This article was downloaded by: [University of Nebraska, Lincoln] On: 04 April 2015, At: 20:26 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Traffic Injury Prevention Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/gcpi20
Demonstrating a Correlation Between the Maturity of Road Safety Practices and Road Safety Incidents a
a
Luis Amador & Christopher Joseph Willis a
Concordia University, Building, Civil and Environmental Engineering, Montreal, Quebec, Canada Accepted author version posted online: 27 Sep 2013.Published online: 27 May 2014.
Click for updates To cite this article: Luis Amador & Christopher Joseph Willis (2014) Demonstrating a Correlation Between the Maturity of Road Safety Practices and Road Safety Incidents, Traffic Injury Prevention, 15:6, 591-597, DOI: 10.1080/15389588.2013.845753 To link to this article: http://dx.doi.org/10.1080/15389588.2013.845753
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Traffic Injury Prevention (2014) 15, 591–597 C Taylor & Francis Group, LLC Copyright ISSN: 1538-9588 print / 1538-957X online DOI: 10.1080/15389588.2013.845753
Demonstrating a Correlation Between the Maturity of Road Safety Practices and Road Safety Incidents LUIS AMADOR and CHRISTOPHER JOSEPH WILLIS Concordia University, Building, Civil and Environmental Engineering, Montreal, Quebec, Canada
Downloaded by [University of Nebraska, Lincoln] at 20:26 04 April 2015
Received 16 August 2013, Accepted 14 September 2013
Objective: The objective of this study is to demonstrate a correlation between the maturity of a country’s road safety practices and road safety incidents. Methods: Firstly, data on a number of road injuries and fatalities for 129 countries were extracted from the United Nations Global Status on Road Safety database. These data were subdivided according to road safety incident and accident causation factors and normalized based on vehicular fleet (per 1000 vehicles) and road network (per meter of paved road). Secondly, a road safety maturity model was developed based on an adaptation of the concept of process maturity modeling. The maturity of countries with respect to 10 road safety practices was determined through the identification of indicators recorded in the United Nations Global Status of Road Safety Database. Plots of normalized road safety performance of the 129 countries against their maturity scores for each road safety practice as well as an aggregation of the road safety practices were developed. An analysis of variance was done to determine the extent of the correlation between the road safety maturity of the countries and their performance. In addition, a full Bayesian analysis was done to confirm the correlation of each of the road safety practices with injuries and fatalities. Results: Regression analysis for fatalities, injuries, and combined accidents identified maturity with respect to road safety practices associated with speed limits and use of alternative modes as being the most significant predictors of traffic fatalities. A full Bayesian regression confirms that there is a correlation between the maturity of road safety practices and road safety incidents. Conclusion: Road safety practices associated with enforcement of speed limits and promotion of alternative modes are the most significant road safety practices toward which mature countries have concentrated their efforts, resulting in a lower frequency of fatalities, injury rates, and property damage accidents. The authors argue that the use of gross domestic product (GDP) as a predictor of road safety incidents suffers from the presumptive assumption that the only criterion that matters is national income, therefore erroneously predicting that richer countries all perform approximately the same in fatalities and that developing nations are at various stages, with high variability and uncertainty in prediction. It is proposed that an aggregation of individual maturity scores from road safety practices will lead to a better indicator for policy because it connects externalities (fatalities/injuries) with intrinsic factors for which policies can be oriented. Keywords: accidents, correlation, fatalities, maturity model, road safety
Introduction It has been recognized that road traffic accidents are a major public health and development issue (Peden et al. 2004). In 2004 it was estimated that road crashes caused approximately 1.27 million deaths in addition to nonfatal injuries affecting between 20 and 50 million people (World Health Organization 2009). In developing countries, road traffic injuries account for approximately 1 percent of the gross national product (Asian Development Bank 2003; Jacobs et al. 2000; Mohan 2002), an amount that sometimes exceeds investment aid provided to these countries (Road Safety Task Force 2011). In addition, Managing Editor David Viano oversaw the review of this article. Address correspondence to Luis Amador, Concordia University, Building, Civil and Environmental Engineering, 1455 De Maisonneuve Blvd. West, Montreal, Quebec H3G 1M8 Canada. E-mail: [email protected]
road accidents place a burden on households. For example, according to Odero et al. (2003), in Kenya more than 75 percent of road traffic casualties are among the country’s economically productive adults. In addition, in Bangalore, Bangladesh, more than 50 percent of households that were considered as being poor after a crash that led to either death or serious injury would not have been classified as such before the crash (Aeron-Thomas et al. 2004). Clearly, road safety in general has a noticeable effect on the economic growth and development of a country. Recent studies have triggered an increase in public attention leading to the ultimate global declaration of a decade for action by the United Nations in 2011 (United Nations 2010). This call has been spurred by various past studies, in particular the World Health Organization/World Bank World Report on Road Traffic Injury Prevention. The aim of a global declaration of a decade for action by the United Nations is to stabilize and reduce fatalities (World Health Organization
Downloaded by [University of Nebraska, Lincoln] at 20:26 04 April 2015
592 2011), although that seems to signify an increase in injuries (light or severe). Researchers such as Koptis and Crooper (2005) have observed a good correlation between fatalities and gross domestic product (GDP) per capita from a global perspective. Koptis and Crooper (2005) found that in general there are few fatalities per capita in low-income nations (low traffic volumes), reaching a peak in middle-income countries, and then dropping in high-income nations. An explanation for this trend according to the International Road Assessment Program (2006) is that road safety schemes adopted during the 1970s by developed nations have contributed to the reductions in fatalities and improvements in road safety. It is suggested that to deal with poor road safety practices that lead to increased accidents, the problem must be dealt with by effective management from a leading agency that measures road safety via a funded strategy with measurable targets. This lead agency will have control in enforcing regulations, assessing road safety, educating drivers, and controlling the use of vehicles in good operating order. One way in which a lead agency of this kind will achieve the aforementioned is by firstly assessing the current state or capability of a country with respect to the implementation of road safety practices. Looking at other disciplines, such as manufacturing and construction, the concept of maturity modeling appears to be a suitable technique that can be used by lead agencies to assess a country’s capabilities with respect to road safety practices. As part of an effort to influence policy with respect to road safety, this article reports on a study that links the maturity of a country’s road safety practices with its road safety performance. The objectives of this article are as follows: (1) discuss the possible relationship between the elements of road safety and elements of collisions from a maturity perspective, (2) provide a general description of the road safety maturity model, (3) apply the road safety maturity model using data from the United Nations and World Health Organization databases, and (4) conduct a statistical analysis of road safety maturity and performance to identify significant indicators in support of the argument that there is a linkage between road safety maturity and road safety performance. In addition to adding to the existing body of knowledge on road safety, this article is intended to provide guidance to road safety policy makers in devising strategies to improve road safety.
