In Human Reliability Analysis, we identify the factors which influence human performance on a critical task so that we can “optimise” them. In response to my recent post on Positive PIFs (Performance Influencing Factors), the question of what we mean by optimising PIFs was raised, and in this article, I thought I would try to answer this question.
We optimise Performance Influencing Factors to reduce the likelihood of human error. There are a lot of different PIFs though, so what does ‘optimise’ mean. Optimise is used as shorthand for an array of different tactics to either reduce the negative effect of a PIF or turn it into a positive influence. In other words – and how I wish I had set up a shortcut key for this phrase 20 years ago – it depends.
There are some Performance Influencing Factors we cannot directly affect. For tasks carried out outside, weather is a good example. However we may wish to, we cannot optimise the weather. What we want to do instead is optimise the conditions in which critical task steps are carried out – can we bring the task indoors, or provide local shelter, or appropriate clothing, or procedural controls to stop the task if the weather exceeds certain limits.
There are some Performance Influencing Factors that either only act negatively or only act positively. How we optimise these depends on whether they are discrete (ie they are either there or not), or act on a continuum. For discrete PIFs, we are in an all or nothing situation by definition. Discrete PIFs with a negative effect need to be removed. For example, if equipment is operated against the design stereotype such as opening a wheel-operated valve by turning it clockwise, the opposite of the “lefty-loosey, righty-tighty” rule I learned as a kid, it needs to be replaced with one that operates as normal. If you can’t remove the PIF immediately, eg you have to wait for a turnaround, then all you can do is mitigate the PIF as a short-term remedy. Discrete PIFs with a positive effect that are absent need to be implemented. For example, not labelling plants increases the likelihood of misidentification, which happens more frequently than you would like to believe – cheap, simple, obvious – why wouldn’t you?
Performance Influencing Factors which act on a continuum are often more complex. Fatigue, for example, always has a negative effect but the degree of effect depends on many other factors including individual susceptibility, time of day etc. In such cases, where we cannot remove the PIF, then we are looking to manage it in line with relevant good practise which should optimise it by reducing its effect as far as possible, acknowledging that there may still be a residual negative effect. Competency is a good example of a PIF that acts positively on a continuum, but once again the degree of effect depends on many others factors such as recency of training, individual aptitude etc. So again we are looking at implementing relevant good practise to increase its effect as far as possible, acknowledging it won’t always be perfectly optimised.
And finally there are PIFs that follow the stress curve – too little or too much are negative and we are looking for the comfort zone or Goldilocks ‘just-right’ effect for a positive influence. Workload, temperature, lighting, there are many examples. Optimising these PIFs means establishing what the comfort zone is – the limits within which the majority of people will be performing well. Sometimes there is evidence that establishes a norm e.g. studies have shown that human performance deteriorates below 17 deg C and above 27 deg C. Where these don’t exist, it will take collaboration with the user group to work out the limits. And given individual differences, allowing people a measure of personal control over the PIF, if possible, would be good practice.
So “optimising PIFs” seems to mean a range of strategies including (but I am sure not limited to) removing the PIF, implementing the PIF, and managing the PIF as best we can.