Likelihood is estimated from how often the task is done, how often and how closely people are near the hazard, and whether it has happened before here or elsewhere. The code offers a five level scale (certain, very likely, possible, unlikely, rare) and questions on how often and how long people are exposed, how effective current controls are, whether organisational change or seasonal demand raises likelihood, whether the working environment (heat, wet floors, poor light, work pace, poor natural surveillance) makes harm more likely, how human behaviour such as fatigue or distraction plays in, and how individual differences matter, including disability, inexperience, visitors and groups more exposed to harassment.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.