Accident Analysis and Prevention 64 (2014) 86–91

Contents lists available at ScienceDirect

Accident Analysis and Prevention journal homepage: www.elsevier.com/locate/aap

The effect of traffic tickets on road traffic crashes Roni Factor ∗ School of Criminology, University of Haifa, Mt. Carmel, Haifa 31905, Israel

a r t i c l e

i n f o

Article history: Received 7 October 2013 Received in revised form 25 November 2013 Accepted 25 November 2013 Keywords: Road traffic accidents Traffic violations High-risk behaviors Socioeconomic status Distance traveled Enforcement

a b s t r a c t Road traffic crashes are globally a leading cause of death. The current study tests the effect of traffic tickets issued to drivers on subsequent crashes, using a unique dataset that overcomes some shortcomings of previous studies. The study takes advantage of a national longitudinal dataset at the individual level that merges Israeli census data with data on traffic tickets issued by the police and official data on involvement in road traffic crashes over seven years. The results show that the estimated probability of involvement in a subsequent fatal or severe crash was more than eleven times higher for drivers with six traffic tickets per year compared to those with one ticket per year, while controlling for various confounders. However, the majority of fatal and severe crashes involved the larger population of drivers who received up to one ticket on average per year. The current findings indicate that reducing traffic violations may contribute significantly to crash and injury reduction. In addition, mass random enforcement programs may be more effective in reducing fatal and severe crashes than targeting high-risk recidivist drivers. © 2013 Elsevier Ltd. All rights reserved.

1. Introduction Road traffic crashes exact a huge economic, social, and human toll. They are the eighth leading cause of death for all ages globally, with an average of more than 3000 mortalities per day (Lozano et al., 2012; World Health Organization, 2013). Hence, elucidating the underlying mechanisms behind traffic crashes in order to better prevent them is an important public health effort. Driving itself is by nature a dangerous activity, one that demands skill, attention, quick decision making, and on-the-go communication between drivers (Rothengatter, 1997; Wilmot and Khanal, 1999; Factor et al., 2011). Driver behavior is therefore an important aspect of road safety – and indeed, studies suggest that the vast majority of traffic crashes are caused by human factors (Shinar, 2007). Other studies conducted over the years have shown that various behaviors usually regarded as violations of traffic regulations – including driving under the influence of alcohol, speeding, failing to obey a red light or stop signal, and nonuse of restraint devices – are associated with an increased risk of road traffic crashes and of being injured when one occurs (Gebers and Peck, 2003; Elvik and Christensen, 2007; Goldenbeld et al., 2011; Factor et al., 2012). These associations have often been studied via driver samples – using self-reports, observations, and comparison groups – and evaluation studies (Macdonald et al., 2004; de Winter and Dodou, 2010; Stanojevic´ et al., 2013). However, most previous studies are marked by limitations that weaken their findings. These include the

∗ Tel.: +972 52 380 1911; fax: +972 77 201 8769. E-mail address: [email protected] 0001-4575/$ – see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.aap.2013.11.010

possibility of self-report bias; a limited number of observations; small sample sizes; limited geographical areas; limited time; or controlling for a limited number of confounders, if any (Parker et al., 1995; Blows et al., 2005; Ivers et al., 2009). The current study aims to overcome these limitations by analyzing the relationship between the number of traffic tickets received by drivers and their subsequent road traffic crashes, using a unique large, longitudinal, individual-level dataset. The dataset merges Israeli census data with data on tickets issued by the police and official data on road traffic crashes over a period of seven years, and so enables robust testing for a correlation between tickets issued and crashes while controlling for various demographic and socioeconomic variables, as well as distance traveled. The findings of the current analysis may enhance our understanding of how risky driving affects road traffic crashes – knowledge crucial for the development of effective interventions and policies to reduce road injuries. 2. Traffic violations and crashes Drivers seem to violate traffic regulations for different reasons. Even good drivers may succumb to poor planning, decision making, or concentration when they are hungry, tired, in a hurry, or ill-tempered. Other cases may involve sensation-seeking or risktaking, deliberate social nonconformism, or even acts of social resistance (Jonah, 1997; Rothengatter, 1997; Shinar, 2007; Factor et al., 2013b). Many studies suggest that traffic law violations, whatever their origin, are – at least at the aggregate level – among the key factors contributing to an increased risk of crashes (Parker et al., 1995;

