A¢¢id. Anal. & Prey.
Vol. 5, PO- 30"7-370.
P*rgarnon Press 1973.
Ptant.ed i~ Great Britain.
RESEARCH NOTE A NOTE ON DIFFERENCES BETWEEN URBAN-RURAL SPEED-LIMIT CATEGORISATIONS OF ROAD ACCIDENT
AND DATA
E. WELDON*
Research Group in Traffic Studies, University College London, London, England (Received 3 January 1973)
WHEN analysing road accident statistics, it has been common practice to regard roads having 30 and 40 m.p.h, speed limits as being, to a considerable extent, equivalent to roads in urban areas. However, an analysis of some of the differences, using data for Great Britain, shows that: (i) The ratio of the difference in lengths of urban roads and roads with speed limits, to length of urban roads, exceeds 11 per cent. (ii) The corresponding ratio in respect of vehicle flow, i.e. miles travelled per day, is about 5 per cent. (iii) Similar ratios in respect of fatal accidents exceed 10 per cent for those involving two cars, 43 per cent for one car but no pedestrian, and 51 per cent for one car and a pedestrian. Corresponding ratios exceed 33, 44 and 57 per cent for serious accidents. The equivalencing of certain speed limits with an urban classification stems from some practical difficulties of comparing data from different sources, and from the practice to categorise accidents in terms of some characteristics of the road, e.g. whether the road is in town or in the country, carries heavy or light traffic, and whether the traffic it carries tends to be fast-moving or otherwise. An early example of this is in the report on fatal road accidents in 1933 (Ministry of Transport, 1934), which distinguished roads on a built-up/ non-built-up dichotomy, corresponding to the urban-rural status of the Local Authority in whose area the particular road is situated. Many authors have followed this in discussing road networks and traffic flows (Tanner, Johnson and Scott, 1962; Timbers, 1968), and also road traffic accidents (Smeed, 1960, 1961; Johnson, 1969, 1970). However much of the reported accident data is differentiated in terms of speed limits as "built-up" and " n o n built-up" types of area (Smeed, 1960, 1961; Department of the Environment, 1971). Comparisons, between urban-rural and speed limit based typologies, have been tabulated by Tanner, Johnson and Scott (1962) both for the length of the road network and for the daily traffic flow in Great Britain. They point out that in general speed limits depend on the presence or absence of street lighting, and also that the urban-rural and the speed limit dichotomies do not quite coincide. Nevertheless, their figures suggest that the urban roads correspond approximately with roads having 30 and 40 m.p.h, speed limits, whilst rural roads similarly correspond to roads with speed limits of 50 m.p.h, and above. Taking the * Present address: University Teaching Methods Unit, Institute of Education, University of London, London, England. 367
368
E.~VELDON
" u r b a n " category as the frame of reference, Tables 1 and 2 show the percentage deviation D from the hypothesis, that the " u r b a n " a n d " b u i l t - u p " categories c o n t a i n the same sample of data.
TABLE 1. MILESOF ROAD(from Tanner et al., 1962) Urban Rural
56277 106292
With speed limit Without speed limit
Total
162569
Total
49875 112694 162569
D = 100 (56277--49875) _ I I . 4 ~ 56277 TABLE2. TRAFFICFLOW(THOUSANDSOF VEHICLESMILESPER DAY) (from Tanner et al., 1962) Urban 101246 Rural 87054 Total
With speed limit Without speed limit
188300
Total
96257 92043 188300
D = 100 (101246--96257) _ 4-9~ 101246
Similar c o m p a r i s o n s between these two typologies were made in respect of data comprising all the fatal a n d serious accidents in G r e a t Britain in 1968, which were reported to the police a n d recorded in the n a t i o n a l "Stats. 19" accident data. Specifically, the object was to determine the percentage deviation D between " l o w " speed limit a n d " u r b a n " accidents. Three categories of accident were e x a m i n e d : those involving (i) only two cars, (ii) single cars without pedestrians a n d (iii) single cars with pedestrians. The u r b a n - r u r a l criterion followed the usual classification by Local A u t h o r i t y status. T w o speed limit b a n d s were defined: the " l o w " band comprising roads with 40 m.p.h, limit or below, and the " h i g h " b a n d comprising roads with a 50 m.p.h, limit or above. T a k i n g the " u r b a n " category as the frame or reference, Tables 3 a n d 4 show the respective values of D.