Relationship Between the Elements of Road Safety and Elements of Collisions Most countries have laws that provide a clear definition of what constitutes a motor vehicle collision (MVC). For example, in Ontario, Canada, an MVC is defined as “any incident in which bodily injury or damage to property is sustained as a result of the movement of a vehicle, or of its load while a motor vehicle is in motion” (Ardal et al. 2002, p. 1). A general definition of an MVC is therefore a collision that involves a motorized vehicle crashing with either a motorized or nonmotorized user or an object. Based on this general definition, there are 3 main elements intrinsic to road safety: road features, human behavior, and vehicle characteristics (El-Basyouny and Sayed
Amador and Willis 2010). In this regard, MVCs are directly influenced by road features, human behavior, and vehicle characteristics, which collectively can be referred to as road safety elements. The 3 aforementioned road safety elements will influence the elements of collisions; that is, severity, likelihood, and exposure (Hauer 1997). The elements of collisions, will in turn determine the number of fatalities and the extent of injuries. An illustrative example of the aforementioned relationship is as follows: the human behavior of driving under the influence of alcohol (DUI) increases the likelihood of a collision. An increase in the likelihood of collisions due to DUIs will likely increase the number of road fatalities and injuries associated with a country. Clearly, any attempt to reduce the number of fatalities and injuries, as well as reduce the extent of injuries, must focus on improving road features, human behavior, and vehicle characteristics. One way to improve aspects of the 3 road safety elements is by improving the maturity of the practices associated with the road safety elements. Improving the maturity of the practices associated with road features, human behavior, and vehicle characteristics is done by firstly improving the consistency of enforcement or implementation of practices. In addition, practices must be periodically changed/improved to address evolving behaviors of road users, evolving features of roads, and evolving vehicle characteristics. Road safety practices address the factors that cause road fatalities and injuries. In this study, these factors are referred to as causation factors. Table 1 lists some of the road safety practices associated with the 3 road safety elements along with the additional element road safety management and their associated causation factors. The road safety practices considered in this study are highlighted in bold and can be traced back to the 3 road safety elements along with the additional element road safety management. The identification and selection of the road safety practices assessed in this study was based on the format of the data extracted the United Nations Global Status of Road Safety Database as well as on some assumptions and understandings. For instance, it is assumed that agencies can manage road safety to reduce severity by enforcing speed and alcohol regulations and the use of seat belt, helmet, and child restraint devices. It is understood that to reduce injuries and fatalities, road users can use safety hardware (Tighe et al. 2001), and to reduce the likelihood of injuries and fatalities agencies can promote alternate means of transportation (United Nations 2011). In addition, it is understood that road safety audits are the means to determine the need for safety hardware, identify inconsistencies in speed along a route, and determine accident hotspots (Zein 2004). Based on the aforementioned, the road safety practices assessed in this study address the following causation factors: seat belt use, child restraint, use of helmets, alcohol, existence of a road safety agency and strategy, use of alternate modes, drivers’ licensing and insurance, and the completion of road safety audits.
Overview of the Road Safety Maturity Model Maturity modeling, more specifically process maturity modeling, has its genesis in the software manufacturing industry
Road Safety Practices and Road Incidents
593
Table 1. List of road safety practices and injury and fatality causation factors addressed by them Road safety elements
Road safety practices
Injury and fatality causation factors
Road features
Warning signs are installed on high-speed roadways High-speed roadways are engineered with guardrails and collision barriers installed at strategic locations High-speed roadways have warning strips installed There are laws that regulate the use of helmets There are laws that address driving under the influence of alcohol Drug and alcohol tests are administered to drivers There are laws that regulate the use of seat belts There are laws that regulate the use of child restraints There are laws that regulate the installation of seat belts in vehicles There are laws that regulate the installation of airbags There are laws that regulate the maximum engine size and power There is a national road safety strategy
Excessive speed Excessive speed
Human behavior
Vehicle characteristics
Downloaded by [University of Nebraska, Lincoln] at 20:26 04 April 2015
Road safety management
There is promotion of alternate modes of transportation Road safety audits are done There are tests for driver licensing There are laws regulating vehicle insurance
Excessive speed No use of helmets Use of alcohol Use of alcohol No use of seat belts No use of child restraints No use of seat belts No use of air bags Unsafe vehicle engine size and power No formal strategy being implemented to deal with road injuries and fatalities No promotion and use of alternate modes of transportation No identification of inconsistent speed limits along a route, no identification of accident hotspots Ineffective licensing process —
Road safety practices highlighted in bold are those that were considered in this study.
(Finnemore et al. 2000) and has been credited to work done by Watts Humphreys and his colleagues at IBM in the early 1980s (Curtis 2001). The road safety maturity model developed and implemented in this study is based on an adaptation of the concept of process maturity exemplified in IBM’s capability maturity model. In particular, the road safety maturity model is patterned after the construction industry macro maturity model developed by Willis and Rankin (2012). The road safety maturity model perceives that every country engages in various road safety practices. A road safety practice is, in general, a behavior demonstrated by a majority of road users and is usually based on laws or regulations governing road use by pedestrians, motor vehicles, and cyclists. In addition, a road safety practice may be a regulation associated with vehicle characteristics and road features. It may also be a management practice used in the administration and coordination of road networks. The relationship between a road safety practice and the road safety goal of a country is a simple one. Basically, a road safety practice seeks to achieve a road safety objective, which, in turn, leads to the realization of a road safety goal. Road safety objectives deal with the severity, likelihood, and exposure of collisions. The level of implementation of a road safety practice will determine the extent to which its objective—for example, prevent injury and death of infants travelling in cars—is achieved. The extent to which each of the many road safety objectives is achieved will determine how well a country realizes its road safety goal, which may be to reduce road fatalities and extent of injuries. Based on the aforementioned, it is clear that the maturity of a country with respect to its road safety practices will serve as a leading or forward-looking indicator of its road safety performance. By looking at the maturity of road safety practices it should be possible to predict the lagging or after-the-fact road safety performance of a country. This leads to the assumption that to improve road safety performance, improvements in the maturity of road safety practices should first be effected.
The maturity of a country with respect to a road safety practice is determined based on the demonstrated capabilities of the country in implementing the road safety practice. The capability of a country with respect to a road safety practice determines the extent to which the road safety practice is institutionalized and made effective and is reflected by various outcomes/indicators. There are 3 capabilities concomitant with a road safety practice. The existence of a capability reflects a level of maturity, which is captured by a maturity score ranging from one third to 1. Obviously, if a road safety practice is nonexistent, the maturity score is zero. The score of one third represents that a country is immature with respect to a road safety practice. At the immature level a country is not capable of fully and consistently implementing a road safety practice; that is, the implementation of a road safety practice is ad hoc. The score of two thirds represents that a country is in a state of transitional maturity with respect to a road safety practice. At the transitional maturity level a country’s implementation of a road safety practice is standard and consistent. In addition, the road safety practice is singular and applies to all road users in all regions of the country. The score of 1 represents that a country is in a state of full maturity with respect to a road safety practice. The main indicator of full maturity is “different regions” of a country tailoring laws and regulations to suit their existing road-use culture and conditions; that is, the required behaviors of road users are slightly different between regions of a country. An example of this would be that seat belts are mandatory for all occupants in a vehicle for a particular region, while in another region seat belts are mandatory for only the driver and front seat passenger. The maturity of a road safety practice is improved via a stepwise progression. This means that a country should pass through the transitional maturity stage before it becomes fully mature with respect to a road safety practice. In this way, the road safety practice becomes institutionalized before it is subject to continuous improvement.
594
Downloaded by [University of Nebraska, Lincoln] at 20:26 04 April 2015
Method The method used to implement this study entailed firstly gathering road safety performance data; that is, road injuries and fatalities for a large sample of countries. The road safety performance of 129 countries was measured based on road accident data from the United Nations Global Status on Road Safety Database. The number of injuries and fatalities per country was subdivided based on the causation factors alcohol use, seat belt use, child restraint use, helmet use, speeding, and licensing. These data were then normalized based on vehicular fleet (per 1000 vehicles) and road network (per meter of paved road). The maturity of the countries with respect to the various road safety practices was determined through the identification of outcomes/indicators that were recorded in the United Nations Global Status of Road Safety Database. The database provided a concise summary of the current state of the countries with respect to laws and regulations, as well as other initiatives associated with the injury and fatality causation factors. Figure A1 (see online supplement) provides an illustrative example of the logic used to derive the maturity of a country with respect to the road safety practice “seat belt use.” It was firstly determined whether or not there was a seat belt law, for which there were 3 possible scenarios; that is, no law, a national law, and subnational laws. If there was no law, a maturity score of zero was given. If the law existed, it was determined whether it applied to either all occupants of the vehicle or to the driver and front passenger only. If the law applied to all occupants of the vehicle and was actively enforced, then a maturity score of 1 was given. If the law was not enforced a maturity score of one third was given. If the seat belt law was limited to the driver and front passenger and was actively enforced, a maturity score of two thirds was given. Likewise, if this law was not actively enforced, a maturity score of one third was given. If the law applied to all occupants of the vehicle and was actively enforced a maturity score of 1 was given. The second step to demonstrate the correlation between maturity of road safety practices and road safety performance was to plot the normalized road safety performance of the 129 countries against the maturity scores of the countries. Plots were done with respect to each road safety practice as well as an aggregation of the road safety practices. It should be noted that a logarithmic transformation was applied to improve the normality of the data. An analysis of variance was done to determine the extent of the correlation between the road safety maturity of the countries and their performance. This also determined the significance of each road safety practice with respect to injuries and fatalities. In addition, a full Bayesian analysis was done to confirm the correlation of each of the road safety practices with injuries and fatalities.