R. Factor / Accident Analysis and Prevention 64 (2014) 86–91

ETSC, 1999; Sullman et al., 2002; Gebers and Peck, 2003; Ayuso et al., 2010; Stanojevic´ et al., 2013). For instance, Rajalin (1994) compared drivers who were involved in fatal crashes with a control group randomly selected from the Finnish drivers registry, and found a higher risk of involvement in crashes among drivers with previous traffic violations. In the Netherlands, Lourens et al. (1999) found that drivers who had received traffic tickets were significantly more involved in road traffic crashes than those without tickets at all distance traveled levels. Another Netherlands study found that vehicles whose drivers had committed more than one violation per year were involved in more road crashes than vehicles with only one violation per year (Goldenbeld et al., 2011). Blows et al. (2005) in a sample of drivers from the northern part of New Zealand, found an association between self-reported traffic violations and road crashes, with those reporting a higher frequency of traffic violations 2–4 times more likely to have been injured while driving. Ivers et al. (2009) in a non-representative Webbased cohort study of young drivers in Australia, calculated that self-reported risky driving behaviors were associated with a 50% increased risk of police-reported crashes, controlling for various factors. Finally, a study in British Columbia based on official data but without controlling for confounders suggests that there is a consistent increase in crashes with an increasing number of convictions (and crashes) (Chen et al., 1995). Beyond the general association between traffic violations and road traffic crashes, researchers have identified a number of specific driving offenses that have significant associations with the prevalence and severity of crashes. These include speeding (Cooper, 1997; Ayuso et al., 2010; Elvik et al., 2012), driving under the influence of alcohol and drugs (Ferrante et al., 2001; Macdonald et al., 2004; Bjerre and Thorsson, 2008; Vingilis and Wilk, 2008), nonuse of seat belts and safety restraint systems (Robertson, 1996; ETSC, 1999; Sivak et al., 2007), and failing to stop at a red light or to yield the right of way (Retting et al., 2003; Pai, 2011). Further evidence for an association between traffic violations and road traffic crashes comes from studies on the effect of enforcement on crashes and driver behavior. Indeed, the positive association between violations and crashes is one of the key assumptions guiding enforcement efforts (Shinar, 2007). Yannis et al. (2007) using aggregate spatial data from Greece, found that intensified enforcement was the main cause of improved road safety. Similarly, a study in Israel suggests a significant reduction is severe road traffic crashes and casualties on highly enforced roads compared to control roads (Hakkert et al., 2001). Stanojevic´ et al. (2013) show that a lack of enforcement results in greater rates of speeding, failure to use seat belts, driving under the influence of alcohol, engagement in both aggressive and ordinary violations, and involvement in risky situations. A closer look at the studies reviewed above reveals that most fall into one of two types: driver samples and evaluation studies. The first type, driver samples, can be further divided into three categories: (a) driver self-reports (Reason et al., 1990; Sullman et al., 2002; Blows et al., 2005; de Winter and Dodou, 2010), (b) observations (Stanojevic´ et al., 2013), and (c) comparisons of traffic violators and/or drivers involved in road traffic crashes with a matched control group of “regular” drivers (Rajalin, 1994; Ferrante et al., 2001; Macdonald et al., 2004). The second type, evaluation studies, examines changes in traffic violations and road traffic crashes following an intervention, such as an enforcement or device initiative (Hakkert et al., 2001; Yannis et al., 2007). However, both types of studies are vulnerable to several limitations that may weaken their findings. These limitations include, among others, reliance on self-reports, small sample sizes, a focus on a relative small geographic area, a limited time of study, controlling for a limited number of confounders, or any combination of these. Possibly as a result of such limitations, some research suggests that the

87

correlation between traffic violations and crashes is complex and relatively low (Gebers and Peck, 2003). The following section describes how the unique dataset of the current study was used to better elucidate the link between the number of tickets received by drivers and subsequent road traffic crashes. 3. Methods 3.1. Dataset The current study uses a unique individual-level dataset that merges Israeli census data with road traffic crash data and data on traffic tickets issued to drivers. For each driver who was asked to respond to the extended 1995 census questionnaire, which was administrated to a representative sample of 20% of the Israeli population, the census data were merged with individual-level information on traffic tickets issued and involvement in road traffic crashes from the years 2002 to 2008 (the different data sources were matched using each driver’s national identification number). This process created, in essence, a set of panel data that made it possible to trace a fifth of the Israeli driver population over a period of 13 years, to identify those drivers who received traffic tickets and/or were involved in road traffic crashes from 2002 to 2008, and to estimate the effect of traffic tickets on crashes while controlling for various demographic and socioeconomic variables. The dataset was obtained especially for the current research from the Israel Central Bureau of Statistics, following approval of an ethics committee which determined that it did not violate drivers’ right to privacy or confidentiality. The quality of the matching process was tested in three stages. The results showed that more than 92% of the crash and traffic ticket records were linked to the census data (the missing records were due either to errors in one or the other data source or to the fact that some of the drivers were new immigrants who did not take part in the census). No meaningful differences were found between the linked and unlinked drivers. Finally, an additional quality check found that among the linked cases, 97.6% fully matched by gender and year of birth, and a further 2.3% matched by gender or birth year only (in the current analyses, only those cases with a perfect match were used). Thus, these tests suggest that the merging procedure was of relatively high quality. The dataset includes 409,051 drivers who received 830,763 traffic tickets (not including parking tickets) and were involved in 22,562 crashes that resulted in casualties, among them 2401 fatal or severe crashes. 3.2. Variables In the following analyses only fatal and severe crash data were used, because crashes that lead to light injuries appear to be underreported (Elvik and Vaa, 2004; Evans, 2004). The road traffic crash variable was dichotomized to represent whether the driver was (1) or was not (0) involved in a fatal or severe crash during the study period. The variable was dichotomized since only 0.005% of the sample had more than one fatal or severe crash during the research period. The traffic tickets variable comprised tickets for all types of violations, including moving violations, equipment violations, etc., with the exception of parking tickets and tickets related to involvement in a fatal or severe crash. In order to standardize the number of traffic tickets drivers received before their first fatal or severe crash (if any), for each driver the mean number of tickets per year was calculated until the end of the study or until the first fatal or severe crash, if there was one. The number of tickets per year per