TABLE 3. NO. OF FATAL ACCIDENTS,GREAT BRITAIN, 1968 (STARS. 19 DATA)
Car-car
1 Car (non-ped.)
I Car (ped.)
Urban Rural Total
209 498 707
191 424 615
806 701 1507
Low High Total
231 476 707
274 341 615
1223 284 1507
-- 10-5~
--43-5 ~
D
--51.7
Differences between urban-rural and speed-limit road accident data
369
TABLE4. No. OF SERIousACCIDENTS,GREATBRITAIN,1968 (STATS. 19 DATA) Car-car
1 Car (non-ped.)
1 Car (ped.)
Urban Rural Total
3779 5567 9346
2349 5400 7749
8433 5971 14404
Low High Total
5031 4315 9346
3404 4345 7749
13259 1145 14404
--33.1%
--44-9%
--57.2 ~
D
The comparison of data on the basis of the two dichotomies showed that for trat~c flows the respective differences were of the order of 5 per cent, and for the length of the road network they were about 11 per cent. However the corresponding differences for accident frequencies ranged from I0 to 57 per cent, so that the hypothesis that the "urban" and "built-up" categories contain the same population of data is not tenable. In respect of accident analysis, the urban-rural differentiation gives quite different results from a differentiation on the basis of speed limits. Furthermore, such differences were found to be greatest for those accidents involving pedestrians. The results also showed (see Tables 3 and 4) that the frequencies of one-car (pedestrian) accidents were consistently greater for roads classified as " u r b a n " and "low" than for "rural" and "high", and the frequencies of one-car (non-pedestrian) accidents smaller for " u r b a n " and "low" than for "rural" and "high". However for the car-car accidents, the greater number of fatalities occurred for "rural" and "high", whereas for the serious accidents the greater frequencies were associated with "rural" and "low". This suggests the possibility that a large number of non-fatal car-car accidents may be associated with rural roads in the " l o w " band of speed limits. Smeed (1960, 1961) points out that "roads on which speed limits were imposed by no means coincided with the town and country division". Tanner, Johnson and Scott (1962) suggest that a break-down in terms of speed limit, rather than urban-rural may be more valuable for some purposes; but they do not discuss this point any further. The results in this paper suggest the need to clarify more precisely what, in fact, is being differentiated by each of these dichotomies. The urban-rural concept would appear to distinguish macroscopically between " t o w n " and "country", whereas speed limit bands (provided the limits themselves are realistic) may be more directly related to the dynamic traffic and driving situations. The mere attachment of a label to a road does not in itself enhance the usefulness or value of the classification--indeed it may lead to stereotyping. There seems to be no a priori reason to limit the number of categories used to two, nor for restricting the number of criteria to one. However, there is always the need to define categories that have both validity and relevance to the particular object of the investigation. Acknowledgement--This research was supported from a grant by the Science Research Council.
REFERENCES DEPARTMENTOF THE ENVIRONMENT(1971) Road Accidents 1969. H.M.S.O. JOHNSOn, H. D. (1969) Rood Accident and Casualty Rates Based on the 1300 Point Census. Road Research Laboratory (M.O.T.), Crowthorne.
370
E. WELDON
Jo~rysoy, H. D. (1970) Road Accident and Casualty Rates in 1968. Road Research Laboratory (M.O.T.), Crowthorne. MrNIs'rR'~ OF Ta.ANSPOe,T (1934) Report on Fatal Road Accidents which Occurred during the Year 1933. H.M.S.O., London. SS~ED, R. J. (1960) The influence of speed and speed regulations on traffic flow and accidents. Rds. & Rd. Constr., 38, 393-397; ibid. (1961), 39, 15-24. TA,'~,~R, J. C., JoI-~soy, H. D. and Sco'rl', J. R. (1962) Sample Survey o f the Roads and Traffic o f Great Britain. Road Research Technical Paper No. 62. H.M.S.O., London. TtsoEv,s, J. A. (1968) Traffic Survey at 1300 Sites. Research Laboratory (M.O.T.), Crowthorne.