AN EXAMINATION OF THE "SEATBELT SAVINGS" CLAIM

C.I.P.S. Report No. 891-1

This report examines the Department of Transport's "200 lives and 7000 serious injuries" claim for seatbelt compulsion. Casualty statistics from the Dept. of Transport are used to examine the claim.

The conclusion is reached that this claim of annual life and injury savings due to seatbelt compulsion is difficult to substantiate.

Prepared for C.I.P.S. by Mr. G. Read.

Copyright C.I.P.S., August 1991.



AN EXAMINATION OF THE "SEATBELT SAVINGS" CLAIM

CONTENTS

1. BACKGROUND
1.1. General.
1.2. Period of Study.

2. STATISTICAL MANIPULATION.

3. SOURCE DATA.

4. METHOD OF EXAMINATION.

5. CASUALTY RATES, 1976-1982 AND 1983-1989.

6. DEDUCTIONS FROM TABLES 1 AND 2.
6.1. Table 1.
6.2. Table 2.

7. CONCLUSIONS FROM TABLES 1 AND 2.
7.1. Table 1.
7.2. Table 2.

8. DISCUSSION OF DEPT. OF TRANSPORT'S CLAIM.

9. REPORT CONCLUSION.

TABLE 1. Road Casualty annual averages by road user group

TABLE 2. Road Casualty annual averages by "belt" and "non-belt" groups.



AN EXAMINATION OF THE "SEATBELT SAVINGS" CLAIM

1.BACKGROUND.

1.1. General.

It is claimed by the Dept. of Transport (DTP) that seatbelt compulsion saves "200 lives and 7000 serious injuries" annually. Seatbelt compulsion was introduced in 1983, and applied to front-seat occupants of cars and light goods vehicles (LGV). Rear seat compulsion for children under 14 was introduced in 1989. In 1991, compulsion was extended to all car occupants and to taxi passengers.

This report examines the DTP claim in the light of road casualty statistics published by the DTP itself.

1.2. Period of Study.

Published official road casualty figures were, at the time of writing, only available up to 1989: this covered 7 years of seatbelt compulsion (1983 to 1989). Rear seat belt compulsion was therefore excluded from the study. The figures covering 1983 to 1989 were compared with the figures which covered 1976 to 1982, i.e. the 7 years preceding belt compulsion.

2. STATISTICAL MANIPULATION.

No statistical manipulations or modifications have been made to the DTP published casualty figures: only plain averaging has been used. This approach has been used because

(a) Many statistical manipulations involve a degree of "estimation", which can be held to be but a polite name for guessing.

(b) The vast majority of the population, from M.P.s to "bus-queue opinion", has a great mistrust of statistically manipulated figures. It was, we believe, Winston Churchill who said "There are three degrees of untruth; lies, damned lies, and statistics - in that order".

3. SOURCE DATA.

The accident injury figures used are taken unchanged from the DTP's annual publication "Road Accidents Great Britain (RAGB)" The figures in RAGB are assembled from the STATS19 forms which police officers fill in at the scenes of road traffic accidents. Specific RAGB sources used are:-
(a) RAGB 1986, page 84, Table 5, "Casualties by road user type and severity, 1976 to 1986"
(b) RAGS 1989, Page 69, Table 5, "Casualties by road user type and severity, 1979 to 1989".

Table 5 of RAGB gives year-by-year figures of casualties in respect of groups of road users (e.g. car drivers, bus passengers, etc.) Casualties are listed in 3 levels of severity, i.e.:-
(a) K (killed)
(b) KSI (killed and severely injured)
(c) All severities.

4. METHOD OF EXAMINATION.

Two seven-year periods were examined, 1976-1982 and 1983-1989 The year 1983 was when car and LGV front-seat belt compulsion was introduced. As in any road user group actual casualty figures vary from year to year, average annual figures were calculated for each of the two periods. These were calculated for both K and KSI categories.

The actual numerical changes in averages between the two periods are calculated, as are the percentage differences. These changes are compared with the DTP's claim figures.

5. CASUALTY RATES, 1976-1982 AND 1983-1989.

The calculated averages referred to in Para. 4 are shown in Table I. Also shown are the numerical and the percentage differences between the averages for the two periods.

Table 1 shows that the only group to show an average annual casualty rise for the 1983-1989 period was pedal cyclists KSI. All other groups showed significant decreases compared with the figures for the 1976-1982 period.

A further comparison is made by Table 2. This table compares the figures for those affected by belt compulsion with the figures for those not so affected. A difficulty arises in this comparison because those in rear seats were not subject to belt compulsion. Assumptions could be made as to the percentage of car occupants unbelted. Trial calculations assuming 1/3 of car passengers were not belted showed that such assumptions made at most only 1% difference to the figures. Such assumptions were therefore not made.

6. DEDUCTIONS FROM TABLES 1 AND 2.

6.1. Table 1.
The annual average casualty rates for all groups of road users except cyclists KSI have dropped significantly for the 1983-1989 period as compared with the 1976-1982 period.

6.2. Table 2.
The annual average casualty rates for "belted" road users dropped (-11%K, -17% KSI) compared with the 1976-1982 period. The annual average casualty rate for other (i.e. "unbelted") road users has dropped by comparable percentages. These are -19%K and -15% KSI.

7. CONCLUSIONS FROM TABLES 1 AND 2.

7.1. Table 1.
The road user groups of Table 1 are diverse: the situations faced by a pedestrian, for example, differ greatly from those faced by an HGV driver. Nevertheless, all groups except one show significant casualty rate drops for 1983-1989 compared with 1976-1982. It is difficult to imagine any road safety regulation which could have produced such an overall effect. The thought is inescapable that there is some factor at work which is not of Government regulatory origin, a factor which has overall effect.