88

R. Factor / Accident Analysis and Prevention 64 (2014) 86–91

Table 1 Descriptive statistics of the research variables. Variables Road traffic crashes (1 = yes) Traffic tickets per year Gender (1 = woman) Age Religion (1 = Jewish) Years of schooling Household social class White collar Blue collar Non-skilled worker Asset index Vehicle type Car Truck Taxi/bus Daily distance traveled in KM

N 409,051 409,051 409,051 409,051 409,051 408,227 320,662 107,020 159,818 53,824 329,601 401,720 337,057 53,545 11,118 399,684

Range

Mean

SD

0/1 0–6 0/1 16–100 0/1 0–24

0.01 0.29 0.37 37.87 0.88 12.52

0.08 0.48 0.48 14.21 0.33 3.19

0.33 0.50 0.17 0.34

0.47 0.50 0.37 0.93

0.84 0.13 0.03 40.07

0.38 0.34 0.16 14.50

0/1 0/1 0/1 (−2.78)–2.31 0/1 0/1 0/1 20.59–75.98

driver ranged from 0 to 33.71. To avoid statistical bias, drivers who received six or more tickets per year were collapsed into one category (fewer than 0.03% of the drivers received six tickets or more during the research period). Exposure has an important effect on road traffic accident involvement, with the rate of crash involvement increasing in line with distance traveled (Lourens et al., 1999). Following Factor et al. (2011) drivers’ average daily distance traveled in kilometers was estimated from the Israeli Travel Habits Survey. The survey participants were divided into 67 groups according to their gender, age, and place of residence. For each group an estimation of the daily distance traveled per driver in kilometers was calculated by dividing the group’s total distance traveled by the number of drivers in the group. These estimations were then added to the merged dataset. The following socioeconomic, demographic, and vehicle type variables that were previously found to have an association with traffic violations and road traffic crashes (Yagil, 1998; Braver, 2003; Romano et al., 2005; Factor et al., 2013a) were included in the analysis as control variables. Driver’s household social class was estimated by the Class Schema (Erikson and Goldthorpe, 1992), which integrated occupation and personal status at work to produce three hierarchical groups: white collar, blue collar, and non-skilled workers. Household asset index, which is presented in standard scores, was assessed through 15 census variables which estimate ownership of various assets and housing quality (Filmer and Pritchett, 1999, 2001). Driver’s education was measured as years of schooling. The analyses also included age, religion (Jewish/non-Jewish), and vehicle type (car, truck, bus/taxi). Table 1 presents the descriptive statistics of the research variables. 3.3. Method of analysis As mentioned above, the dependent variable, involvement in fatal and severe crashes, is a binary variable which represents whether the driver was or was not involved in such a crash during the research period. Multiple logistic regressions are well suited for testing this type of variable, as they express a dichotomous response variable as a function of several explanatory variables, and they fit these situations better than least-squares linear regressions (Fox, 2008). Thus, using multiple logistic regressions, involvement in fatal and severe crashes was regressed on the rate of traffic tickets per year till the first fatal or severe crash (if any) and the demographic, socioeconomic, car type, and distance traveled variables. In addition, several possible interactions between traffic tickets per year and demographic, socioeconomic, and car type variables were included in some of the models to test if the effects of these variables

Fig. 1. Cumulative percentages of traffic tickets per year and fatal and severe road traffic crashes, 2002–2008. Note: the number of traffic tickets per year in each category is presented in brackets.

