EFFECTIVENESS
The baseline measure of seat belt safety is
"Effectiveness".
The NHTSA has defined an "Effectiveness" factor (E) for seat belts (in "Estimating Lives saved by Restraint Use in Potentially Fatal Crashes. 1995") as "the proportion of lives that would be saved if a safety belt was worn." The value of E can theoretically vary from +100% (everybody belted lives) to 0% (seat belts have no effect) to minus infinity (everybody who puts on a seat belt would die.).
The original estimate of the effectiveness factor "E" was 45% and was obtained by the crash testing of dummies. This was a highly artificial test, since it only tested the belt's effectiveness in frontal collisions with an inert puppet. Most fatal collisions are not frontal, nor is the victim inert, but could be expected to brace his or her body to reduce the severity of injury. So legislators required evidence of the actual effectiveness of seat belts before they enacted laws compelling people to wear belts. Production of this evidence was commissioned to various scientists and engineers.
Most actual proofs of effectiveness have relied upon
finding
(or producing with legislative action) a time-step variation in the
proportion
of people using seat belts, and then comparing the count of
fatalities before and after that step. Most studies of this type
were inconclusive, the few that on first flush appeared to produce
significant results
were found by later reviewers to have serious deficiencies.
An innovative method for estimating effectiveness was the "double pair comparison" developed by Dr.L. Evans. Evans compared belted and unbelted people in the same fatal car accident, and compared their survival rates. Evans has agreed that his method would produce an overestimate of effectiveness if survivors over-report their own use, but was of the opinion that people would not lie about wearing a seat belt unless there was a penalty for doing so. It has since become evident that drivers will dissimulate about belt wearing status even when there is no penalty for failing to wear a belt. (See e.g. Fig. 1 of Reinfurt DW, Campbell BJ, Stewart JR Stutts JC.)
For a review of some of the important studies and their deficiencies, see the excellent books by John Adams "RISK" and Gerald Wilde "Target Risk".
Two theories of human behavior have been proposed to explain the difficulty that has been found in establishing that seat belts have saved the tens of thousands of lives that theory predicted.
LITERATURE REVIEW
In the last 25 years there have been many studies that have attempted to confirm that seat belts actually reduce the road toll. In the literature of any discipline, some papers are cited more frequently than others. With apologies to the authors of numerous other worthy papers, I have selected the following papers as signposts in the development of our cultural attitude to seat belts.
In 1976, Andreassend estimated that deaths to drivers and front seat passengers in Victoria, Australia had declined about 12.5% after introduction of legislation compelling the wearing of seat belts. This estimate has since been discredited.
In 1986, Harvey & Durbin estimated that deaths to drivers and front seat passengers in the UK had declined about 20% after introduction of legislation compelling the wearing of seat belts. This estimate has since been discredited.
In 1987 Evans did a double pair comparison of the fatality rates between belted and unbelted front seat occupants on data collected between 1975 and 1983, and calculated that seat belt effectiveness was within the range predicted by crash testing of dummies. This estimate has since been discredited.
All of the evidence (most of which has been sponsored by various
government
traffic safety authorities or automobile manufacturers) that has been
used
to establish that seat belts save lives has been discredited. No
undiscredited study has been able to show that nett lives have been
saved by adoption of seat belts.
Scientists have constructed theories to explain this failure to
establish in the real world that safety belts are saving lives.
One theory is the "Selective Recruitment Theory" by Evans.
Another is the risk
homeostasis theory, (Wilde) otherwise known as the "risk
compensation theory"(Adams).
Those theories were independently
produced
by scientists
who accepted that seat belts actually do reduce the likelihood that the
wearer would be a fatality, but recognized that
the evidence for that proposition did not meet the expectations
generated
by predictions calculated from seat belt effectiveness data.
The
problem was that fatality predictions did not include an assessment of
the interaction between a safety belt and driver behaviour.
That seatbelts might actually be a nett hazard might be concluded from the 10% increase in fatalities that accompanied the increase to 95% in seat belt wearing compliance in Hawaii between 1986 and 1990. These increased fatalities occurred during a period when the fatality rate was declining for the rest of the United States.
The bibliography is at the bottom of the data/evidence page.