This website is no longer being updated. Please go to GOV.SCOT

Impact of the Road Network on Scotland's Accident Rates - Research Findings

DescriptionThis report was commissioned to explore the reasons behind the higher severity rate and to investigate whether the difference was caused by variations in the type of road network in both countries.
ISBN0 7480 6450 8
Official Print Publication Date
Website Publication DateDecember 29, 1998
Development Department Research Programme Research Findings No 36 (1997)
Impact of the Road Network on Scotland's Accident Rates

The MVA Consultancy

ISBN 0-7480-6450-8Publisher The Scottish officePrice £5.00
The number of fatal and serious road accidents as a proportion of all road accidents involving injury is approximately 40% higher in Scotland than in England and Wales. The Scottish Office commissioned The MVA Consultancy to explore the reasons behind this higher severity rate and to investigate whether the difference was caused by variations in the type of road network in both countries. As the severity variance had previously been identified as mainly a rural problem, the study was restricted to an analysis of accidents on trunk and principal roads in non-built up areas.
Main findings
  • The study was able to account for 60% of the severity difference between Scotland and England/Wales on the non-built up trunk and principal road network in terms of different distributions on the network of accident types.
  • Accidents on low flow roads are on average more severe than those on high flow roads, and estimated traffic flow at time of accident was found to account for 23% of the severity difference.
  • Similarly, accidents on links between junctions are more severe; Scotland had proportionately more accidents occurring on road links and this accounted for a further 20% of the severity difference.
  • Road type accounted for 10% of the severity difference, with Scotland having proportionally more single carriageway roads and more fatal and severe accidents occurring on them.
  • Other network variables tested, including bendiness, number of intersections and type of junction, did not contribute significantly to the differences in severity distribution between Scotland and England/Wales.
  • Non-network factors such as differences in weather and light conditions, response time of emergency vehicles and size and/or age of cars were not tested but may in part explain part of the residual 40% difference in severity rate.
Background
The number of fatal and serious road accidents as a proportion of all road accidents involving injury - 'the severity rate' - is approximately 40% higher in Scotland than in England/Wales.
Although this has been the case for many years, the reasons have not been fully understood. The Scottish Office commissioned an earlier study, undertaken by the Institute of Transport Studies (ITS) at The University of Leeds, to review the literature on the subject of accident causation and severity, to conduct preliminary analysis of such summarised accident data as were available and to examine the feasibility of investigating the issue further 1. One of the conclusions of this study was that, once pedestrian accidents were removed, the higher severity rate problem in Scotland was largely a rural one.
From their literature review, the ITS report listed a number of variables which have been identified by previous research as affecting accident severity. These were:
  • road type (motorway/dual carriageway/single carriageway);
  • level of traffic flow;
  • frequency of intersections;
  • average vehicle speed and speed variance;
  • horizontal curvature;
  • gradient;
  • carriageway width;
  • level and quality of lighting;
  • driver visibility/sight distance; and
  • median carriageway width.
It was hypothesised that differences in the distribution of these network variables between Scotland and England/Wales could explain the overall severity difference between the countries. For example, it was known that accidents on single carriageways were, on average, likely to be more severe than those on motorways or dual carriageways and Scotland had proportionally more single carriageway roads than England/Wales.
The ITS review also identified a number of non-network variables which might explain some of the severity difference. These included:
  • differing weather conditions;
  • differing light conditions;
  • differences in response time of emergency services;
  • differences in the level of drink-driving; and
  • differences in the size and/or age of cars.
It was recognised that further investigation of the severity difference was required and the MVA Consultancy was commissioned to carry out the research.
The MVA Study
A specific aim of this new work was to investigate the hypothesis that the severity rate difference was caused by variations in the distribution of network characteristics between the two countries. The study remit did not include detailed investigation of other factors that affected accident or severity rates.
When the ITS study assessed the availability of network and traffic data relating to different road types, it found that the most comprehensive and reliable information was maintained for trunk and principal roads. For this reason, and because the severity variance between Scotland and England/Wales had been identified as largely a rural one, the scope of this study was limited to the investigation of accident severity on trunk and principal roads where the speed limit was 50mph or over.
Methodology
The theoretical basis for the study method was the assumption that accident severity will be similar for any particular subset of accidents for both Scotland and England/Wales. For example, single vehicle accidents occurring on low flow dual carriageways in Scotland would, on average, have similar accident severity rates as single vehicle accidents occurring on low flow dual carriageways in England/Wales. Any network related difference in the severity rate for all accidents must therefore be largely due to differences in the distribution of subsets of accident. In other words, Scotland must have proportionally more of those types of accident which are most likely to involve serious injury or death. The research approach, therefore, was founded on identifying different distributions of accident types between the countries, not a detailed investigation of factors that affected accident severity.
Study Datasets
The study used information held on nearly 115,000 Accident Attendant Circumstances (STATS 19) forms for accidents in non-built up areas on trunk and principal roads between 1992 and 1994 inclusive. Accidents involving injuries to pedestrians only were excluded. Table 1 below shows the severity rates for the study dataset.
Table 1 - Severity Rates for Study Dataset
Country

