What are probabilistic approaches?

Current methods for risk assessment are mostly "deterministic". This means they treat factors such as the toxicity of pesticides as if they were fixed, and precisely known. But in the real world, factors such as toxicity are not fixed but variable. For example, the same pesticide could be more toxic to some species of wildlife, and less toxic to others. What's more, the factors affecting risk are not precisely known but uncertain. For example, toxicity is measured for only a very small number of species, so scientists have to estimate toxicity to all the other species that we want to protect.

Current methods for risk assessment try to allow for variability and uncertainty by using "fixed safety factors", but this fails to give a complete description of the full range of the possible risks. Also, it is difficult to decide how big the safety factors should be.

Probabilistic approaches enable variation and uncertainty to be quantified, mainly by using distributions instead of fixed values in risk assessment. A distribution describes the range of possible values (e.g. for toxicity), and shows which values within the range are most likely. The result of a probabilistic risk assessment can also be shown as a distribution, showing the range of environmental impacts that are possible, and which impacts within that range are most likely. This should provide a better basis for making decisions about pesticide risks, because the full range of possible outcomes can be taken into account.

More information and examples will be added to this website as the project progresses. Technical summaries of recent work on probabilistic approaches, including some worked examples, can be found in the EUPRA workshop report - for more information go to www.eupra.com.