Road safety risk evaluation by means of ordered weighted averaging operators
and expert knowledge
, Da Ruan
, Tom Brijs
, Geert Wets
, Koen Vanhoof
Hasselt University, Transportation Research Institute IMOB, Wetenschapspark 5 bus 6, 3590 Diepenbeek, Belgium
Belgian Nuclear Research Centre, Boeretang 200, 2400 Mol, Belgium
Available online 3 August 2009
Road safety indicators
Ordered weighted averaging operators
The road safety performance of countries is conducted by combining seven main risk indicators into one
index using a particular weighting and aggregation method. Weights can be determined with respect to
the assumed importance of the indicator, whereas aggregation operators can be used to stress better per-
formances differently from worse performances irrespective of the indicator’s meaning. In this research,
both expert weights and ordered weighted averaging operators are explored, evaluated and integrated
resulting in a ranking of countries based on a road safety index.
Ó 2009 Elsevier B.V. All rights reserved.
Research on road safety has lately received a lot of attention in
trafﬁc science. Worldwide, an estimated 1.2 million persons are
killed in road accidents every year and as many as 50 million are in-
jured . Given this high number of casualties and the resulting
suffering and costs, road safety enhancing measures are urgently
needed. Better insight into the relative risk performance of coun-
tries is vital in this respect. Existing research has basically focused
on the identiﬁcation of risk factors leading to road accidents and
casualties (e.g., [2,3]). At the European level, seven main risk factors
have been agreed upon, namely alcohol and drugs, speed, protec-
tive systems, daytime running lights (DRL), vehicle, roads and trau-
ma management [4,5]. The risk factors are captured by quantiﬁable
safety performance indicators (e.g., the alcohol and drugs factor by
the share of road users with a blood alcohol content above the legal
limit). A safety performance indicator is deﬁned as any measure-
ment that is causally related to accidents and casualties and used
in addition to a count of accidents and casualties in order to indicate
the safety performance or understand the process that leads to acci-
dents . Indicator values can then be compared across countries
and key problem areas revealed. This makes it possible to select
appropriate measures that tackle the main risk aspects and conse-
quently enhance the level of road safety in a country.
Countries could be compared on each indicator individually.
However, given the large number of relevant road safety perfor-
mance indicators, the creation of an overall road safety perfor-
mance index – which is a combination of road safety performance
indicators – is valuable. One of the main advantages of an index
over a set of individual indicators is that the overall road safety pic-
ture is presented as the different risk factors are joined in this index.
Moreover, based on the index scores the countries’ ranking can be
set up representing the combined performance on essential road
safety risk indicators. Similar to the common road safety ranking
based on the number of road fatalities per million inhabitants, an
easy comparison across countries can be made based on one score
and the relative position of a country in the ranking identiﬁed. By
contrast, the indicators of which the index consists help in taking
The combination of road safety performance indicators in an in-
dex is a methodologically intensive process consisting of various
steps . In general, an index results from the aggregation of a
set of indicator values and a set of weights. In this paper, we focus
on two essential methodological aspects, that is the weighting of
road safety indicators and the aggregating of road safety indicators.
As the concept of indicators and indexes is relatively new in the
road safety ﬁeld, not much attention has been paid to these topics
so far. Composing an index as the average over all indicator values
is simple yet inappropriate. The importance of two indicators
might signiﬁcantly differ from each other and the idea of full com-
pensation between good scores and bad scores might be unaccept-
able. Given the policy implications the index should be constructed
in a sound way. Therefore, the concept of weighting as well as the
ﬁeld of aggregation should be thoroughly evaluated.
The remaining of this paper is structured as follows: the indica-
tor data are presented in Section 2. Relevant weighting and aggre-
gation concepts are described and applied in Section 3. The results
in terms of the countries’ ranking are discussed in Section 4. This
paper closes with the main conclusions.
0950-7051/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved.
* Corresponding author. Tel.: +32 11 26 91 41; fax: +32 11 26 91 99.
E-mail address: firstname.lastname@example.org (E. Hermans).
Knowledge-Based Systems 23 (2010) 48–52
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/knosys