SWOV Workshop, Haarlem, 16 November 2009
Challenge the future
DelftUniversity ofTechnology
Quantified road safety target- Is the effect sustainable?
S.C. Wong The University of Hong KongTony Sze Delft University of Technology
2Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
Structure of Presentation
• Introduction• Data• Method• Results• Discussions• Conclusions
3Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
IntroductionRoad safety strategy framework
Policy formulation Policy implementation
VisionObjectiveTargetAction plan
EvaluationQuantitative ModelingInstitutional frameworkFunding
• FOCUS: Development of objective-related, cost effective, and practical measures that contribute to target achievement
4Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
IntroductionAchievable and challenging road safety target
Ambition
Capability
Social acceptability
• When a target is too ambitous, it becomes disappointing with no proven effectiveness, and demotivates
• When a target is too ‘easy’, and can be achieved without a high level of implementation, complacency is induced
• THEREFORE, setting up an appropriate target that strikes a balance among ambition, capability and social acceptability is necessary
5Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
IntroductionPrevious work
• Wong, S.C., Sze, N.N., Yip, H.F. et al. (2006) Association between setting quantified road safety targets and road fatality reduction. Accident Analysis and Prevention 38, 997-1005.
1984 1986 1988 1990 1992 1994 1996 1998 2000
Finland
Netherlands
United Kingdom
Norway
Denmark
Sweden
New Zealand
Spain
Australia
Hungary
Poland
1983 1984 1985 1986 1987 1988 1989 1990
Short-term effect on average fatality
6Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
IntroductionCurrent perspectives
1984 1986 1988 1990 1992 1994 1996 1998 2000
Finland
Netherlands
United Kingdom
Norway
Denmark
Sweden
New Zealand
Spain
Australia
Hungary
Poland1982 1984 1986 1988 1990 1992 1994 1996
Long-term effect on trend of fatality
7Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
DataSetting road safety target in OECD countries
1971 1976 1981 1986 1991 1996 2001 2006
FinlandNetherlands
United KingdomNorway
DenmarkSweden
New ZealandSpain
AustraliaHungary
PolandUnited States
FranceAustriaCanada
Czech RepublicGreeceIreland
ItalyPortugal
SwitzerlandBelgium
GermanyJ apanKorea
Turkey
• 14 OECD countries first set quantified road safety target in 1970s – 1990s
• 7 of which further set a new target before the expiration of 1st target
• 16 OECD countries have no evidence of target setting till year 2001
8Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
DataFatality dataCountry Fatality definition Correction factor
Austria 3 days 1965-1982: +15%1983-1991: +12%
Czech Republic Until 1979: 24 hours Until 1979: +30%
France 1970-1992: 6 days 1970-1992: +9% 1993-2000: +5.7%
Greece Until 1995: 3 days Until 1995: +15%
Hungary Until 1975: 48 hours Until 1975: +20%
Italy 7 days +8%
Japan Until 1992: 24 hours Until 1992: +30%
Korea 3 days +15%
Portugal 24 hours +30%
Spain Until 1993: 24 hours Until 1993: +30%
Switzerland Until 1991: >30 days Until 1991: -3%
Turkey 24 hours +30%
Source of data for current study• International Road Traffic & Accident Database (IRTAD)• World Road Statistics (WRS)
9Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
DataSample and study period
1971 1976 1981 1986 1991 1996 2001 2006
FinlandNetherlands
United KingdomNorway
DenmarkSweden
New ZealandSpain
AustraliaHungary
PolandUnited States
FranceAustriaCanada
Czech RepublicGreeceIreland
ItalyPortugal
SwitzerlandBelgium
GermanyJ apanKorea
Turkey
Sample• 1st target set in OECD countries during the period 1973-1997
Study period for the effect of target setting• y = 1: year of target setting• y = Y: target year
or year preceding setting of new target
y=0 (reference)
10Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
DataSample and study period
1971 1976 1981 1986 1991 1996 2001 2006
FinlandNetherlands
United KingdomNorway
DenmarkSweden
New ZealandSpain
