Uses exposure data for road safety analysis

In road safety analysis, exposure data can be used in two manners:

• To obtain risk data in the form of outcome per unit of exposure
• To describe differences in the road safety situation

An example of the first type is the number of fatalities per inhabitant. Such a figure is called a risk. In general it is defined by "the number of outcomes" divided by "the amount of exposure". An example for the second case is the number of motor vehicles per inhabitant.

Due to the potentially non-linear relationship between the number of outcomes (i.e. road accidents) and exposure, care has to be taken when risk estimates are multiplied by exposure figures to obtain estimates of the number of outcomes under different circumstances. Risk and exposure are not independent. When exposure changes, other factors determining the risk are likely to change too. It is advisable to focus, if not restrict, the use of risk figures to compare different risks and acknowledge that risks may be different because exposure and other related factors may be different.

When risks are compared, the potential biases in both accident outcome and exposure measure should be considered and, when possible, corrected for. When estimates are considered sufficiently useful, their measurement error has to be accounted for.

In road safety analyses, different exposure measures are used, according to data availability and quality, as well as the particular objective of the analysis. These measures may vary significantly in terms of the potential level of desegregation and the possible underlying bias in their estimates. No general rule is available concerning the preferred measures of exposure. Vehicle- and person kilometres of travel, as well as the time spent in traffic, are conceptually closer to the theoretical definition of exposure and can be theoretically available to a satisfactory level of detail. However, under certain conditions, other available exposure measures may be equally efficient for the purposes of a particular analysis and/or more reliable. These alternative exposure measures may also have other, explanatory or descriptive uses.