Amador and Willis lack of normality and homoscedasticity was observed. A logarithmic transformation was applied to the responses in order to improve normality. It was also learned that more sophisticated indicators for the normalization of the responses were required. It was observed that a natural logarithmic transformation of normalized fatalities per vehicular fleet and the extent of paved network [ln(fatalities/1000 vehicles/meter of paved road)] had the best fit to the data and resulted in normal and homoscedastic residual plots. This is shown in Figure A2 (see online supplement). Similar expressions were tested for injuries alone, and the total cost of accidents was normalized. As can be seen in Figure A2, they all returned good fits. Two models were developed in MINITABS; that is, a generalized linear model and a multivariate regression analysis. Generalized linear models are used if some predictors come from a qualitative variable and simple regression is used if all predictors are quantitative. Both analyses used the same response, [ln(fatalities/1000 vehicles/meter paved road)], with the 10 individual maturity scores as predictors. Regression analysis for fatalities, injuries, and combined accidents all identified establishing and enforcing speed limits and promotion of alternative modes as the most significant road safety practices with a 95 percent confidence interval (CI). Regression analysis for normalized fatalities also identified the road safety practice associated with road audits as a significant predictor at a 90 percent CI. The road safety practice addressing alcohol use has a slightly lower CI. Table 2 summarizes the statistic results of the regression analysis between individual maturity scores and normalized fatality exposure. Equation (1) presents the regression equation, which after removing the transformation becomes an exponential function of the independent variables and calibrating coefficients. These are shown in Table 3, which highlights that the maturity of road safety practices associated with the establishment and enforcement of speed limits and the promotion and use of alternate modes of transportation have the greatest impact on traffic fatalities. The more mature a country is with respect to these 2 practices, the greater the likelihood of there being a reduction in traffic fatalities. 1 5 ln(F) = −0.14 − 1.15x1 − 0.932x2 − x3 − x4 − 1.91x5 3 3 + 0.243x6 − 0.010x7 − 2.96x8 + 0.97x9 + 0.19x10 (1) where x1 = seat belt; x2 = child restraint; x3 = use of helmet; x4 = speed; x5 = alcohol; x6 = agency; x7 = strategy; x8 = promotion of alternative modes; x9 = road safety audits; x10 = driver’s licensing and insurance.
Analysis of Results Individual maturity scores were correlated with road fatalities and injuries. Several models for normalized fatalities and injuries (per capita, per vehicular fleet) were tested. However, a
Validation The classical statistical analysis presented in the previous section used P values as an indicator of significance and
Road Safety Practices and Road Incidents
595
Predictor Constant Seat belt (x1 ) Child restraint (x2 ) Helmet (x3 ) Speed (x4 ) Alcohol (x5 ) Agency (x6 ) Strategy (x7 ) Alternate modes (x8 ) Audits (x9 ) Licensing/insurance (x10 )
Coefficient
Standard error of coefficients
T
P
−0.143 −1.150 −0.932 −0.333 −1.666 −1.914 0.242 −0.010 −2.957 0.970 0.189
2.018 1.007 0.776 0.681 0.613 1.032 0.636 0.485 0.546 0.476 1.977
−0.07 −1.14 −1.20 −0.49 −2.71 −1.85 0.39 −0.02 −5.41 2.04 0.10
.944 .257 .233 .627 .008 .067 .699 .983 .000 .044 .924
16000 14000
Injuries per 100,000 people
Table 2. Summary statistics for natural logarithm of normalized fatalities
R2 = 0.4718
12000 10000 8000 6000 4000 2000 0 0
10000
20000
30000
40000
50000
60000 70000 80000
Bold values the most significant road safety practices based on regression analysis.
compared it with a predefined level of confidence. A full Bayesian linear regression was conducted in order to validate the results from the classical approach. Two hypotheses were tested in order to identify significant predictors. The null hypothesis set the mean of all coefficients (accompanying each individual maturity score) to zero, an indication of null contribution of each term to the model’s predictive capability. The prior specification was noninformative with a broad variance along with zero mean. The alternate hypothesis was that a particular coefficient was significant at a 95 percent CI. Probabilistic distributions of every coefficient were produced in order to refute or accept the null hypothesis. A full Bayesian regression was done to estimate the distributions from the observed data. Two hundred thousand iterations were run and a burn-in phase of 100,000 iterations was removed from the analysis. Figure A3 (see online supplement) shows the model specification for the full Bayesian regression with OpenBugs. Table A1 (see online supplement) shows that the road safety practice addressing the causation factor speed (a4 ) and the road safety practice addressing the use of alternate modes (a8 ) clearly refute the null hypothesis and are both significant at a 95 percent CI. Their values of 2.5 and 97.5 percent are both negative and away from the zero mean. Also significant is the road safety practice associated with safety audits (a9 ) and the road safety practice addressing Table 3. Summary statistics—natural log of normalized total cost − injuries and fatalities Predictor Constant Seat belt (x1 ) Child restraint (x2 ) Helmet (x3 ) Speed (x4 ) Alcohol (x5 ) Agency (x6 ) Strategy (x7 ) Alternate modes (x8 ) Audits (x9 ) Licensing/insurance (x10 )
Coefficient
Standard error of coefficient
T
P
−0.280 −0.780 −0.594 −0.168 −1.732 −1.569 0.215 0.309 −2.803 0.523 0.914
1.951 0.965 0.749 0.657 0.586 0.987 0.590 0.462 0.525 0.459 1.927
−0.14 −0.81 −0.79 −0.26 −2.95 −1.59 0.37 0.67 −5.34 1.14 0.47
.886 .421 .431 .799 .004 .116 .716 .506 .000 .258 .636
Bold values are road safety practices whose level of maturity have the greatest impact on traffic fatality.
35 30
Fatalities per 100000
Downloaded by [University of Nebraska, Lincoln] at 20:26 04 April 2015
GDP per Capita (US$)
25 20
R2 = 0.2702
15 10 5 0 0
10000
20000
30000
40000
50000
GDP (per capita US$)
Fig. 1. GDP per capita as an explanatory variable of road fatalities and injuries.
the use of alcohol (a5 ). This clearly demonstrates that there is a correlation between the maturity of road safety practices and road safety incidents.