All Accidents

Fatal and Serious Accidents

Severity Rate

England/Wales

102,595

22,817

0.222

Scotland

10,756

3,400

0.316

Total

113,351

26,217

0.231

It can be calculated from the above that Scotland had 1,008 'extra' fatal and serious accidents over a three year period than would be the case if Scotland had the same overall severity rate as England/Wales. It is these extra fatal and severe accidents that this study was seeking to explain.
The STATS 19 data were then enhanced with data from a digitised network base. This included information on road and junction type, road link bendiness, number of minor road intersections and the location of roads relative to major conurbations. Link sections within areas defined as built up were excluded from the analysis.
For each accident Annual Average Daily Flows (AADFs) from the Department of Transport's National Traffic Census Database were assigned using a Geographic Information System (GIS). The Department of Transport also supplied factors for conversion of AADFs to average hourly flows. Since the time of accident was recorded on the STATS 19 form, it was possible to estimate an average flow per lane at the time of accident using the appropriate factors, the STATS 19 data (for the number of carriageways and lanes) and the AADFs.
The above data preparation yielded two datasets for analysis:
  • a set of accident records augmented by flow, geometric and locational characteristics of the link on which they occurred; and
  • a link dataset, which included a count of the number of fatal, serious and slight accidents assigned to each link.
Results
The study method was to split, or partition, the total set of accidents in such a way that differences in accident severity could be due to differences in the distribution of accident types. Differences in distribution of accident types - each with associated severity rates - could be then be examined with reference to the relative characteristics of the two networks.
Given the differences in road design and layout, the most appropriate top level partition was road type (motorway/dual carriageway/single carriageway). Analysis of the data showed that accidents on single carriageways were, on average, more severe than those on motorways or dual carriageways. Scotland had proportionally more accidents occurring on single carriageway roads than England/Wales. This is not unexpected - Scotland has proportionally more single carriageway roads and proportionally more vehicle kilometres driven on this road type than England/Wales. These differences therefore account for some of the differences in severity rate.
Partitioning the data by road type accounted for 102.9 of the 1,008 'extra' fatal and serious accidents, i.e 10.2% of the overall difference in severity rate between Scotland and England/Wales can be explained by the fact that Scotland had proportionally more accidents occurring on single carriageway roads than in England/Wales.
Having split the accident data into different road types, the data were further partitioned within each road type into bands of estimated flow at the time of accident. The survey data showed that accidents occurring on low flow roads were on average more severe than those occurring on high flow roads. As Scotland had proportionally more accidents occurring at lower flows than England/Wales, this resulted in a higher severity rate than if the flow profiles for Scotland and England/Wales had been the same. Partitioning the data by estimated flow at time of accident accounted for a further 227.7 (22.6%) of the 1,008 'extra' fatal and serious accidents in Scotland.
Accidents on dual and single carriageways were further classified using the STATS 19 data according to whether they occurred at 'slow' junctions (signal controlled or roundabouts), 'fast' junctions (all other junction accidents) or on links (accidents not taking place at junctions). Accidents occurring at slow junctions were on average very much less severe than those occurring at fast junctions, which were in turn less severe than those which did not take place at a junction.
Scotland had proportionally more accidents which took place on links (ie not at a junctions) than England/Wales. This resulted in a higher severity rate than would be the case if the distribution of accidents at junctions for Scotland and England/Wales were the same. Partitioning the data by junction type accounted for a further 200.6 (19.9%) of the 1,008 `extra' fatal and serious accidents in Scotland.
Partitioning the data to account for different distributions of accidents by road type, estimated flow at time of accident and junction type between Scotland and England/Wales explained 531 or 53% of the overall severity difference. These partitions matched the three highest ranked hypotheses on reasons for differences in severity set out in the ITS report.
Four more variables were used to further partition the data, though the level of additional explanation each provided was relatively small. Table 2 below shows the summary of the total severity reductions associated with each partition in order of their contribution.
Table 1 - Summary of Total Effects of Partitions
Partitions

Severity Difference

% of Severity Difference

% of all Severity Explained

Estimated flow at
time of accident

227.7

22.6

38.5

Junction type

200.6

19.9

33.9

Road type

102.9

10.2

17.4

Location relative to conurbation

32.4

3.2

5.5

Trunk road status

15.8

1.6

2.7

Number of vehicles involved

6.4

0.6

1.1

Location of junction accidents

5.3

0.5

0.9

All partitions

591.02

58.6

100.0

Other network characteristics variables were tested but did not have significant differences in distribution between Scotland and England/Wales. These included:
  • road bendiness (expressed in degrees turned per km);
  • number of minor road intersections;
  • maximum deflection angle on a link; and
  • if a link lay in a tourist area or not.
The possibility that non-network effects such as those described earlier may be the cause of some of the severity difference between Scotland and England/Wales was suggested by the ITS feasibility study.
Conclusions
This research has provided some answers to the conundrum of Scotland's high accident severity rates in non-built up areas, and has demonstrated that over half of the difference can be accounted for by differences in the road network between Scotland and the rest of Great Britain. The earlier ITS study suggested that other factors such as light and weather conditions, response time by emergency services, and size or age of vehicles may also play a role in the different severity rates. However, accounting for these residual differences, ie those not explained by network factors, may prove to be more difficult.
1 Feasibility Study Into the Impact of Scotland's Road Network on Accident Rates (1994). Institute of Transport Studies, University of Leeds. The Scottish Office (unpublished).
Any queries regarding the statistical, technical or methodological content of this Report should be addressed to:-
Charlie Henderson
The MVA Consultancy
27 York Place
Edinburgh
EH1 3HP
"Impact of the Road Network on Scotland's Accident Rates", the research report summarised in this Research Findings, is available priced £7.50.
Cheques should be made payable to The Stationery Office and addressed to:
The Stationery Office Bookshop,
Mail Order Department,
21 South Gyle Crescent,
Edinburgh EH12 9EB
Telephone: 0131-479 3141, or Fax: 0131-479 3142.

The report can also be ordered online from:www.thestationeryoffice.co.uk

Further copies of this Research Findings may be obtained from:
The Scottish Office Central Research Unit
Area 2J
Victoria Quay
Edinburgh
EH6 6QQ
Telephone: 0131-244 7560