why does it work?

Volume and Severity of Injuries are not separable pieces of information. The two may move together, or they may diverge. It is important to know whether an injury rate is climbing while the average severity of included events is dropping, or vice versa.

Under-reporting is real and present, always; it's a natural part of any measurement system. It is more prominent in lower-severity events where there is more room for interpretation. An injury rate that treats all injuries as equal is highly sensitive to under-reporting of lower-severity events. A weighted frequency rate is less sensitive if a number of lower-severity events are not included. The average SAIF also highlights differences in the relative magnitudes of events included across groups, for whatever reason.

SAIF works by setting a benchmark injury and assessing others relative to it. The benchmark for SAIF is an injury of two weeks' lost time. This is the injury for which TRIF and SAIF would both be 1. For an Injury Rate, all injuries are assessed as 1; for SAIF, injuries are adjusted based on whether they could reasonably be considered better or worse than a two-week LTI. The result is as follows:

                  TRIF Rate = People Hurt / Exposure Hours

                  SAIF Rate = Relative number of people with two-week LTI / Exposure Hours

                  Average SAIF = The average score of injuries included in the SAIF Rate

Using this method, it is possible to see the difference between an organisation with a high volume of low-severity events or the reverse. The SAIF, as a weighted rate, allows a more realistic comparison than TRIF alone. Understanding both the rate and the relative severity of included events yields a more dynamic picture of injury outcomes. The SAIF rate is also statistically less sensitive to the under-reporting of lower-severity events.