Discussion Previous research done by Koptis and Crooper (2005) identified an intrinsic correlation between GDP per capita and fatalities per capita. It has been argued that for countries with very low GDP (low-income nations) fatalities per 100,000 people are on the order of 5 to 10. Such a low number of fatalities can be attributed to lower volumes of traffic from a modest intensity of economic activities. An alarming increase in road deaths has been observed in developing nations (on the order of 15 to 25). As countries become more industrialized and achieve a status of economic development, fatalities lower to similar levels as those in low-income nations. Figure 1 illustrates GDP per capita versus fatalities and injuries for 129 countries around the world used in this research. In its decade for action in road safety (2011–2020), the United Nations’ main objective was to lower fatalities from road accidents (collisions), therefore accepting a trade-off between deaths and injuries. The argument is that enhanced safety hardware can help prevent fatalities because it mitigates
Normalized Fatality-Exposure Versus Maturity
ln (fatalities/1000 vehicles/meter paved road)
4 2 0 0
1
2
3
4
5
6
7
8
9
10
-2 -4 -6
y = -0.918x + 1.6604 R2 = 0.3992
-8 -10
ln (total cost of accidents per 1000 vehicles per meter paved road)
an accident’s severity. In addition, injuries may even drop in severity to become property damage-only accidents. It is clear from a worldwide analysis that GDP exhibits a good correlation with injuries per capita (R2 = 0.47), which is in line with the interpretation of the data shown in Figure 1. This study has taken a different perspective. It is proposed that the combined maturity score is a better basis for policy. Three analyses were conducted to support this perspective. Fatalities and injuries were independently correlated with total maturity and they were also combined to produce a monetized indicator of total safety cost assuming $1.4 million per fatality and $85,000 per injury. All 3 analyses demonstrated a good correlation with the degree of maturity of the road safety system for a sample of 129 nations. In fact, all 3 support that increased maturity translated into a reduction in the number of fatalities and injuries. From the tested normalization factors it was found that the most accurate ones were those that took into consideration the size of vehicular fleet as a proxy of traffic volume in the country and the extent of paved roads. The approach used here differs from the use of GDP per capita to identify road externalities expectations at an international level, which follows a complex function (Figure 1). The latter (GDP per capita) suffers from the presumptive assumption that the only criterion that matters is national income, therefore erroneously predicting that richer countries all perform about the same in fatalities and that developing nations are at various stages with high variability and uncertainty in the prediction. The use of maturity overcomes these issues and returns not only a monotone function as shown in Figure 2 but also connects externalities (fatalities/injuries) with the intrinsic factors to which policies can be oriented. More mature countries do not necessarily happen to be richer but have built a higher degree of capability in terms of safety strategies, management, and operation. Both accident-based safety externalities and monetized cost yield similar trends. In Figure 2, all individual maturity scores are combined into an overall score. It is assumed that the relevance of each criterion carries equal weight. This assumption can be further refined; for instance, to account for cultural, environmental, and political differences, one could use weights to account for such differences across countries. Another important element is that of regional and national calibration. Regional organizations can use this approach to conduct a peer review comparison of their performance with that of other regions. Countries can use this method to encourage evidence-based implementation of policies; that is, by comparing road safety levels across states/provinces with dissimilar road safety maturity capabilities. In addition, international organizations can make use of this approach to guide resource allocation (both monetary aid and technical support) to countries competing for foreign aid or technical support or for conditional granting of support to bring about improvement in weak safety capabilities. The analysis done in this study reveals that from a global perspective, the road safety practices associated with establishing and enforcing speed limits and promotion of alternative modes are the most significant road safety practices on which mature countries have concentrated their efforts, resulting in a lower frequency of fatalities, injury rates, and property dam-
Amador and Willis
Total Maturity Score (base 10)
Normalized Total Cost of Safety Issues versus Maturity 4 2 0 0
1
2
3
4
5
6
7
8
9
10
-2 -4 -6
-8
2
y = -0.0847x + 0.1101x - 0.1003 R2 = 0.3894
-10
Total Maturity Score (base 10)
Normalized Injury Exposure versus Maturity 6 4
ln (fatalities/veh/meter road)
Downloaded by [University of Nebraska, Lincoln] at 20:26 04 April 2015
596
2 0 0
1
2
3
4
5
6
7
8
9
10
-2 -4 -6
y = -0.0979x2 + 0.4407x - 0.0332 R2 = 0.2752
-8
Total Maturity Score
Fig. 2. Maturity and fatalities, total cost of road safety and injuries.
age accidents. In addition, road safety practices associated with road audits and alcohol were found useful to further reduce fatalities. It is therefore recommended that developing countries should firstly focus on improving the maturity of the road safety practices associated with establishing and enforcing speed limits, promotion of alternative modes, completing road audits, and establishing and enforcing alcohol restrictions for drivers.
Downloaded by [University of Nebraska, Lincoln] at 20:26 04 April 2015
Road Safety Practices and Road Incidents This study is not without difficulties and weaknesses. A significant difficulty associated with this study is in estimating the monetary cost of collisions of various levels of severity, because this differed significantly between rich and poor countries. A significant weakness associated with this study is that all road safety practices were assumed to have the same level of importance with respect to reducing injuries and fatalities. Realistically, this is not the case and each road safety practice should have a weight of importance assigned to it. Based on the aforementioned, future work associated with this study includes (1) determining and comparing the relative importance of road safety practices in both developed and developing countries and (2) collecting and analyzing time series data of the maturity of road safety practices and observed accidents to further test the hypothesis that improvements in maturity of road safety practices will cause a reduction in road fatalities and extent of injuries. In summary, this study has answered the call made by the United Nations for a decade of action for road safety. Based on the general findings of this study, countries are provided with a more realistic basis on which to formulate road safety policies that are more in tune with their current realities.
Supplemental Material Supplemental data for this article can be accessed on the publisher’s website.
References Aeron-Thomas A, Jacobs GD, Sexton B, Gururaj G, Rahman F. The Involvement and Impact of Road Crashes on the Poor: Bangladesh and India Case Studies. Crowthorne: UK: Transportation Research Laboratory Limited; 2004. Published Project Report PPR 010. Ardal S, Berkowitz-Sandhu M, Woodward G. Driver Residence Analysis of Motor Vehicle Collisions: A Discussion Paper on the Potential of MTO Data to Address Population Health Issues. Toronto, ON, Canada: Central East Health Information Partnership; 2002. Asian Development Bank. Road Safety Guidelines for the Asian and Pacific Regions. Manila, Philippines: Asian Development Bank; 2003.
597 Curtis B. The Capability Maturity Model. Stamford, CT: Gartner Measurement; 2001. El-Basyouny K, Sayed T. Safety performance functions with measurement errors in traffic volume. Saf Sci. 2010;48:1339–1344. Finnemore M, Sashar M, Haigh R. Case studies in construction process improvement. Paper presented at: Arcom Construction Process Improvement Workshop; February 16, 2000; Loughborough, UK. Hauer E. Observational Before–After Studies in Road Safety: Estimating the Effect of Highway and Traffic Engineering Measures on Road Safety. Oxford: Pergamon Press; 1997. International Road Assessment Program. Road Deaths in Developing Countries: The Challenge of Dysfunctional Roads. Basingstoke, Hampshire, UK: John Mumford; 2006. Jacobs G, Aeron-Thomas A, Astrop A. Estimating Global Road Fatalities. Crowthorne, UK: Transport Research Laboratory Limited; 2000. Research Report RR445. Koptis E, Crooper M. Traffic fatalities and economic growth. Accid Anal Prev. 2005;37(1):169–178. Mohan D. Road safety in less motorized environments: Future concerns. Int J Epidemiol. 2002;31:527–532. Odero W, Khayesi M, Heda PM. Road traffic injuries in Kenya: magnitude, cause and status of intervention. Inj Control Saf Promot. 2003;10:53–61. Peden M, Scurfield R, Sleet D, et al. World Report on Road Traffic Injury Prevention. Geneva, Switzerland: World Health Organization; 2004. Road Safety Task Force. Making the Business Case for Road Safety Investment to achieve Sustainable Road Mobility. Michelin; 2011. White Paper. Tighe S, Li N, Falls LC, Haas R. Incorporating road safety into pavement management. Transp Res Rec. 2000;1699:1–10. United Nations. General Assembly adopts text proclaiming decade of action for road safety (2011–2020), aimed at reducing traffic related deaths and injuries. Sixty-fourth General Assembly Plenary [press release]. 2010. Available at: http://www.un.org/News/Press/docs/ 2010/ga10920.doc.htm. Accessed February 7, 2013. United Nations. Global Plan for Decade of Action for Road Safety 2011–2020. New York, NY: United Nations; 2011. Willis CJ, Rankin JH. The construction industry macro maturity model (CIM3): theoretical underpinnings. Int J Productivity Perform Manag. 2012;61(4):338–402. World Health Organization. Global Status Report on Road Safety: Time for Action. Geneva Switzerland: WHO Publications; 2009. World Health Organization. Global Plan for the Decade for Action for Road Safety 2011–2020. 2011. Available at: http://www.who.int/ roadsafety/decade of action/. Accessed February 7, 2013. Zein S. The Canadian Guide to In-Service Road Safety Reviews. Ottawa, ON, Canada: Transportation Association of Canada; 2004.