on involvement in fatal/severe crashes vary across different levels of traffic tickets per year. The inclusion of traffic tickets received per year prior to the first fatal and severe crash as an exploratory variable while controlling for various confounders enables testing for temporal order and causality. One of the weaknesses of logistic models is the difficulty of interpreting the coefficients. Hence, in order to illustrate the probability of involvement in a subsequent fatal or severe crash for different rates of traffic tickets received, the relevant regression coefficients were transformed to probabilities while all other variables in the model were held constant to their mean (see, e.g., Pampel, 2000; Fox, 2008). Due to missing cases, the regression analysis was performed for 274,131 drivers. An analysis comparing the characteristics of these drivers to the missing cases (N = 134,920) found no meaningful differences between them. 4. Results Over the seven years from 2002 through 2008, 0.58% of the drivers were involved in a fatal or severe road traffic crash, and about 60% received at least one traffic ticket. Among drivers who were involved in a fatal or severe crash during those years, 55% received at least one traffic ticket before the crash, while 60% of those not involved in a crash received at least one traffic ticket. Among all drivers who received traffic tickets, 99% were not involved in a fatal or severe crash. Fig. 1 presents the cumulative percentage of traffic tickets per year and the cumulative percentage of fatal and severe traffic crashes during the research period (for ease of presentation the traffic tickets per year were collapsed into six groups). As the figure shows, up to 94% of drivers received one traffic ticket or less per year during the research period, and these drivers were involved in 82% of the fatal and severe crashes. In other words, the bulk of crashes involved that portion of drivers – the vast majority – who received a relatively small number of tickets. On the other hand, the 6% of drivers who received more than one ticket per year during the research period were involved in 18% of the fatal and severe crashes. Logistic regressions with four models are presented in Table 2. As the table shows, the data reveal a positive and significant association between receiving traffic tickets and subsequent involvement in a fatal or severe crash. Controlling for estimated daily distance traveled, an increase of one unit in the number of traffic tickets received per year increases the odds of involvement in a fatal or

R. Factor / Accident Analysis and Prevention 64 (2014) 86–91

89

Table 2 Logistic regression of fatal and severe crashes on traffic tickets per day, distance traveled, demographic, socioeconomic, and car type variables, 2002–2008. Variables

Traffic tickets per year Daily distance traveled Gender (1 = woman) Age Religion (1 = Jewish) Years of schooling Asset index Household social class White collar Blue collar Non-skilled worker Vehicle type Taxi/bus Truck Car Car (0 = truck) Gender × tickets Religion × tickets Car × tickets Constant

Model 1

Model 2

Model 3

Model 4

OR

95% CI

OR

95% CI

OR

95% CI

OR

95% CI

1.81*** 1.03***

(1.72, 1.91) (1.03, 1.04)

1.65*** 1.02*** 0.58*** 0.99* 0.83* 1.00 0.98

(1.56, 1.75) (1.01, 1.02) (0.49, 0.70) (0.99, 0.99) (0.72, 0.96) (0.98, 1.02) (0.92, 1.05)

1.52*** 1.02*** 0.53*** 0.99* 0.75*** 1.00 0.98

(1.39, 1.66) (1.01, 1.02) (0.44, 0.65) (0.99, 0.99) (0.63, 0.89) (0.98, 1.02) (0.92, 1.04)

1.55*** 1.02*** 0.60*** 1.00* 0.78*** 1.00 0.98

(1.43, 1.69) (1.01, 1.02) (0.50, 0.72) (0.99, 0.99) (0.67, 0.90) (0.98, 1.02) (0.92, 1.04)

1.00 1.11 1.26**

(0.97, 1.27) (1.06, 1.48)

1.00 1.11 1.26**

(0.97, 1.27) (1.07, 1.48)

1.00 1.12 1.17

(0.97, 1.28) (0.99, 1.40)

1.00 0.69*** 0.47***

(0.55, 0.85) (0.39, 0.58)

1.00 0.70*** 0.48***

(0.56, 0.86) (0.39, 0.58) 0.62***

(0.53, 0.72)

1.58** 1.13*

(1.18, 2.12) (1.01, 1.27) 1.18** 0.00***

(1.05, 1.32)

0.00***

0.01***

0.01***

Note: CI = confidence interval; OR = odds ratio. * p < .05. ** p < .01. *** p < .001.

severe crash during the research period by 81% (Model 1). This association remains even when controlling for various confounders, as can be seen in Model 2, where an increase of one unit per year in the rate of traffic tickets increases the odds of involvement in a fatal or severe crash by 65%. Model 2 also indicates that drivers who are male, younger, non-Jewish (a minority group in Israel), nonskilled workers, and who drive trucks, taxis or buses have higher odds of involvement in a fatal or severe crash, while controlling for the other variables in the model. Models 3 and 4 reveal significant interactions between traffic tickets per year and gender, religion, and car type, suggesting that the association between the rate of traffic tickets and fatal/severe crashes is conditioned by the driver’s gender and religion and by car type (other interaction terms were considered but were not significant, and so were not included in the models). The meaningful and positive effect of traffic tickets per day on fatal and severe crashes can be easily observed in Fig. 2, which illustrates the likelihood of involvement in a subsequent fatal or severe

