How do we collect and interpret human safety performance data?

Since human performance demands human perceptions and observations we need to accept its subjectivity – both good and bad...

Since humans are the largest influencing factor on safety climate and performance the logical deduction must be that – if we are to assess, correct and develop our safety climate – we need to publicly show, discuss and elaborate on human safety performance. Since human performance demands the involvement of human perceptions and observations we also need to accept its subjectivity.

In our article ’The art of measuring safety performance’ we tried to extend our understanding of what additional safety data is needed (going beyond the use of lagging and leading indicators only) to provide us with a more robust picture of our present safety conditions,  defined as Safety Condition Indicators. Also we argued that this data should serve as a tool to help us articulate and understand our safety performance and conditions better – especially our safety behaviour. The article recommended focusing on four areas: Crew safety competencies; Crew satisfaction with company safety tools; Employee perception of safety climate; Frequency of safety processes actually taking place.

But how do we gather the data within such ‘uncountable’ subjects and what challenges do we need to overcome in order to ensure a lean process? This article describes typical challenges we face when trying to gather, process and present safety (behaviour) performance data and a few methods believed to help shipping companies manage this challenge.

The challenges we face

Robust safety performance data is critically important to ensure a meaningful assessment of the potential risk and hazards at a work place. The motivation is simple: we cannot improve or discuss what we do not measure. However, to ensure robust data management it is the belief of Green-Jakobsen that the key focus should be on management of data rather than extreme data gathering. Arduous and excessive data collection or poor interpretation of data may be a major disincentive among employees to participate actively and constructively in the use of the data provided.

So, how do we gather data on human behaviour and performance? How do we have a systematic approach and still embrace the different perceptions, and who does what, when and where?

Evaluating, gathering and measuring human safety performance (and behaviour) is different from gathering other types of performance data such as time, temperatures, technical tolerances etc. Shipping companies will need to discover other ways of gathering data, and as described in our article ‘The art of measuring safety performance’  Green-Jakobsen sees two predominate ways of doing this:

1. Crew appraisal on safety performance
2. Signing off survey to gather crew perceptions and observations on safety climate.

To get these processes on track and let them become an integral part of our work we have challenges to overcome. If we don’t overcome these we risk ending up having to accept John Dewey’s statement indirectly claiming the fact that humans are reactive in their approach towards life: Humans first think when they are confronted with problems. Apparently this is not appropriate since our aim is to equip humans to be prepared and not surprised!

Since humans are the largest influencing factor on safety climate and performance the logical deduction must be that – if we are to assess, correct and develop our safety climate – we need to publicly show, discuss and elaborate on employee safety performance

Typical safety performance measurement challenges and how we overcome these

Collecting data on human safety behaviour and performance will, like any other data gathering process, demand input from some kind of data source. Since the data source relies on human perceptions and observations the data collection process relies on employee input, which then again results in a subjective assessment. But the question is: What is the alternative? Should we not discuss our safety behaviour and performance? Should we give up and make excuses by claiming that it is too subjective and difficult to manage?

The short answer must be that have to accept subjectivity, manage it and subscribe to the fact that human safety performance demands the involvement of humans. We also need to accept the fact that human safety performance and behaviour – to be of value – has to be publicly addressed and that our observations and perceptions need to be subject to discussion.

As described in our article ‘The art of measuring safety performance’ we have identified 4 key safety behaviour and performance areas to be measured. These are:

Employee perception of the workplace safety climate
Employee perception of the quality of the toolbox-talks made, risk assessments, etc.
Manager/officer assessment of employee safety competencies
Employee perception of the extent of the use of safety tools, PPE, etc.
We need to accept the fact that human safety performance and behaviour has to be publicly addressed and that our observations need to be subject to discussion

Once again – common for all these 4 areas is that they all rely on employee perceptions and therefore to a high degree are subjective. But what is the alternative? Since we now have defined appropriate human safety behaviours (Green-Jakobsen Safety I’s) we want our employees to apply them, and how do we overcome this?

In many cases, safety data is unavailable or unknown. A variety of strategies can be employed to collect quality data, perform analysis, and ensure that safety stakeholders can access the data and the analysis. The table below has been prepared to describe challenges encountered when building a robust performance measurement.

Typically Green-Jakobsen describes these challenges through the application of our Performance Butterfly model splitting the process into 4 categories: 1) Standards and direction 2) Supporting means 3) Human performance and leadership and 4) Organisational anchorage. (For a full description of the Green-Jakobsen performance butterfly we can refer to: Making safety initiatives last). The table highlights some key challenges and offers a few ideas on how to overcome these challenges.