Breath testing started in 1967, so could reasonably be held to have effect in both periods. Seatbelt compulsion started in 1983, so could only have effect in the second period: even in that period, it could only have effect on the casualty rates of the affected road user groups (car and LGV) The "energy crisis" speed limits applied between 1973 and 1977, and therefore could only have effect in the first of the two time periods.

The overall annual average casualty drop for the second period accords with the results produced by two models. These are the Smeed and the Partyka mathematical models. The Smeed model relates casualties to the number of cars in relation to the size of the population. The Partyka model relates casualties to economic conditions. NOTE. Data on these models can be found in "Risk and Freedom", Adams, TPP, ISBN 0 948537 05 1.

7.2. Table 2.
Table 2 presents figures from DTP-published sources which can be compared with the oft-repeated DTP claim that "belt compulsion saves 200 lives and 7000 serious injuries annually".

Table 2 classifies the "belted" group as car and LGV occupants. The average annual death rate for this group is lower by 272 for the 1983-1989 period than for the 1976-1982 period. If the claim of "200 lives" is true, then it logically follows that all other road safety measures together were only responsible for saving (for this group) 72 lives annually. This is difficult to believe, and we feel that it would be strongly contested by those who operate anti-drink legislation.

In respect of serious injuries, Table 2 shows that the annual average KSI for the "belted" group is 6225 less for the second period than for the first. The figure of 6255 is 745 less than the DTP claim of "7000 serious injuries" annually saved by seatbelts. The claim is therefore not believable.

8. DISCUSSION OF DTP CLAIM.

If one rejects the possibility that the DTP is deliberately falsifying figures, the question arises as to how these figures have been arrived at. It is notable that the initial claims for belt compulsion savings were even greater, and were scaled down as time progressed. Examples are:-
(a) Wm. Rogers, Sec. State Transport (Hansard 22-3-79) "On the best available evidence ... compulsion could save up to 1000 lives and 10,000 injuries a year"
(h) TRRL Report, Grime, 1979. "Seatbelts reduce deaths of car occupants by at least forty percent"
(c) Hurst, Accident Analysis and Prevention, 1979. "Belt use reduces chances of fatal injury by about 83% for drivers and about 80% for front seat passengers"
(d) David Ennals, Sec. State Health (Hansard 13-1-81) "Not wearing a belt increases 6-fold a motorist's chance of being killed in an accident"
(e) TRRL Research Report 9, Scott & Willis, 1985. "The law was responsible for a saving in the 1st year of 500 fatalities, 6500 serious injuries, and 13000 slight casualties"

Most of these quoted claims were based upon various forms of statistical estimations of varying degrees of validity. The current "200/7000" claim may well be based upon similar estimation, possibly that used in the Durbin and Harvey report on seatbelt compulsion effects. The method used by that report took no account of the effects of anti-drink legislation, a fact admitted by its authors to the Royal Statistical Society.

As can be seen from Tables 1 and 2, any plain man's interpretation of the actual casualty figures for the 7 years before and after the introduction of belt laws show beyond doubt that the claims for casualty savings by seatbelts are not believable. It is notable that no other field of road safety legislation makes such claims: the section of RAGB 1989 on drink, for example (pages 20 to 25) carefully avoids figures which can be so construed. The same is true of DTP Accident Fact Sheet 3/90 on the same subject.

9. REPORT CONCLUSION

The public at large have neither the means nor the ability to check the validity of official statistical claims. So, such claims are widely accepted as true on the basis that "they must know best". This public attitude means that figures of dubious validity could be published with every chance of uncritical acceptance.

Whatever justification the DTP may offer for their figures, their claim does not appear to stand up to the test of simple arithmetic as understood by the average man. They might say "your evaluation is too simple", to which we would reply "why does such a thing have to be complicated to be true?"

If the DTP claim is correct, it then remains to be explained why the casualty drop for "unbelted" road user groups is comparable with the drop for "belted" road users. It also remains to be explained why all other road safety measures appear to have minimal effect upon car and LGV occupants.


TABLE 1: Road Casualty Annual Averages by Road User Group.

USER GROUP 1976-1982 1983-1989 CHANGE No. CHANGE%
Pedestrians
K 2125 1796 -329 -15%
KSI 20055 18533 -1522 -8%
Cyclists
K306289-17-6%
KSl52795576+297+6%
Motorcyclists
K1126795-331-29%
KSI2223216286-5946-27%
Car Drivers
K14331305-128-8%
KSI1950616919-2587-13%
Car Passenger
K991876-115-12%
KSI1503112074-2957-20%
Bus Drivers
K62-4-33%
KSI8858-30-34%
Bus Passenger
K3824-14-37%
KSI1032849-183-18%
LGV Drivers
K9781-16-16%
KSI14721069-403-27%
LGV Passenger
K5845-13-22%
KSI961653-308-32%
HGV Driver & Passenger
K8774-13-15%
KSI960783-177-18%
All
K63085306-1002-16%
KSI8654673071-13475-16%

SOURCE DATA: RAGB Table 5.


TABLE 2. Road Casualty Annual Averages, "Belt" & "Non-belt" Groups

USER GROUP 1976-1982 1983-1989 C8ANGE No. CHANGE%
Belted
K25792307-272-11%
KSI3697030715-6255-17%
Other
K36882980-708-19%
KSI4964642085-7561-15%

DATA SOURCE: RAGB Table 5.


Return to C.I.P.S. home page