Fig. 2. Effects display of the estimated probability of involvement in a fatal or severe crash by traffic tickets per year, 2002–2008. Note: the effects displays were modeled from the logistic regression (Model 2) when all other variables in the model were held constant to their means.

crash for different rates of traffic tickets when all other variables in Model 2 are held constant (Pampel, 2000). The figure shows that the probability of involvement in a fatal or severe crash is 65% higher for drivers who received one ticket per year (0.0063) compared to drivers who received no tickets (0.0038) during the seven years of the study. The probability of fatal/severe crash involvement is 1051% higher for drivers who received six tickets per year (0.0725) compared to those who received one ticket per year (0.0063) during the study period. It is worth noting that similar results were obtained when light crashes were also analyzed, as well as when two different datasets were considered – a dataset which included all violations, even those occurring after the crash; and a dataset that merged only traffic tickets and fatal/severe crashes occurring between 2002 and 2008, without the census data.

5. Discussion Studies indicate that risky driving and violations of traffic rules are associated with road traffic crashes (Parker et al., 1995; Gebers and Peck, 2003). However, studies exploring these associations often use less than optimal data, which may bias and weaken their results. The current study aimed to overcome these limitations by examining the effect of traffic tickets on subsequent fatal and severe crashes, using a unique longitudinal dataset at the individual level that merges Israeli census data with traffic tickets issued by the police and official data on involvement in road traffic crashes. The dataset covers 20% of Israeli drivers, identifying those who received tickets and/or were involved in a crash over seven years, while providing demographic, socioeconomic, distance traveled, and car type information for each driver. The results show that a large majority of the fatal and severe crashes (82%) involved the broad population of drivers (94% of all drivers) who received a relatively small number of traffic tickets (up to one per year over seven years). The other 18% of fatal/severe crashes involved the small group of drivers (6% of all drivers) who received more than one ticket per year. The multivariate analyses suggest a strong positive association between the number of tickets drivers received and their subsequent involvement in a fatal or

90

R. Factor / Accident Analysis and Prevention 64 (2014) 86–91

severe crash. The estimated probability of involvement in a subsequent fatal or severe crash is 65% higher for drivers who received one ticket per year compared to those who received no tickets, and more than eleven times higher for drivers with six tickets per year compared to drivers who received one ticket. This effect is in line with previous studies (Parker et al., 1995; ETSC, 1999; Blows et al., 2005; Ivers et al., 2009), but is stronger than those previously reported (Gebers and Peck, 2003; Shinar, 2007). One explanation frequently offered for the effect of traffic violations on road traffic crashes is that “people drive as they live” (Tillmann and Hobbs, 1949; Factor et al., 2007). In other words, it is reasonable to assume that drivers who engage in risky driving behavior (and who therefore receive many traffic tickets) have similar characteristics as those who are involved in crashes. Under this argument, while only a thin line divides cases where risky driving behavior merely causes annoyance from those where risky behavior results in a crash, when that line is crossed it is usually a matter of chance, depending upon the simultaneous occurrence of a number of additional “random” factors. However, although this explanation is attractive, if crashes were indeed dependent upon such chance occurrences, we would not see the strong correlation found here between traffic violations (as measured via traffic tickets received) and crashes – a correlation found despite controlling for many known confounders that are associated with both crashes and violations. A more likely explanation relates to the fact that driving inherently makes high demands on the driver’s cognitive resources, requiring as it does steady attention to numerous foci (the road ahead; other vehicles; non-vehicle road users, such as pedestrians or cyclists; and the condition of the vehicle being driven). Risky driving behavior (e.g., speeding, weaving in and out of lanes, or failing to yield right of way) overloads these cognitive resources, making the driver more prone to errors which may precipitate a crash (Stradling et al., 1998). It remains for future studies to uncover the precise mechanism by which risky behavior and traffic violations influence subsequent involvement in road traffic crashes. An important implication of findings such as the current ones is that reducing traffic violations and risky driving may contribute to a reduction in crashes. Improved traffic enforcement seems to have such an effect by promoting adherence to traffic laws (though if such enforcement programs are then curtailed, their effect may be relatively short-lasting) (Stanojevic´ et al., 2013). An important question in situations of scarce resources is whether costly enforcement initiatives should target the entire population of drivers, or should focus on the small population of high-risk recidivist drivers. In keeping with the prevention paradox (Rose, 1981; Davison et al., 1991), the current study shows that while drivers who receive more traffic tickets are more likely to be involved in a crash than those with fewer tickets, the bulk of crashes actually involve drivers with a small number of tickets, who form the majority of drivers. The current data suggest that hypothetically, and leaving all other factors unchanged, removing from the road the 6% of drivers who receive the most tickets would reduce the total number of fatal and severe crashes by about 18% over seven years. Meanwhile, a mass random enforcement program that cut tickets issued to each driver by two tickets over seven years (i.e., 0.29 fewer tickets per year) would reduce the number of fatal and severe crashes by 27% over that time. Thus, as suggested by Shinar (2007) in relation to DWI offenders, a mass random enforcement program would have a greater impact on road safety than targeting high-risk offenders. The current findings should be interpreted in light of the study’s limitations. First, the analyses are based on data on traffic tickets issued, and not on actual traffic violations. As the degree to which the number of tickets issued mirrors actual traffic violations is unknown (Elvik et al., 2012), we must be cautious before drawing conclusions regarding the effect of total traffic violations on crashes.

