducted the definitive meta-analytic comparison. Across 120 independent samples and 81,213 participants, they systematically compared the predictive validity of industry-specific safety climate measures against universal (cross-industry) measures. The results were clear and practically important: industry-specific measures displayed better predictive power for safety behaviour and risk perceptions, while universal measures displayed better predictive power for adverse events other than accidents and injuries.

This means the two types of instrument are not interchangeable — they tap into different aspects of safety climate and predict different outcomes. An industry-specific survey for construction will better predict whether workers follow safe work practices on site. A universal survey will better predict the broader range of adverse outcomes — near misses, property damage, and organisational-level safety incidents.

Christian, Bradley, Wallace and Burke (2009) provide additional context. Their meta-analysis of 90 studies confirmed that group-level safety climate is the strongest situational predictor of safety outcomes overall, but they also noted substantial measurement heterogeneity across the included studies. Different instruments measured safety climate in different ways, and this measurement variation contributed to variability in effect sizes. In practical terms, your choice of instrument affects not just what you measure but how strong the relationship between your scores and actual outcomes appears to be.

The measurement landscape includes several well-validated options. Zohar's Group-Level Safety Climate Scale (originally 1980, updated 2002 and 2010) focuses on supervisor safety practices and has been extensively validated in manufacturing, rail, and construction. The Neal and Griffin Safety Climate Scale distinguishes management, supervisor, and co-worker levels and has been used in multiple meta-analyses. The Nordic Safety Climate Questionnaire (NOSACQ-50) offers a comprehensive 50-item, 7-dimension instrument validated across multiple industries and languages. The NIOSH Safety Climate Survey provides a short 8-item universal measure with good cross-industry applicability.

Beus et al. (2010) added an important methodological warning: content contamination in safety climate instruments — where items overlap with the outcomes they are supposed to predict — inflates apparent predictive validity. If your survey includes items about safety behaviour frequency alongside items about management commitment, the resulting score will appear to predict behaviour more strongly than it genuinely does. This contamination is more common in older, less carefully designed instruments.

Implications

Match the instrument to the measurement purpose. The Jiang et al. finding is unambiguous: if you want to predict and change frontline safety behaviour, use an industry-specific instrument that captures the particular hazards, norms, and practices of your sector. If you want to track broader organisational safety outcomes for benchmarking or governance, a universal instrument may serve better.

Consider using both. Some organisations will benefit from a two-tier approach: a short universal measure for cross-site or cross-industry benchmarking, combined with an industry-specific module for targeted behavioural improvement in high-risk operations.

Beware of consultant-proprietary instruments without published validation data. With over 150 instruments available, many in active commercial use, the quality varies enormously. Validated instruments with published psychometric properties and peer-reviewed evidence should be preferred over proprietary tools with no independent validation.

Check for content contamination. Review the items in your chosen instrument to ensure they measure perceptions of safety priority (climate) rather than safety behaviour frequency. If the survey asks "how often do your co-workers follow safety procedures?" alongside "how committed is management to safety?", the first item contaminates the climate measure.

What You Can Do

  1. 1
    ODesignDefine your measurement purpose before selecting an instrument. The evidence shows that industry-specific and universal measures predict different outcomes. Decide whether you are targeting frontline behaviour change, organisational-level adverse event tracking, or both, and select accordingly.
  2. 2
    ODesignUse validated instruments with published evidence. The evidence supports instruments like Zohar's scale, Neal and Griffin's scale, the NOSACQ-50, and the NIOSH Safety Climate Survey. Require published psychometric data (reliability, factor structure, criterion validity) from any survey provider.
  3. 3
    ODiagnoseAudit your instrument for content contamination. The evidence warns that items overlapping with safety behaviour outcomes inflate apparent predictive validity. Review each item and ensure it taps into perceptions of priority, commitment, and communication rather than behaviour frequency.
  4. 4
    ODesignConsider a two-tier measurement strategy. The evidence suggests combining a short universal measure for benchmarking with an industry-specific module for targeted action planning. This gives you both comparability and specificity.
  5. 5
    ODesignUse group-level aggregation for the strongest predictions. The evidence consistently shows that group-level safety climate (team or unit averages) is a stronger predictor of safety outcomes than individual-level scores. Aggregate your data to the team or workgroup level for the most actionable insights.
Intervention Level:IndividualGroupLeaderOrganisation

The Bottom Line

There is no single "best" safety climate survey. Over 150 instruments exist, and the evidence shows that industry-specific measures predict safety behaviour better, while universal measures predict broader adverse events better. Your choice of instrument should be driven by your measurement purpose — whether you are targeting behaviour change on the frontline or tracking overall safety outcomes at the organisational level.

Share:

Evidence Quality Note

We rate this evidence as strong. The Jiang et al. NIOSH meta-analysis provides a definitive head-to-head comparison based on 120 samples and over 81,000 participants — an exceptionally large evidence base for a measurement question. Christian et al.'s 90-study meta-analysis and Beus et al.'s methodological analyses add important context. Limitations include that the comparison is at the aggregate level — specific instruments were not compared head-to-head — and the evidence base reflects predominantly Western industrial settings. The practical guidance on which specific instrument to choose remains partially a matter of professional judgement.

Source Citation

  1. Jiang, L., Yu, G., Li, Y., & Li, F. (2018). Safety climate and safety outcomes: A meta-analytic comparison of universal vs. industry-specific safety climate predictive validity. NIOSH Report / Journal of Occupational Health Psychology. https://stacks.cdc.gov/view/cdc/207485
  2. Christian, M. S., Bradley, J. C., Wallace, J. C., & Burke, M. J. (2009). Workplace safety: A meta-analysis of the roles of person and situation factors. Journal of Applied Psychology, 94(5), 1103–1127. https://doi.org/10.1037/a0016172
  3. Beus, J. M., Payne, S. C., Bergman, M. E., & Arthur, W. Jr. (2010). Safety climate and injuries: An examination of theoretical and empirical relationships. Journal of Applied Psychology, 95(4), 713–727. https://doi.org/10.1037/a0019164