Learning from incidents is vital in aiding organizations in the elimination of Serious Injuries and Fatalities (SIFs).  The goal of this project was to determine what happened and why, highlighting any causal factors that led to the incident.  This approach enables organizations to learn from incidents and make appropriate changes to ensure they won’t happen again.

 

Most incident investigation resources focus on the analysis of data collected, and there are many different academic and commercial tools available to help with that step.  However, within these tools there is often a presumption that the information collected is itself of high quality – but that is not always the case.

Our research found that this first stage in the process is often neglected, and how to ensure high-quality information collection has not seen much attention in either research or practice.  The quality of this information is critical to the rest of the process.  Poor quality information will result in poor analysis, inform poor organizational learning and thus limit the implementation of positive change.

​

​

We undertook interviews with industry experts, ran focus groups and held a number of workshops with the team. Analysis of the data generated enabled us to also distil best practices from the data, grounded in the collective 300+ years of experience within our industry team.

​

We also ran a unique simulated experiment to examine incident investigation interviews in depth.  Analysis of the resultant data revealed that bias is a common problem in incident investigations.  This is not itself a surprise – as we are all biased! – but our research enabled us to highlight which biases are most prominent and problematic in this type of interview and how they emerge.  

​

From this research we have developed bespoke guidance and tools to help investigators overcome issues in information collection through interviews within the construction industry.  All the tools are available through the Knowledge Center on this website, or you can just…