Second, the socioeconomic data were drawn from the 1995 census, seven years before the start of the study period, raising the possibility that some socioeconomic information was no longer accurate when a driver received a ticket or was involved in a crash. However, several analyses indicate that, on average, changes over time in socioeconomic variables – most of which were in any case calculated at the household level – are not meaningful. Third, the current analysis did not include drivers under 25 (the youngest drivers were 18 years old at the time of the census, and so were 25 at the start of the study period in 2002). However, since younger drivers are known to have a higher risk of involvement in crashes and to commit more traffic violations (Ivers et al., 2009), the effects found here might be even stronger had younger drivers been included. Finally, the analyses included all fatal and severe crashes in which a driver was involved without taking into account whether that driver was formally named as culpable in the crash. There were two reasons for this decision. First, the police often assign culpability in a crash in order to prosecute the offender and not in order to determine the ultimate cause of the crash. Hence, police reports are not necessarily reliable for this purpose (Shinar, 2007). Second, it is reasonable to assume that in many cases where more than one driver is involved in a crash, the behavior of both drivers contributes to the crash to some degree – an assumption supported by the current findings of an association between tickets and all types of fatal and severe crashes. In conclusion, using a unique longitudinal individual-level dataset, the current study suggests that the number of traffic tickets drivers received strongly affects their subsequent involvement in fatal and severe crashes. Moreover, most crashes involve the population of relatively law-abiding drivers, who receive relatively few tickets. It thus appears that mass random enforcement programs may be more effective in reducing fatal and severe crashes than targeting only high-risk drivers. Further studies in different countries over longer periods are required to support the current findings. In addition, to better understand the effect of traffic violations per se on crashes, researchers should seek ways to measure traffic violations directly while continuing to control for various confounders. Acknowledgment The study was supported by the Israel National Road Safety Authority. References ˜ M., 2010. The impact of traffic violations on the estiAyuso, M., Guillén, M., Alcaniz, mated cost of traffic accidents with victims. Accident Analysis and Prevention 42 (2), 709–717. Bjerre, B., Thorsson, U., 2008. Is an alcohol ignition interlock programme a useful tool for changing the alcohol and driving habits of drink-drivers? Accident Analysis and Prevention 40 (1), 267–273. Blows, S., Ameratunga, S., Ivers, R.Q., Lo, S.K., Norton, R., 2005. Risky driving habits and motor vehicle driver injury. Accident Analysis and Prevention 37 (4), 619–624. Braver, E.R., 2003. Race, Hispanic origin, and socioeconomic status in relation to motor vehicles occupant death rates and risk factors among adults. Accident Analysis and Prevention 35 (3), 295–309. Chen, W., Cooper, P., Pinili, M., 1995. Driver accident risk in relation to the penalty point system in British Columbia. Journal of Safety Research 26 (1), 9–18. Cooper, P.J., 1997. The relationship between speeding behaviour (as measured by violation convictions) and crash involvement. Journal of Safety Research 28 (2), 83–95. Davison, C., Smith, G.D., Frankel, S., 1991. Lay epidemiology and the prevention paradox – the implications of coronary candidacy for health-education. Sociology of Health & Illness 13 (1), 1–19. de Winter, J.C.F., Dodou, D., 2010. The driver behaviour questionnaire as a predictor of accidents: a meta-analysis. Journal of Safety Research 41 (6), 463–470. Elvik, R., Christensen, P., 2007. The deterrent effect of increasing fixed penalties for traffic offences: the Norwegian experience. Journal of Safety Research 38 (6), 689–695.