AustraliaHungary
PolandUnited States
FranceAustriaCanada
Czech RepublicGreeceIreland
ItalyPortugal
SwitzerlandBelgium
GermanyJ apanKorea
Turkey
1971 1976 1981 1986 1991 1996 2001 2006
FinlandNetherlands
United KingdomNorway
DenmarkSweden
New ZealandSpain
AustraliaHungary
PolandUnited States
FranceAustriaCanada
Czech RepublicGreeceIreland
ItalyPortugal
SwitzerlandBelgium
GermanyJ apanKorea
Turkey
y=0 (reference)
Sample• 1st target set in OECD countries during the period 1973-1997
Study period for the effect of target setting• y = 1: year of target setting• y = Y: target year
or year preceding setting of new target
11Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
DataComparison group
1971 1976 1981 1986 1991 1996 2001 2006
FinlandNetherlands
United KingdomNorway
DenmarkSweden
New ZealandSpain
AustraliaHungary
PolandUnited States
FranceAustriaCanada
Czech RepublicGreeceIreland
ItalyPortugal
SwitzerlandBelgium
GermanyJ apanKorea
Turkey
12Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
DataComparison group (cont’d)
1971 1976 1981 1986 1991 1996 2001 2006
FinlandNetherlands
United KingdomNorway
DenmarkSweden
New ZealandSpain
AustraliaHungary
PolandUnited States
FranceAustriaCanada
Czech RepublicGreeceIreland
ItalyPortugal
SwitzerlandBelgium
GermanyJ apanKorea
Turkey
13Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
DataTreatment group
1971 1976 1981 1986 1991 1996 2001 2006
FinlandNetherlands
United KingdomNorway
DenmarkSweden
New ZealandSpain
AustraliaHungary
PolandUnited States
FranceAustriaCanada
Czech RepublicGreeceIreland
ItalyPortugal
SwitzerlandBelgium
GermanyJ apanKorea
Turkey
14Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
Method
Year 1968 1969 1970 1971 1972 1973 1974 1975 1976
y -4 -3 -2 -1 0 1 2 3 4
Finland 939 1006 1055 1143 1156 1086 865 910 804
United Kingdom 6810 7365 7499 7699 7763 7406 6876 6366 6570
Norway 479 496 560 533 490 511 509 539 471
Denmark 1096 1190 1208 1215 1116 1132 766 827 857
Sweden 1262 1275 1307 1213 1194 1177 1197 1172 1168
Scaling factor of annual fatality number at year y,
where y = 0 denotes reference year (year preceding a target setting)
0
yy
nf n
Normalization of fatality data
15Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
Method
Year 1968 1969 1970 1971 1972 1973 1974 1975 1976
y -4 -3 -2 -1 0 1 2 3 4
Finland 0.812 0.870 0.913 0.989 1 0.939 0.748 0.787 0.696
United Kingdom 0.877 0.949 0.966 0.992 1 0.954 0.886 0.820 0.846
Norway 0.978 1.012 1.143 1.088 1 1.043 1.039 1.100 0.961
Denmark 0.982 1.066 1.082 1.089 1 1.014 0.686 0.741 0.768
Sweden 1.057 1.068 1.095 1.016 1 0.986 1.003 0.982 0.978
Normalization of fatality data
Scaling factor of annual fatality number at year y,
where y = 0 denotes reference year (year preceding a target setting)
0
yy
nf n
16Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
Regression analysis
• Historical fatality trend in the ‘before’ period
• Trend of log-transformed scaling factor, f and year, y,
where i = t denotes treatment group
i = c (c = 1, 2, 3, … C ) denotes comparison group
α denotes time-series trend of ln (f )
Method
Year 1978 1979 1980 1981 1982 1983 1984 1985
y -7 -6 -5 -4 -3 -2 -1 0
United Kingdom 1.311 1.219 1.143 1.122 1.139 1.045 1.075
Greece 0.981 0.950 0.950 0.880 0.893 0.993 0.986
Italy 1.131 1.181 1.212 1.146 1.094 1.091 1.020
Qualification test for comparison group
ln i if y y
17Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
Regression analysis
• Difference in time-series trend between treatment & comparison group
Treatment group
Comparison group
where δ is a dummy variable to signify a comparison group
• Therefore, the regression equation for qualification test,
Method
Year 1978 1979 1980 1981 1982 1983 1984 1985
y -7 -6 -5 -4 -3 -2 -1 0
United Kingdom 1.311 1.219 1.143 1.122 1.139 1.045 1.075
Greece 0.981 0.950 0.950 0.880 0.893 0.993 0.986
Italy 1.131 1.181 1.212 1.146 1.094 1.091 1.020
i t
i t c c c
Qualification test for comparison group
ln , 0i t i i tf y y i t c
18Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
Method
Country f ln f y δ-Greece δ-Italy