Traffic Injury Prevention Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/gcpi20
Demonstrating a Correlation Between the Maturity of Road Safety Practices and Road Safety Incidents a
a
Luis Amador & Christopher Joseph Willis a
Concordia University, Building, Civil and Environmental Engineering, Montreal, Quebec, Canada Accepted author version posted online: 27 Sep 2013.Published online: 27 May 2014.
Click for updates To cite this article: Luis Amador & Christopher Joseph Willis (2014) Demonstrating a Correlation Between the Maturity of Road Safety Practices and Road Safety Incidents, Traffic Injury Prevention, 15:6, 591-597, DOI: 10.1080/15389588.2013.845753 To link to this article: http://dx.doi.org/10.1080/15389588.2013.845753
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Traffic Injury Prevention (2014) 15, 591–597 C Taylor & Francis Group, LLC Copyright ISSN: 1538-9588 print / 1538-957X online DOI: 10.1080/15389588.2013.845753
Demonstrating a Correlation Between the Maturity of Road Safety Practices and Road Safety Incidents LUIS AMADOR and CHRISTOPHER JOSEPH WILLIS Concordia University, Building, Civil and Environmental Engineering, Montreal, Quebec, Canada
Downloaded by [University of Nebraska, Lincoln] at 20:26 04 April 2015
Received 16 August 2013, Accepted 14 September 2013
Objective: The objective of this study is to demonstrate a correlation between the maturity of a country’s road safety practices and road safety incidents. Methods: Firstly, data on a number of road injuries and fatalities for 129 countries were extracted from the United Nations Global Status on Road Safety database. These data were subdivided according to road safety incident and accident causation factors and normalized based on vehicular fleet (per 1000 vehicles) and road network (per meter of paved road). Secondly, a road safety maturity model was developed based on an adaptation of the concept of process maturity modeling. The maturity of countries with respect to 10 road safety practices was determined through the identification of indicators recorded in the United Nations Global Status of Road Safety Database. Plots of normalized road safety performance of the 129 countries against their maturity scores for each road safety practice as well as an aggregation of the road safety practices were developed. An analysis of variance was done to determine the extent of the correlation between the road safety maturity of the countries and their performance. In addition, a full Bayesian analysis was done to confirm the correlation of each of the road safety practices with injuries and fatalities. Results: Regression analysis for fatalities, injuries, and combined accidents identified maturity with respect to road safety practices associated with speed limits and use of alternative modes as being the most significant predictors of traffic fatalities. A full Bayesian regression confirms that there is a correlation between the maturity of road safety practices and road safety incidents. Conclusion: Road safety practices associated with enforcement of speed limits and promotion of alternative modes are the most significant road safety practices toward which mature countries have concentrated their efforts, resulting in a lower frequency of fatalities, injury rates, and property damage accidents. The authors argue that the use of gross domestic product (GDP) as a predictor of road safety incidents suffers from the presumptive assumption that the only criterion that matters is national income, therefore erroneously predicting that richer countries all perform approximately the same in fatalities and that developing nations are at various stages, with high variability and uncertainty in prediction. It is proposed that an aggregation of individual maturity scores from road safety practices will lead to a better indicator for policy because it connects externalities (fatalities/injuries) with intrinsic factors for which policies can be oriented. Keywords: accidents, correlation, fatalities, maturity model, road safety
Introduction It has been recognized that road traffic accidents are a major public health and development issue (Peden et al. 2004). In 2004 it was estimated that road crashes caused approximately 1.27 million deaths in addition to nonfatal injuries affecting between 20 and 50 million people (World Health Organization 2009). In developing countries, road traffic injuries account for approximately 1 percent of the gross national product (Asian Development Bank 2003; Jacobs et al. 2000; Mohan 2002), an amount that sometimes exceeds investment aid provided to these countries (Road Safety Task Force 2011). In addition, Managing Editor David Viano oversaw the review of this article. Address correspondence to Luis Amador, Concordia University, Building, Civil and Environmental Engineering, 1455 De Maisonneuve Blvd. West, Montreal, Quebec H3G 1M8 Canada. E-mail: [email protected]
road accidents place a burden on households. For example, according to Odero et al. (2003), in Kenya more than 75 percent of road traffic casualties are among the country’s economically productive adults. In addition, in Bangalore, Bangladesh, more than 50 percent of households that were considered as being poor after a crash that led to either death or serious injury would not have been classified as such before the crash (Aeron-Thomas et al. 2004). Clearly, road safety in general has a noticeable effect on the economic growth and development of a country. Recent studies have triggered an increase in public attention leading to the ultimate global declaration of a decade for action by the United Nations in 2011 (United Nations 2010). This call has been spurred by various past studies, in particular the World Health Organization/World Bank World Report on Road Traffic Injury Prevention. The aim of a global declaration of a decade for action by the United Nations is to stabilize and reduce fatalities (World Health Organization
Downloaded by [University of Nebraska, Lincoln] at 20:26 04 April 2015
592 2011), although that seems to signify an increase in injuries (light or severe). Researchers such as Koptis and Crooper (2005) have observed a good correlation between fatalities and gross domestic product (GDP) per capita from a global perspective. Koptis and Crooper (2005) found that in general there are few fatalities per capita in low-income nations (low traffic volumes), reaching a peak in middle-income countries, and then dropping in high-income nations. An explanation for this trend according to the International Road Assessment Program (2006) is that road safety schemes adopted during the 1970s by developed nations have contributed to the reductions in fatalities and improvements in road safety. It is suggested that to deal with poor road safety practices that lead to increased accidents, the problem must be dealt with by effective management from a leading agency that measures road safety via a funded strategy with measurable targets. This lead agency will have control in enforcing regulations, assessing road safety, educating drivers, and controlling the use of vehicles in good operating order. One way in which a lead agency of this kind will achieve the aforementioned is by firstly assessing the current state or capability of a country with respect to the implementation of road safety practices. Looking at other disciplines, such as manufacturing and construction, the concept of maturity modeling appears to be a suitable technique that can be used by lead agencies to assess a country’s capabilities with respect to road safety practices. As part of an effort to influence policy with respect to road safety, this article reports on a study that links the maturity of a country’s road safety practices with its road safety performance. The objectives of this article are as follows: (1) discuss the possible relationship between the elements of road safety and elements of collisions from a maturity perspective, (2) provide a general description of the road safety maturity model, (3) apply the road safety maturity model using data from the United Nations and World Health Organization databases, and (4) conduct a statistical analysis of road safety maturity and performance to identify significant indicators in support of the argument that there is a linkage between road safety maturity and road safety performance. In addition to adding to the existing body of knowledge on road safety, this article is intended to provide guidance to road safety policy makers in devising strategies to improve road safety.