R. Factor / Accident Analysis and Prevention 64 (2014) 86–91 Elvik, R., Sogge, C.V., Lager, L., Amundsen, F.H., Pasnin, L.T., Karlsen, R., Fosli, K., 2012. Assessing the efficiency of priorities for traffic law enforcement in Norway. Accident Analysis and Prevention 47, 146–152. Elvik, R., Vaa, T., 2004. The Handbook of Road Safety Measures. Elsevier, Amsterdam, New York. Erikson, R., Goldthorpe, J.H., 1992. The Constant Flux: A Study of Class Mobility in Industrial Societies. Clarendon Press, Oxford University Press, New York, Oxford, England. ETSC, 1999. Police Enforcement Strategies to Reduce Traffic Casualties in Europe. European Transport Safety Council, Brussels. Evans, L., 2004. Traffic Safety. Science Serving Society, Bloomfield Hills, MI. Factor, R., Mahalel, D., Rafaeli, A., Williams, D.R., 2013a. A social resistance perspective for delinquent behavior among non-dominant minority groups. British Journal of Criminology 53 (5), 784–804. Factor, R., Mahalel, D., Yair, G., 2007. The social accident: a theoretical model and a research agenda for studying the influence of social and cultural characteristics on motor vehicle accidents. Accident Analysis and Prevention 39 (5), 914–921. Factor, R., Prashker, J.N., Mahalel, D., 2012. The flashing green light paradox. Transportation Research Part F: Traffic Psychology and Behavior 15 (3), 279–288. Factor, R., Williams, D.R., Kawachi, I., 2013b. Social resistance framework for understanding high-risk behavior among nondominant minorities: preliminary evidence. American Journal of Public Health 103 (12), 2245–2251. Factor, R., Yair, G., Mahalel, D., 2011. Acciphilia on the road: an analysis of severe collisions. Journal of Safety Research 42 (5), 367–374. Ferrante, A.M., Rosman, D.L., Marom, Y., 2001. Novice drink drivers, recidivism and crash involvement. Accident Analysis and Prevention 33 (2), 221–227. Filmer, D., Pritchett, L.H., 1999. The effect of household wealth on educational attainment: evidence from 35 countries. Population and Development Review 25 (1), 85–120. Filmer, D., Pritchett, L.H., 2001. Estimating wealth effects without expenditure data – or tears: an application to educational enrollments in states of India. Demography 38 (1), 115–132. Fox, J., 2008. Applied Regression Analysis and Generalized Linear Models, 2nd ed. SAGE, Los Angeles. Gebers, M.A., Peck, R.C., 2003. Using traffic conviction correlates to identify high accident-risk drivers. Accident Analysis & Prevention 35 (6), 903–912. Goldenbeld, C., Reurings, M., Van Norden, Y., Stipdonk, H.L., 2011. The Relation Between Traffic Offences and Road Crashes: Preliminary Study Making use of Central Fine Collection Agency (CJIB) Data. Institute for Road Safety Research (SWOV). Hakkert, A.S., Gitelman, V., Cohen, A., Doveh, E., Umansky, T., 2001. The evaluation of effects on driver behavior and accidents of concentrated general enforcement on interurban roads in Israel. Accident Analysis & Prevention 33 (1), 43–63. Ivers, R., Senserrick, T., Boufous, S., Stevenson, M., Chen, H.-Y., Woodward, M., Norton, R., 2009. Novice drivers’ risky driving behavior, risk perception, and crash risk: findings from the drive study. American Journal of Public Health 99 (9), 1638–1644. Jonah, B.A., 1997. Sensation seeking and risky driving: a review and synthesis of the literature. Accident Analysis and Prevention 29 (5), 651–665. Lourens, P.F., Vissers, J.a.M.M., Jessurun, M., 1999. Annual mileage, driving violations, and accident involvement in relation to drivers’ sex, age, and level of education. Accident Analysis and Prevention 31 (5), 593–597. Lozano, R., Naghavi, M., Foreman, K., Lim, S., Shibuya, K., Aboyans, V., Abraham, J., Adair, T., Aggarwal, R., Ahn, S.Y., Almazroa, M.A., Alvarado, M., Anderson, H.R., Anderson, L.M., Andrews, K.G., Atkinson, C., Baddour, L.M., Barker-Collo, S., Bartels, D.H., Bell, M.L., Benjamin, E.J., Bennett, D., Bhalla, K., Bikbov, B., Abdulhak, A.B., Birbeck, G., Blyth, F., Bolliger, I., Boufous, S., Bucello, C., Burch, M., Burney, P., Carapetis, J., Chen, H., Chou, D., Chugh, S.S., Coffeng, L.E., Colan, S.D., Colquhoun, S., Colson, K.E., Condon, J., Connor, M.D., Cooper, L.T., Corriere, M., Cortinovis, M., De Vaccaro, K.C., Couser, W., Cowie, B.C., Criqui, M.H., Cross, M., Dabhadkar, K.C., Dahodwala, N., De Leo, D., Degenhardt, L., Delossantos, A., Denenberg, J., Des Jarlais, D.C., Dharmaratne, S.D., Dorsey, E.R., Driscoll, T., Duber, H., Ebel, B.,