United Kingdom 1.139 0.130 -1 0 0
United Kingdom 1.045 0.044 -2 0 0
United Kingdom 1.075 0.072 -3 0 0
… … … … … …
Greece 0.893 -0.113 -1 1 0
Greece 0.993 -0.007 -2 1 0
Greece 0.986 -0.014 -3 1 0
… … … … … …
Italy 1.094 0.090 -1 0 1
Italy 1.091 0.087 -2 0 1
Italy 1.020 0.020 -3 0 1
.. … … … … …
Qualification test for comparison group (cont’d)
ln 1, 2,i t i i t Greece Greece Italy Italyf y y y y y y
19Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
MethodQualification test for comparison group (cont’d)
0.7
0.8
0.9
1
1.1
1.2
1.3
1.4
1974 1976 1978 1980 1982 1984
United Kingdom Greece Italy
Expon. (United Kingdom) Expon. (Greece) Expon. (Italy)
20Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
Method
Year 1968 1969 1970 1971 1972 1973 1974 1975 1976
y -4 -3 -2 -1 0 1 2 3 4
Finland 0.812 0.870 0.913 0.989 1 0.939 0.748 0.787 0.696
United Kingdom 0.877 0.949 0.966 0.992 1 0.954 0.886 0.820 0.846
Sweden 1.057 1.068 1.095 1.016 1 0.986 1.003 0.982 0.978
Effectiveness evaluation
Regression analysis
• Regression equation,
where g (y ) = 1 for i = t & y > 0, and g (y ) = 0 if otherwises
h (y ) = a + b y
• Existence of target, g(y )
• Effect of target, h(y )
• a- absolute change in fatality number; b- change in the trend of fatality number
ln i if y y g y h y
21Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
Method
Country f ln f y g (y) h (y)
Finland 0.790 -0.236 -9 0 -
Finland 0.834 -0.182 -8 0 -
… … … … … …
Finland 0.989 -0.011 -1 0 -
Finland 1 0 0 0 -
Finland 0.939 -0.062 1 1 a + b
Finland 0.748 -0.290 2 1 a + 2b
… … … … … …
Finland 0.696 -0.363 4 1 a + 4b
Norway 0.757 -0.278 -9 0 -
… … … … … …
Norway 1.043 0.042 1 0 -
Norway 1.039 0.038 2 0 -
… … … … … …
Effectiveness evaluation (cont’d)
ln , 1,..., 2, 1,0,1,2, ,i i i if y y g y h y y a g y b g y y y m m k
22Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
MethodEffectiveness evaluation (cont’d)
ln i if y y g y h y h y a by
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
1978 1980 1982 1984 1986 1988 1990 1992
(I) a>0, b>0
(II) a>0, b<0
(III) a<0, b>0
(IV) a<0, b<0
a- absolute change in fatality numberb- change in the trend of fatality number
23Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
Results
• Immediate fatality reduction- consistent to the results of previous study• Desirable changes in the trend of fatality number
• Diminishing the rate of increase• Changing an increase to a decrease• Yielding a greater reduction
Effects of quantified road safety target in long termTreatment country h (y) Comparison group Previous study
Wong & Sze et al. (2006)a b
Finland -0.03 -0.10 11 N/A
United Kingdom 0.10 -0.02* 4 N/A
Norway -0.06 -0.07** 2 -4.7%
Denmark 0.03 -0.03^ 2 -37.7%**
Sweden 0.21** -0.03^ 8 +6.1%
New Zealand -0.04 -0.04^ 4 -8.8%*
Spain -0.31** -0.05* 1 -15.3%**
24Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
Discussion
Sustainable goal and target
• Rational, motivating and action guiding goal
• More modest and short-term targets as interim milestones
Comprehensive road safety management framework
• Sub-targets for various road safety indicators
• Periodoic progress review for different road entities and sub-targets
• Robust institutional framework
• Strong political foundation
25Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
Discussion (cont’d)
Target setting and fatality reduction
• Intermediate factors between road safety target and fatality reduction
• Factors, circumstances and developments that were anticipated prior
to target setting
26Is the effect of quantified road safety targets sustainable? S.C. Wong and N.N. Sze
Conclusions
Effectiveness of road safety target set in OECD countries (1970s-1990s)
• Regression approach effectiveness evaluation
• Trend of fatality number in a longer time span
• Desirable effect on the trend of fatality number is revealed
Future prospects: a sustainable road safety target
• Ambitiousness of target
• Existence of long-term committment
• Use of sub-targets
• Presence of detailed road safety program
• National income and expenditure on road safety measures