Relationship Between the Elements of Road Safety and Elements of Collisions Most countries have laws that provide a clear definition of what constitutes a motor vehicle collision (MVC). For example, in Ontario, Canada, an MVC is defined as “any incident in which bodily injury or damage to property is sustained as a result of the movement of a vehicle, or of its load while a motor vehicle is in motion” (Ardal et al. 2002, p. 1). A general definition of an MVC is therefore a collision that involves a motorized vehicle crashing with either a motorized or nonmotorized user or an object. Based on this general definition, there are 3 main elements intrinsic to road safety: road features, human behavior, and vehicle characteristics (El-Basyouny and Sayed
Amador and Willis 2010). In this regard, MVCs are directly influenced by road features, human behavior, and vehicle characteristics, which collectively can be referred to as road safety elements. The 3 aforementioned road safety elements will influence the elements of collisions; that is, severity, likelihood, and exposure (Hauer 1997). The elements of collisions, will in turn determine the number of fatalities and the extent of injuries. An illustrative example of the aforementioned relationship is as follows: the human behavior of driving under the influence of alcohol (DUI) increases the likelihood of a collision. An increase in the likelihood of collisions due to DUIs will likely increase the number of road fatalities and injuries associated with a country. Clearly, any attempt to reduce the number of fatalities and injuries, as well as reduce the extent of injuries, must focus on improving road features, human behavior, and vehicle characteristics. One way to improve aspects of the 3 road safety elements is by improving the maturity of the practices associated with the road safety elements. Improving the maturity of the practices associated with road features, human behavior, and vehicle characteristics is done by firstly improving the consistency of enforcement or implementation of practices. In addition, practices must be periodically changed/improved to address evolving behaviors of road users, evolving features of roads, and evolving vehicle characteristics. Road safety practices address the factors that cause road fatalities and injuries. In this study, these factors are referred to as causation factors. Table 1 lists some of the road safety practices associated with the 3 road safety elements along with the additional element road safety management and their associated causation factors. The road safety practices considered in this study are highlighted in bold and can be traced back to the 3 road safety elements along with the additional element road safety management. The identification and selection of the road safety practices assessed in this study was based on the format of the data extracted the United Nations Global Status of Road Safety Database as well as on some assumptions and understandings. For instance, it is assumed that agencies can manage road safety to reduce severity by enforcing speed and alcohol regulations and the use of seat belt, helmet, and child restraint devices. It is understood that to reduce injuries and fatalities, road users can use safety hardware (Tighe et al. 2001), and to reduce the likelihood of injuries and fatalities agencies can promote alternate means of transportation (United Nations 2011). In addition, it is understood that road safety audits are the means to determine the need for safety hardware, identify inconsistencies in speed along a route, and determine accident hotspots (Zein 2004). Based on the aforementioned, the road safety practices assessed in this study address the following causation factors: seat belt use, child restraint, use of helmets, alcohol, existence of a road safety agency and strategy, use of alternate modes, drivers’ licensing and insurance, and the completion of road safety audits.
Overview of the Road Safety Maturity Model Maturity modeling, more specifically process maturity modeling, has its genesis in the software manufacturing industry
Road Safety Practices and Road Incidents
593
Table 1. List of road safety practices and injury and fatality causation factors addressed by them Road safety elements
Road safety practices
Injury and fatality causation factors
Road features
Warning signs are installed on high-speed roadways High-speed roadways are engineered with guardrails and collision barriers installed at strategic locations High-speed roadways have warning strips installed There are laws that regulate the use of helmets There are laws that address driving under the influence of alcohol Drug and alcohol tests are administered to drivers There are laws that regulate the use of seat belts There are laws that regulate the use of child restraints There are laws that regulate the installation of seat belts in vehicles There are laws that regulate the installation of airbags There are laws that regulate the maximum engine size and power There is a national road safety strategy
Excessive speed Excessive speed
Human behavior
Vehicle characteristics
Downloaded by [University of Nebraska, Lincoln] at 20:26 04 April 2015
Road safety management
There is promotion of alternate modes of transportation Road safety audits are done There are tests for driver licensing There are laws regulating vehicle insurance
Excessive speed No use of helmets Use of alcohol Use of alcohol No use of seat belts No use of child restraints No use of seat belts No use of air bags Unsafe vehicle engine size and power No formal strategy being implemented to deal with road injuries and fatalities No promotion and use of alternate modes of transportation No identification of inconsistent speed limits along a route, no identification of accident hotspots Ineffective licensing process —
Road safety practices highlighted in bold are those that were considered in this study.
(Finnemore et al. 2000) and has been credited to work done by Watts Humphreys and his colleagues at IBM in the early 1980s (Curtis 2001). The road safety maturity model developed and implemented in this study is based on an adaptation of the concept of process maturity exemplified in IBM’s capability maturity model. In particular, the road safety maturity model is patterned after the construction industry macro maturity model developed by Willis and Rankin (2012). The road safety maturity model perceives that every country engages in various road safety practices. A road safety practice is, in general, a behavior demonstrated by a majority of road users and is usually based on laws or regulations governing road use by pedestrians, motor vehicles, and cyclists. In addition, a road safety practice may be a regulation associated with vehicle characteristics and road features. It may also be a management practice used in the administration and coordination of road networks. The relationship between a road safety practice and the road safety goal of a country is a simple one. Basically, a road safety practice seeks to achieve a road safety objective, which, in turn, leads to the realization of a road safety goal. Road safety objectives deal with the severity, likelihood, and exposure of collisions. The level of implementation of a road safety practice will determine the extent to which its objective—for example, prevent injury and death of infants travelling in cars—is achieved. The extent to which each of the many road safety objectives is achieved will determine how well a country realizes its road safety goal, which may be to reduce road fatalities and extent of injuries. Based on the aforementioned, it is clear that the maturity of a country with respect to its road safety practices will serve as a leading or forward-looking indicator of its road safety performance. By looking at the maturity of road safety practices it should be possible to predict the lagging or after-the-fact road safety performance of a country. This leads to the assumption that to improve road safety performance, improvements in the maturity of road safety practices should first be effected.
The maturity of a country with respect to a road safety practice is determined based on the demonstrated capabilities of the country in implementing the road safety practice. The capability of a country with respect to a road safety practice determines the extent to which the road safety practice is institutionalized and made effective and is reflected by various outcomes/indicators. There are 3 capabilities concomitant with a road safety practice. The existence of a capability reflects a level of maturity, which is captured by a maturity score ranging from one third to 1. Obviously, if a road safety practice is nonexistent, the maturity score is zero. The score of one third represents that a country is immature with respect to a road safety practice. At the immature level a country is not capable of fully and consistently implementing a road safety practice; that is, the implementation of a road safety practice is ad hoc. The score of two thirds represents that a country is in a state of transitional maturity with respect to a road safety practice. At the transitional maturity level a country’s implementation of a road safety practice is standard and consistent. In addition, the road safety practice is singular and applies to all road users in all regions of the country. The score of 1 represents that a country is in a state of full maturity with respect to a road safety practice. The main indicator of full maturity is “different regions” of a country tailoring laws and regulations to suit their existing road-use culture and conditions; that is, the required behaviors of road users are slightly different between regions of a country. An example of this would be that seat belts are mandatory for all occupants in a vehicle for a particular region, while in another region seat belts are mandatory for only the driver and front seat passenger. The maturity of a road safety practice is improved via a stepwise progression. This means that a country should pass through the transitional maturity stage before it becomes fully mature with respect to a road safety practice. In this way, the road safety practice becomes institutionalized before it is subject to continuous improvement.
594
Downloaded by [University of Nebraska, Lincoln] at 20:26 04 April 2015
Method The method used to implement this study entailed firstly gathering road safety performance data; that is, road injuries and fatalities for a large sample of countries. The road safety performance of 129 countries was measured based on road accident data from the United Nations Global Status on Road Safety Database. The number of injuries and fatalities per country was subdivided based on the causation factors alcohol use, seat belt use, child restraint use, helmet use, speeding, and licensing. These data were then normalized based on vehicular fleet (per 1000 vehicles) and road network (per meter of paved road). The maturity of the countries with respect to the various road safety practices was determined through the identification of outcomes/indicators that were recorded in the United Nations Global Status of Road Safety Database. The database provided a concise summary of the current state of the countries with respect to laws and regulations, as well as other initiatives associated with the injury and fatality causation factors. Figure A1 (see online supplement) provides an illustrative example of the logic used to derive the maturity of a country with respect to the road safety practice “seat belt use.” It was firstly determined whether or not there was a seat belt law, for which there were 3 possible scenarios; that is, no law, a national law, and subnational laws. If there was no law, a maturity score of zero was given. If the law existed, it was determined whether it applied to either all occupants of the vehicle or to the driver and front passenger only. If the law applied to all occupants of the vehicle and was actively enforced, then a maturity score of 1 was given. If the law was not enforced a maturity score of one third was given. If the seat belt law was limited to the driver and front passenger and was actively enforced, a maturity score of two thirds was given. Likewise, if this law was not actively enforced, a maturity score of one third was given. If the law applied to all occupants of the vehicle and was actively enforced a maturity score of 1 was given. The second step to demonstrate the correlation between maturity of road safety practices and road safety performance was to plot the normalized road safety performance of the 129 countries against the maturity scores of the countries. Plots were done with respect to each road safety practice as well as an aggregation of the road safety practices. It should be noted that a logarithmic transformation was applied to improve the normality of the data. An analysis of variance was done to determine the extent of the correlation between the road safety maturity of the countries and their performance. This also determined the significance of each road safety practice with respect to injuries and fatalities. In addition, a full Bayesian analysis was done to confirm the correlation of each of the road safety practices with injuries and fatalities.
