Keeping people safe through the power of behavioural science and data science
Almost all workplace accidents and incidents are down to human error – despite organisations increasingly seeking to either improve or put in place new safety processes and policies. A better way is needed to understand – and improve – the link between people’s behaviours and improved safety performance.
Using a combination of behavioural and data science, organisations can gain a much clearer understanding of why accidents and incidents occur, identify targeted interventions, and predict and prevent incidents from happening.
Drive understanding with data
Historically, health and safety functions have relied on quantitative data, such as KPIs, to determine and record unsafe behaviour. The only way to understand not just what happened, but why, is to read reports verbatim. Yet this way of moving from data to insight isn’t practical at an organisational level. It also subjects reports to individual biases and judgements.
With Natural Language Processing (NLP) techniques, data scientists can analyse masses of unstructured textual information to uncover actionable insights based around the why. The key with NLP is to ensure your data analysis is compliant with data protection regulations, such as the General Data Protection Regulation. Using incident reports as a core data source is generally good in this regard, as they often exclude data related to any individual but still contain actionable insight that can drive improved safety performance.
It’s an approach we used in partnership with a transport client to analyse a large number of incidents, categorise and visualise them, and then establish the behavioural causes behind them. This method also provided the organisation with the ability to segment incidents by locality, roles, or times – all helping to target interventions and improve safety.
Pursue data-driven targeted inventions
NLP analysis is only part of the answer, however. With the definition of ‘unsafe behaviour’ often industry- or even organisation-specific, there needs to be a well-defined framework, backed by science and research, that can make any behavioural data insight truly actionable. This is where a behavioural scientist or psychologist will play their part. They have the means to apply a human lens to data-led insight based on their expert knowledge of how people behave – and to deliver effective interventions.
Insight from these learnings can help shape recruitment, training, and leadership. If, for example, it’s found that a lack of conscientiousness results in a higher rate of fatalities within certain job roles, it would benefit recruiters to screen for this behaviour during the interview process or undertake aptitude tests to assess people’s levels of conscientiousness.
For in-house training, a behavioural scientist or psychologist can create conditions that replicate real-life scenarios, producing the same psychological response. This approach ensures that any interventions are likelier to be more effective in the real world. PA’s Occupational Psychology team recently reviewed a client’s training approach to a safety-critical role. The team provided guidance on the most effective methods to integrate the non-technical skills and behavioural material into the technical skills material, developed behavioural dilemmas and assessments to be used throughout, and reviewed the skills required by the trainers to successfully deliver the learning programme.
Insights can also be used to develop strategic approaches to safety management, drive safety conversations, and allow front-line team leaders to identify behavioural signs that could lead to a safety incident. We recently supported a large transport organisation with this by identifying key leadership accountabilities required during the selection, training, mentoring, and development of safety role-holders.
Preventing future incidents
By understanding why incidents are occurring through behavioural patterns over time, organisations can start to link the cause and effect of incidents, reducing – and even stopping – accidents and fatalities.
By actively responding to clearly defined and understood behavioural indicators, leaders are likely to prevent an incident from happening. They’ll also be better placed to understand previously data-rich but insight-lite metrics. For instance, a recent client safety analysis using NLP showed that accidents and incidents dropped during Bank Holidays, when more stringent planning was in place. This allowed the organisation to learn from the behaviours at that time and update their ways of working.
There are great opportunities in different sectors to apply data science and behavioural science together. Their combined power can help drive much better data-driven decisions, better protecting organisations and their people.