 Standards and direction – How can we…
…ensure that all employees have a shared understanding of the expected safety performance and behaviour? The short answer here is: Implement and apply the Green-Jakobsen Safety I’s. Green-Jakobsen’s Safety I’s define what is believed to be good safety behaviours understandable, applicable and measurable at all levels of the organisation. The application of Safety I’s offers the company a tested and manageable safety behaviour programme and process.
…ensure that all employees have a shared perception of the safety performance and behaviour? Let’s be honest – this is a delicate process demanding great persistency with many change tactical considerations. Developing uniform safety behaviour based on shared perceptions of what is good and bad is a process not to be dealt with lightly. Imagine the scenario where your employees feel unsafe if a safety debrief has not taken place. Imagine that you have employees who all feel comfortable correcting a superior officer if he or she is doing something wrong. Developing a safety culture of this depth and effectiveness demands patience and persistency but what is the alternative? Not doing it? If improvements are needed – this is hardly the way forward. We must believe that we can achieve changes for the better, which research within this field of work also confirms.
Supporting systems – How do we…
…gather data? If we wish to measure performance we need to gather data.  Some data is easy to gather whereas other data require larger employee efforts. How do we e.g. gather data addressing crew safety behaviour and competencies? The short answer is 1) Crew signing off survey and 2) Appraisal of crew safety performance
 …minimise the need for coordination of data collection? Who does what, when and where are all important questions to address and answer. If data collection becomes a logistical challenge data will not be found. Minimising the number of ´handlings’ is extremely important and a number of effective and cheap IT tools are on the market. If needed, replication of data from HR management systems could be an option.
…ensure that the data we gather is of value? Performance data has to be relevant for the receiver. The closer data relates to the receiver’s world of performance the more likely it is that the data presented will result in reflection and action. For various reasons LTIF and FAC result indicators are important but for ship crews the data is very high-level and without any evidence of specific ship crews safety performance. Data has to be ’practice close’.
…ensure consistent reporting? Data demands a strong and robust reporting culture. Without data we might as well forget all about it. But where do we find data and who shall report it. However, one thing is certain; data on human safety performance requires data input from those who observe the performance. Consistent reporting requires employees who report as agreed and in the spirit it was designed.
…ensure the data is timely? On-going and daily dialogue is the most effective way of keeping a close eye on our safety behaviour and performance. But the open and trusting dialogue between peers, officers and subordinates can often be questioned. The systems to support the process formalise data gathering.
Human reaction – How do we…
…communicate towards our employees how the data is processed and presented? Ethical guidelines (expressing the necessary respect for privacy and sensitive information, as well as a guarantee that data will not be abused) are important and it has to be absolutely clear how data is processed and to what degree the reporting is anonymous. A typical challenge is to what degree it should be visible how senior officers have a positive or negative impact on the on board safety performance. Will the company allow publicising performance data on e.g. specific Masters or not? All this and other issues have to be stated in the ethical guidelines.
…convince our employees that the data they get is of value? The best way to convince employees that the data is of value is to ensure data employee relevance. Like life in general our responsive behaviour increases if what we see is relevant for our life or wellbeing. In other  words, we will have to gather, process and present data employees can relate to and which are appropriate to their position and responsibilities. Moreover, the data shall provide them with a safety performance understanding that helps them articulate safety performance and issues. If we fail this objective employees can and will not respond.
…ensure that our employees aren’t reluctant to share their true perceptions? This question is very important, but unfortunately also often difficult to control. Many factors can influence the willingness to express true perceptions so if we wish to overcome this we have to be persistent in our focus in this area. Constant communication on the company’s need for the true perceptions of employees is pivotal. Any opportunity to express this need has to be included in interaction, process or programme. An influential factor is the character of the applied leadership.
…minimise data subjectivity? The short answer is that we accept subjectivity, but manage it. What employee perceptions offer is a starting point for a dialogue on specific and desired safety behaviour and performance. The belief is – if we are to create a shared and strong understanding of our present safety behaviour and performance – that we need to make it ‘public’ what we see. If perceptions are not made public our beliefs are never challenged and the company’s employees ‘can get away with stuff’ they shouldn’t have done or believed.
Organisational process – How do we…
…ensure that crew respond and correct in accordance with data provided? Performance data without appropriate response is of no organisational value. To avoid a devaluation of the need to gather, process, present and respond to performance data the company ensures and legalises performance data discussions and response during existing and regular work processes. Safety performance could e.g. be discussed during on board safety meetings or during ship/shore interaction on ship operations in general.
…advise our crew on how to follow-up? The use of Safety Action Plans is a commonly used tool to organise, prioritise and implement safety performance improvement initiatives. However, most importantly a verification feedback loop shall be applied. Safety performance data is only of value if it is publicly discussed and used to guide future changes and corrections.
…use the data gathered from the sign off survey and crew appraisal? The data gathered should be discussed during e.g. safety meetings, toolbox talks, preparation of risk assessments, etc. However, most importantly crews or departments need to react to any indications of poor performance. A passive or reluctant reaction will undermine the value of the Safety Condition Indicators.
….encourage positive safety behaviour? Far too often safety is about what to avoid, how to learn from what went wrong or how to manage the risk possessed by human beings. But in this respect it is important to note that human beings most of the time do their job without getting hurt* so why don’t we spend more time discussing the things that we do well and make us work safer? The key message is that we need to emphasise and reinforce what goes well but of course also manage the bad. Discussing positive safety behaviour will create a stronger willingness to discuss safety performance.


* Hollnagel, various articles

Performance Butterfly model

Need help?

Contact us for personal advice

Erik Green
Erik Green
Managing Director and Partner

Would you like us to call you?

Follow the flow of inspiration - Subscribe to GJ Academy
Thank you for your Subscription