This is a really interesting study that assessed the shelf life of safety climate (SC) assessments and how far into the past and future can it predict incidents (and incidents predict SC). As SC is a snapshot at that particular moment, it’s unknown how long that assessment provides meaningful info about the organisation.
Surveys were run at a multinational chemical company with >7k responses at 42 worksites linked to >14k incidents over a 4 year period.
Near misses, learning events (events that prompt info sharing) & 4 levels of incidents were included. Level 1 are minor first aid injuries, level 2 were more serious but not permanently disabling or fatal, and levels 3/4 were more catastrophic but due to the low numbers couldn’t be included into the models.
The 2 research questions were:
- How far into the future does a SC assessment predict safety incident rate? This was for SC as a leading indicator.
- How far into the past from a SC assessment does the incident predict SC assessments? SC as a lagging indicator.
Results
Results indicate that SC predicts incidents of varying levels of severity but it predicts the most severe over the shortest period of time with the same relationship with more severe incidents predicting SC over the shortest period. Notably, the incidents predicted by SC aren’t entirely the same incidents that go on to predict SC.
As a leading indicator: When predicting the more severe incidents is strongest in the month following the SC survey and “expires after 3 months”. Thus, quoting the authors “results indicate that the assumption (often unstated in organizations) that a safety climate survey can be conducted annually—like a typical job satisfaction survey or a health insurance satisfaction survey— and be assumed to retain its explanatory power for 12 months is incorrect, at least for more threatening events” (p533).
Further, “the typical aggregation of incidents over a year-long period—or even a 6-month period—would suggest that safety climate can do little to predict these more severe incidents” (p530). However, if very short time periods are used between SC assessments (e.g. 1-3 months), then it can be predictive of level 2 events.
An exception to the immediacy of prediction for level 2 is periods over 15 months *prior* to the incident where data at this longer prior period predicted subsequent SC assessments.
A single SC assessment can nevertheless contribute to the prediction of level 1 events, the least damaging and for up to at least 2 years following the survey period. Therefore for predicting minor events, annual SC surveys appears valid.
For learning events, greater time periods ahead (upwards of 2 years) further increases the predictive relationship.
SC as lagging indicator: in contrast to SC as leading, results indicate that each of the four incident types predict later SC performance. Nevertheless, “safety climate was never a good leading or lagging indicator of Near Misses” (p534). That is, near misses weren’t predicted by SC, nor did were a good predictor of subsequent SC performance.
Authors note that this lack of relationship between SC & near misses is disconcerting, but suggest that, perhaps, near misses may often get classified as learning events, thus explaining why learning events are predictive but not near misses.
This may suggest that “organizations must be as concerned about their incident recording processes as they are about their survey data” (p534), because the data isn’t as objective as they seem to think.
Even so, results suggest that the organisation “does not learn from Learning Events, whether they were Near Misses or something else, because there is no disruption in the prediction of safety climate from much earlier events” (p534).
Authors: Bergman, M. E., Payne, S. C., Taylor, A. B., & Beus, J. M. (2014). Journal of business and psychology, 29, 519-540.
Study Link: https://dx.doi.org/10.1007%2Fs10869-013-9337-2
LinkedIn post: https://www.linkedin.com/pulse/shelf-life-safety-climate-assessment-how-long-until-ben-hutchinson-usx9c
Safety & Performance Research Summaries 