Amador and Willis lack of normality and homoscedasticity was observed. A logarithmic transformation was applied to the responses in order to improve normality. It was also learned that more sophisticated indicators for the normalization of the responses were required. It was observed that a natural logarithmic transformation of normalized fatalities per vehicular fleet and the extent of paved network [ln(fatalities/1000 vehicles/meter of paved road)] had the best fit to the data and resulted in normal and homoscedastic residual plots. This is shown in Figure A2 (see online supplement). Similar expressions were tested for injuries alone, and the total cost of accidents was normalized. As can be seen in Figure A2, they all returned good fits. Two models were developed in MINITABS; that is, a generalized linear model and a multivariate regression analysis. Generalized linear models are used if some predictors come from a qualitative variable and simple regression is used if all predictors are quantitative. Both analyses used the same response, [ln(fatalities/1000 vehicles/meter paved road)], with the 10 individual maturity scores as predictors. Regression analysis for fatalities, injuries, and combined accidents all identified establishing and enforcing speed limits and promotion of alternative modes as the most significant road safety practices with a 95 percent confidence interval (CI). Regression analysis for normalized fatalities also identified the road safety practice associated with road audits as a significant predictor at a 90 percent CI. The road safety practice addressing alcohol use has a slightly lower CI. Table 2 summarizes the statistic results of the regression analysis between individual maturity scores and normalized fatality exposure. Equation (1) presents the regression equation, which after removing the transformation becomes an exponential function of the independent variables and calibrating coefficients. These are shown in Table 3, which highlights that the maturity of road safety practices associated with the establishment and enforcement of speed limits and the promotion and use of alternate modes of transportation have the greatest impact on traffic fatalities. The more mature a country is with respect to these 2 practices, the greater the likelihood of there being a reduction in traffic fatalities. 1 5 ln(F) = −0.14 − 1.15x1 − 0.932x2 − x3 − x4 − 1.91x5 3 3 + 0.243x6 − 0.010x7 − 2.96x8 + 0.97x9 + 0.19x10 (1) where x1 = seat belt; x2 = child restraint; x3 = use of helmet; x4 = speed; x5 = alcohol; x6 = agency; x7 = strategy; x8 = promotion of alternative modes; x9 = road safety audits; x10 = driver’s licensing and insurance.
Analysis of Results Individual maturity scores were correlated with road fatalities and injuries. Several models for normalized fatalities and injuries (per capita, per vehicular fleet) were tested. However, a
Validation The classical statistical analysis presented in the previous section used P values as an indicator of significance and
Road Safety Practices and Road Incidents
595
Predictor Constant Seat belt (x1 ) Child restraint (x2 ) Helmet (x3 ) Speed (x4 ) Alcohol (x5 ) Agency (x6 ) Strategy (x7 ) Alternate modes (x8 ) Audits (x9 ) Licensing/insurance (x10 )
Coefficient
Standard error of coefficients
T
P
−0.143 −1.150 −0.932 −0.333 −1.666 −1.914 0.242 −0.010 −2.957 0.970 0.189
2.018 1.007 0.776 0.681 0.613 1.032 0.636 0.485 0.546 0.476 1.977
−0.07 −1.14 −1.20 −0.49 −2.71 −1.85 0.39 −0.02 −5.41 2.04 0.10
.944 .257 .233 .627 .008 .067 .699 .983 .000 .044 .924
16000 14000
Injuries per 100,000 people
Table 2. Summary statistics for natural logarithm of normalized fatalities
R2 = 0.4718
12000 10000 8000 6000 4000 2000 0 0
10000
20000
30000
40000
50000
60000 70000 80000
Bold values the most significant road safety practices based on regression analysis.
compared it with a predefined level of confidence. A full Bayesian linear regression was conducted in order to validate the results from the classical approach. Two hypotheses were tested in order to identify significant predictors. The null hypothesis set the mean of all coefficients (accompanying each individual maturity score) to zero, an indication of null contribution of each term to the model’s predictive capability. The prior specification was noninformative with a broad variance along with zero mean. The alternate hypothesis was that a particular coefficient was significant at a 95 percent CI. Probabilistic distributions of every coefficient were produced in order to refute or accept the null hypothesis. A full Bayesian regression was done to estimate the distributions from the observed data. Two hundred thousand iterations were run and a burn-in phase of 100,000 iterations was removed from the analysis. Figure A3 (see online supplement) shows the model specification for the full Bayesian regression with OpenBugs. Table A1 (see online supplement) shows that the road safety practice addressing the causation factor speed (a4 ) and the road safety practice addressing the use of alternate modes (a8 ) clearly refute the null hypothesis and are both significant at a 95 percent CI. Their values of 2.5 and 97.5 percent are both negative and away from the zero mean. Also significant is the road safety practice associated with safety audits (a9 ) and the road safety practice addressing Table 3. Summary statistics—natural log of normalized total cost − injuries and fatalities Predictor Constant Seat belt (x1 ) Child restraint (x2 ) Helmet (x3 ) Speed (x4 ) Alcohol (x5 ) Agency (x6 ) Strategy (x7 ) Alternate modes (x8 ) Audits (x9 ) Licensing/insurance (x10 )
Coefficient
Standard error of coefficient
T
P
−0.280 −0.780 −0.594 −0.168 −1.732 −1.569 0.215 0.309 −2.803 0.523 0.914
1.951 0.965 0.749 0.657 0.586 0.987 0.590 0.462 0.525 0.459 1.927
−0.14 −0.81 −0.79 −0.26 −2.95 −1.59 0.37 0.67 −5.34 1.14 0.47
.886 .421 .431 .799 .004 .116 .716 .506 .000 .258 .636
Bold values are road safety practices whose level of maturity have the greatest impact on traffic fatality.
35 30
Fatalities per 100000
Downloaded by [University of Nebraska, Lincoln] at 20:26 04 April 2015
GDP per Capita (US$)
25 20
R2 = 0.2702
15 10 5 0 0
10000
20000
30000
40000
50000
GDP (per capita US$)
Fig. 1. GDP per capita as an explanatory variable of road fatalities and injuries.
the use of alcohol (a5 ). This clearly demonstrates that there is a correlation between the maturity of road safety practices and road safety incidents.
