When I started my research project, I thought I had a good idea about what the problems were and how I could identify them. Needless to say, life is much more complicated than that!! The SCUBA diving industry has its ideas about what the problems are, the divers who dive for fun have their ideas, the lawyers have their ideas and those who just complain for a living, have their ideas too. Of course, very few of these opinions align!
This is where the Delphi method comes in. It has its roots in financial forecasting but can also be used to provide anonymous and confidential consensus opinions from a knowledgeable group. See Wiki for more information on this.
So how does Delphi fit in with my research? As I said there are a considerable number of opinions of how safety could be improved in the industry, the majority of which try to get the user/diver to take more personal responsibility for their actions, recognise the risks they are taking when they go diving, and when things go wrong, look to themselves first rather than try to blame someone else. However, there is considerable evidence from literature to show that if incidents are happening with similar outcomes, then there is likely to be a reason higher up the organisational or supervisory tree leading to the incident at the user level. The Delphi method will aim to identify those higher-level factors by getting a consensus opinion.
The following image shows how the Delphi method works. A problem space is defined and subject matter experts are chosen to try and come up with a solution to the problem. In the case of my research I am going to randomly select approximately 250 instructors covering OC recreational, OC technical and CCR diving and ask them to rate the top 10% of factors from the organisational and supervisory layers of HFACS model which would influence divers to make a mistake at the individual level. The factors provided will not be the standard HFACS model but rather a conglomerate of a number of HFACS models.
Once I have the first set of results back, I will assess which factors occur the most frequently and resend the questionnaire back out for reassessment by the instructors given the reordering that has taken place.
Another pass may be required to get a good consensus opinion. There are no rules about how many passes are required, but rather ‘law of diminishing returns’ is applied.
The output of this research module will identify to all within the community, those areas which the organisations need to focus on to prevent incidents from occurring at the user/individual layer.
To ensure that the research is conducted as robustly as possible, a control group of senior instructors/instructor trainers, who have a positive attitude to safety and have already provided some feedback into the individual layer model validation, will provide an independent Delphi assessment.