The project’s aim appears to be positive in nature, to aim to reward good practice with positive feedback, and to try to spot and prevent situations that might lead to an accident. The promoters of the plan say that it is about keeping an eye on each other, since anyone can make mistakes.

The aim is not to reward good safety practice, which the organisation says it expects as a condition of employment, but to reward the time taken to write up the interaction.

The project documentation introduces how observations work: “The observation can be in any work situation, whether on plant, in the office, working outdoors, or travelling on and around our sites. Observation is about looking at what we are doing and providing each other with feedback. In most instances this will involve giving positive feedback because generally we are all working safely and to the right standards. There will however be occasions when we spot ‘at risk’ behaviours and/or ‘at risk’ conditions. This provides us with the opportunity, through a simple five minute observation, to help protect each other.”

Making an observation is a simple step-by-step process. The observer needs to engage on an adult level with the person being observed, and discuss safety without being critical. Questions to ask include, ‘why are you doing it that way’, ‘had you considered doing it another way’, ‘what might be the worst thing that could happen if you did it like that?’ Then the observer records the conversation on a paper form or on one of the many intranet computer terminals scattered around the site. The observer does not give the name of the person observed, but does give their department.

A human performance team reviews the observation, and credits the observer with shopping voucher points at the end of the month. Each documented observation, positive or negative, nets 10 points, which is roughly equivalent to GBP 1; in addition 5 points go to a charity nominated by staff groups. The review team also select the best observation of the month, which wins an extra 50 points and coverage in site media.

“Most at risk behaviours can be changed straight away. Most of the time people are working safely,” says Sellafield’s former head of human performance Joe McCluskey, who recently left the organisation. “The key issue is how people are approached, how a discussion begins, and part of that is the way you might explain that something is not as safe as it could be,” McCluskey says.

“The balance has to be to encourage our people; this is not just another programme to beat people up,” McCluskey says. Promoters say that they are trying to create a workforce that is confident to question, and be questioned, about poor behaviour and conditions. This questioning attitude, they say, is the ideal working environment, tempering confidence with a bit of criticism. Too much criticism makes people unsure; too little criticism is also bad, since people can start to feel too certain, or complacent.

The organisers are hoping that every employee will record one observation per month. Since the scheme began in April, more than 20,000 observations have been made. The data from observations recorded in the system are used for making conclusions about safety trends on site. Sellafield also has an observation and reporting scheme for non-interactive observations of unsafe events. In the future, the programme could respond to safety trends on site by incentivising observations in areas that turn out to be problematic, for example working at height, by offering a greater number of points.

&#8220The balance has to be to encourage our people; this is not just another programme to beat people up”

The scheme has three levels. All employees are requested to perform peer-to-peer observations. A more in-depth interaction, which might last 15 minutes, considers what it calls error traps: “latent conditions or situations that people can potentially be put into that may put them, or the business, at risk,” according to McCluskey. For example, a simple peer-to-peer observation would look at an activity, such as drilling in a wall. The error trap extends beyond this to look at how the employee is doing the job, how many other jobs he or she is responsible for, what sort of time pressures there are, how well he or she understands the work brief, and other factors. This second level would be conducted by team leaders and support staff. In a third level of observation, scheduled to be introduced in late 2010, managers would dig deeper still into the environment around an action, and the organisational context that surrounds it: what sort of training does the employee have, how isolated is he or she, what sort of communications links does this person have with colleagues, what is the equipment like. As a guide, such an interaction might take two hours.

McCluskey gave a few examples of real-life observations from the programme. One observation submitted warned that a worker looked ill, with a recommendation that his colleagues should keep an eye on him; the employee was eventually admitted to hospital suffering from the early stages of a stroke. Another observation had to do with material handling on site. A worker was transferring items from a wagon to a work site using a rough terrain forklift, a telehandler.

The machine was loaded so that the worker could not see out the windscreen, so he had to lean out of the window to drive. An observation and discussion led to a reorganisation of the job by using a third worker, a signaller, to help move the materials more safely.

McCluskey elaborates the aspirations behind the scheme. “If you think of an environment of an office, with PCs, if your perception is that this is a negative process, that it is about things that are wrong, and in an office of 50 employees, there is only one trailing wire, we can’t all report it.” This is the wrong idea, McCluskey says. “What we are asking people to do is engage in a conversation with people about those things that might lead them to have an accident…the process itself is ad hoc. There is also an element of planned work, in which a team leader for example could do a walk down the corridor, and could for example see someone fixing a handle on a door or fixing a ceiling tile; those observations can be unsafe ones. But it is also about getting positive actions reinforced, and people talking to people, putting in positive observations, workers using their eyes and their ears to be watching and be protecting others.”

McCluskey admits that disagreements about safe working can happen in the observation process. And sometimes observers flag an issue inappropriately due to their own lack of understanding of a situation. But he says that workers with a genuine safety concern should simply report it to their line manager, “and under no circumstances to get into an argument or a conflict; the purpose of this is to protect our colleagues.”

McCluskey says that the difficulty with any cultural or behavioural programme is sustainability; it can be difficult to maintain the focus of a long-term programme. He says that Sellafield has had a behavioural safety-style programme since 1994, which has had varying degrees of success. Every three years or so it is reinvigorated with a new push of some sort.

The programme is based on a incentive-driven observation scheme at the USA’s Idaho National Laboratory, which collected 70,000 observations over a two-and-a-half year period. In that programme, McCluskey says, although one point was awarded per observation, only 30,000 points were claimed. “That is a reflection that people were doing this to share information rather than for personal gain. We liked that principle, and decided to adopt it.”




The balance has to be to encourage our people; this is not just another programme to beat people up



Sellafield’s list of human performance tools

– Questioning attitude

– Self/check or STAR: stop, think, act, review

– Peer checking

– Pre-job brief

– Post-job review

– Independent verification

– Three-part communication

– NATO phonetic alphabet (alpha, bravo, charlie, delta…)