Discussion Previous research done by Koptis and Crooper (2005) identified an intrinsic correlation between GDP per capita and fatalities per capita. It has been argued that for countries with very low GDP (low-income nations) fatalities per 100,000 people are on the order of 5 to 10. Such a low number of fatalities can be attributed to lower volumes of traffic from a modest intensity of economic activities. An alarming increase in road deaths has been observed in developing nations (on the order of 15 to 25). As countries become more industrialized and achieve a status of economic development, fatalities lower to similar levels as those in low-income nations. Figure 1 illustrates GDP per capita versus fatalities and injuries for 129 countries around the world used in this research. In its decade for action in road safety (2011–2020), the United Nations’ main objective was to lower fatalities from road accidents (collisions), therefore accepting a trade-off between deaths and injuries. The argument is that enhanced safety hardware can help prevent fatalities because it mitigates
Normalized Fatality-Exposure Versus Maturity
ln (fatalities/1000 vehicles/meter paved road)
4 2 0 0
1
2
3
4
5
6
7
8
9
10
-2 -4 -6
y = -0.918x + 1.6604 R2 = 0.3992
-8 -10
ln (total cost of accidents per 1000 vehicles per meter paved road)
an accident’s severity. In addition, injuries may even drop in severity to become property damage-only accidents. It is clear from a worldwide analysis that GDP exhibits a good correlation with injuries per capita (R2 = 0.47), which is in line with the interpretation of the data shown in Figure 1. This study has taken a different perspective. It is proposed that the combined maturity score is a better basis for policy. Three analyses were conducted to support this perspective. Fatalities and injuries were independently correlated with total maturity and they were also combined to produce a monetized indicator of total safety cost assuming $1.4 million per fatality and $85,000 per injury. All 3 analyses demonstrated a good correlation with the degree of maturity of the road safety system for a sample of 129 nations. In fact, all 3 support that increased maturity translated into a reduction in the number of fatalities and injuries. From the tested normalization factors it was found that the most accurate ones were those that took into consideration the size of vehicular fleet as a proxy of traffic volume in the country and the extent of paved roads. The approach used here differs from the use of GDP per capita to identify road externalities expectations at an international level, which follows a complex function (Figure 1). The latter (GDP per capita) suffers from the presumptive assumption that the only criterion that matters is national income, therefore erroneously predicting that richer countries all perform about the same in fatalities and that developing nations are at various stages with high variability and uncertainty in the prediction. The use of maturity overcomes these issues and returns not only a monotone function as shown in Figure 2 but also connects externalities (fatalities/injuries) with the intrinsic factors to which policies can be oriented. More mature countries do not necessarily happen to be richer but have built a higher degree of capability in terms of safety strategies, management, and operation. Both accident-based safety externalities and monetized cost yield similar trends. In Figure 2, all individual maturity scores are combined into an overall score. It is assumed that the relevance of each criterion carries equal weight. This assumption can be further refined; for instance, to account for cultural, environmental, and political differences, one could use weights to account for such differences across countries. Another important element is that of regional and national calibration. Regional organizations can use this approach to conduct a peer review comparison of their performance with that of other regions. Countries can use this method to encourage evidence-based implementation of policies; that is, by comparing road safety levels across states/provinces with dissimilar road safety maturity capabilities. In addition, international organizations can make use of this approach to guide resource allocation (both monetary aid and technical support) to countries competing for foreign aid or technical support or for conditional granting of support to bring about improvement in weak safety capabilities. The analysis done in this study reveals that from a global perspective, the road safety practices associated with establishing and enforcing speed limits and promotion of alternative modes are the most significant road safety practices on which mature countries have concentrated their efforts, resulting in a lower frequency of fatalities, injury rates, and property dam-
Amador and Willis
Total Maturity Score (base 10)
Normalized Total Cost of Safety Issues versus Maturity 4 2 0 0
1
2
3
4
5
6
7
8
9
10
-2 -4 -6
-8
2
y = -0.0847x + 0.1101x - 0.1003 R2 = 0.3894
-10
Total Maturity Score (base 10)
Normalized Injury Exposure versus Maturity 6 4
ln (fatalities/veh/meter road)
Downloaded by [University of Nebraska, Lincoln] at 20:26 04 April 2015
596
2 0 0
1
2
3
4
5
6
7
8
9
10
-2 -4 -6
y = -0.0979x2 + 0.4407x - 0.0332 R2 = 0.2752
-8
Total Maturity Score
Fig. 2. Maturity and fatalities, total cost of road safety and injuries.
age accidents. In addition, road safety practices associated with road audits and alcohol were found useful to further reduce fatalities. It is therefore recommended that developing countries should firstly focus on improving the maturity of the road safety practices associated with establishing and enforcing speed limits, promotion of alternative modes, completing road audits, and establishing and enforcing alcohol restrictions for drivers.
Downloaded by [University of Nebraska, Lincoln] at 20:26 04 April 2015
Road Safety Practices and Road Incidents This study is not without difficulties and weaknesses. A significant difficulty associated with this study is in estimating the monetary cost of collisions of various levels of severity, because this differed significantly between rich and poor countries. A significant weakness associated with this study is that all road safety practices were assumed to have the same level of importance with respect to reducing injuries and fatalities. Realistically, this is not the case and each road safety practice should have a weight of importance assigned to it. Based on the aforementioned, future work associated with this study includes (1) determining and comparing the relative importance of road safety practices in both developed and developing countries and (2) collecting and analyzing time series data of the maturity of road safety practices and observed accidents to further test the hypothesis that improvements in maturity of road safety practices will cause a reduction in road fatalities and extent of injuries. In summary, this study has answered the call made by the United Nations for a decade of action for road safety. Based on the general findings of this study, countries are provided with a more realistic basis on which to formulate road safety policies that are more in tune with their current realities.
Supplemental Material Supplemental data for this article can be accessed on the publisher’s website.
References Aeron-Thomas A, Jacobs GD, Sexton B, Gururaj G, Rahman F. The Involvement and Impact of Road Crashes on the Poor: Bangladesh and India Case Studies. Crowthorne: UK: Transportation Research Laboratory Limited; 2004. Published Project Report PPR 010. Ardal S, Berkowitz-Sandhu M, Woodward G. Driver Residence Analysis of Motor Vehicle Collisions: A Discussion Paper on the Potential of MTO Data to Address Population Health Issues. Toronto, ON, Canada: Central East Health Information Partnership; 2002. Asian Development Bank. Road Safety Guidelines for the Asian and Pacific Regions. Manila, Philippines: Asian Development Bank; 2003.
597 Curtis B. The Capability Maturity Model. Stamford, CT: Gartner Measurement; 2001. El-Basyouny K, Sayed T. Safety performance functions with measurement errors in traffic volume. Saf Sci. 2010;48:1339–1344. Finnemore M, Sashar M, Haigh R. Case studies in construction process improvement. Paper presented at: Arcom Construction Process Improvement Workshop; February 16, 2000; Loughborough, UK. Hauer E. Observational Before–After Studies in Road Safety: Estimating the Effect of Highway and Traffic Engineering Measures on Road Safety. Oxford: Pergamon Press; 1997. International Road Assessment Program. Road Deaths in Developing Countries: The Challenge of Dysfunctional Roads. Basingstoke, Hampshire, UK: John Mumford; 2006. Jacobs G, Aeron-Thomas A, Astrop A. Estimating Global Road Fatalities. Crowthorne, UK: Transport Research Laboratory Limited; 2000. Research Report RR445. Koptis E, Crooper M. Traffic fatalities and economic growth. Accid Anal Prev. 2005;37(1):169–178. Mohan D. Road safety in less motorized environments: Future concerns. Int J Epidemiol. 2002;31:527–532. Odero W, Khayesi M, Heda PM. Road traffic injuries in Kenya: magnitude, cause and status of intervention. Inj Control Saf Promot. 2003;10:53–61. Peden M, Scurfield R, Sleet D, et al. World Report on Road Traffic Injury Prevention. Geneva, Switzerland: World Health Organization; 2004. Road Safety Task Force. Making the Business Case for Road Safety Investment to achieve Sustainable Road Mobility. Michelin; 2011. White Paper. Tighe S, Li N, Falls LC, Haas R. Incorporating road safety into pavement management. Transp Res Rec. 2000;1699:1–10. United Nations. General Assembly adopts text proclaiming decade of action for road safety (2011–2020), aimed at reducing traffic related deaths and injuries. Sixty-fourth General Assembly Plenary [press release]. 2010. Available at: http://www.un.org/News/Press/docs/ 2010/ga10920.doc.htm. Accessed February 7, 2013. United Nations. Global Plan for Decade of Action for Road Safety 2011–2020. New York, NY: United Nations; 2011. Willis CJ, Rankin JH. The construction industry macro maturity model (CIM3): theoretical underpinnings. Int J Productivity Perform Manag. 2012;61(4):338–402. World Health Organization. Global Status Report on Road Safety: Time for Action. Geneva Switzerland: WHO Publications; 2009. World Health Organization. Global Plan for the Decade for Action for Road Safety 2011–2020. 2011. Available at: http://www.who.int/ roadsafety/decade of action/. Accessed February 7, 2013. Zein S. The Canadian Guide to In-Service Road Safety Reviews. Ottawa, ON, Canada: Transportation Association of Canada; 2004.
