Search Results

Most of our understanding of traffic safety comes from analyzing data. Derivations from data, strictly speaking, tell us only about properties of data sets. They become important only with the assumption that the data reasonably represent reality. Yet what is included in a data set rarely corresponds to exactly what really happened. Cases that should be included are not included, and cases that should not be included are included. The most reliable information is for fatalities, yet even fatality data are far from perfect. For non-fatal crashes the problems are vastly greater. Indirect means can be employed to compare expected and reported injuries. The number of injuries per fatality, and the number of injuries in similar crashes, should remain fairly constant in time and between countries. This is examined using data from the US, Canada, Great Britain, Northern Ireland, Ireland, and Lithuania. Large discrepancies between reported and inferred injuries are found.