91

Erwin, P.J., Espindola, P., Ezzati, M., Feigin, V., Flaxman, A.D., Forouzanfar, M.H., Fowkes, F.G.R., Franklin, R., Fransen, M., Freeman, M.K., Gabriel, S.E., Gakidou, E., Gaspari, F., Gillum, R.F., Gonzalez-Medina, D., Halasa, Y.A., Haring, D., Harrison, J.E., Havmoeller, R., Hay, R.J., Hoen, B., Hotez, P.J., Hoy, D., Jacobsen, K.H., James, S.L., Jasrasaria, R., Jayaraman, S., Johns, N., Karthikeyan, G., Kassebaum, N., Keren, A., Khoo, J.-P., Knowlton, L.M., Kobusingye, O., Koranteng, A., Krishnamurthi, R., Lipnick, M., Lipshultz, S.E., et al., 2012. Global and regional mortality from 235 causes of death for 20 age groups in 1990 and 2010: a systematic analysis for the global burden of disease study 2010. Lancet 380 (9859), 2095–2128. Macdonald, S., Mann, R.E., Chipman, M., Anglin-Bodrug, K., 2004. Collisions and traffic violations of alcohol, cannabis and cocaine abuse clients before and after treatment. Accident Analysis and Prevention 36 (5), 795–800. Pai, C.-W., 2011. Motorcycle right-of-way accidents—a literature review. Accident Analysis and Prevention 43 (3), 971–982. Pampel, F.C., 2000. Logistic Regression: A Primer. Sage Publications, Thousand Oaks, CA. Parker, D., Reason, J.T., Manstead, A.S.R., Stradling, S.G., 1995. Driving errors, driving violations and accident involvement. Ergonomics 38 (5), 1036–1048. Rajalin, S., 1994. The connection between risky driving and involvement in fatal accidents. Accident Analysis and Prevention 26 (5), 555–562. Reason, J.T., Manstead, A.S.R., Stradling, S., Baxter, J., Campbell, K., 1990. Errors and violations on the roads. Ergonomics 33, 1315–1332. Retting, R.A., Ferguson, S.A., Hakkert, A.S., 2003. Effects of red light cameras on violations and crashes: a review of the international literature. Traffic Injury Prevention 4, 17–23. Robertson, L.S., 1996. Reducing death on the road: the effects of minimum safety standards, publicized crash tests, seat belts, and alcohol. American Journal of Public Health 86 (1), 31–34. Romano, E., Tippetts, S., Voas, R., 2005. Fatal red light crashes: the role of race and ethnicity. Accident Analysis and Prevention 37 (3), 453–460. Rose, G., 1981. Strategy of prevention: lessons from cardiovascular disease. British Medical Journal 282 (6279), 1847–1851. Rothengatter, J.A., 1997. Errors and violations as factors in accident causation. In: Rothengatter, T., Carbonell Vaya, E. (Eds.), Traffic and Transport Psychology: Theory and Application. , 1st ed. Pergamon, Amsterdam, New York, pp. 59–64. Shinar, D., 2007. Traffic Safety and Human Behavior. Elsevier, Amsterdam, Boston. Sivak, M., Luoma, J., Flannagan, M.J., Bingham, C.R., Eby, D.W., Shope, J.I., 2007. Traffic safety in the US: re-examining major opportunities. Journal of Safety Research 38 (3), 337–355. ´ P., Jovanovic, ´ D., Lajunen, T., 2013. Influence of traffic enforcement on the Stanojevic, attitudes and behavior of drivers. Accident Analysis and Prevention 52, 29–38. Stradling, S., Parker, D., Lajunen, T., Meadows, M.L., Xie, Q.C., 1998. Drivers’ violations, errors, lapses and crash involvement: international comparisons. In: Proceedings of the Road Safety in Europe Conference, Bergisch Gladbach, Germany, pp. 31–45. Sullman, M.J.M., Meadows, M.L., Pajo, K.B., 2002. Aberrant driving behaviours amongst New Zealand truck drivers. Transportation Research Part F: Traffic Psychology and Behavior 5 (3), 217–232. Tillmann, W.A., Hobbs, G.E., 1949. The accident-prone automobile driver. American Journal of Psychiatry 106 (5), 321–331. Vingilis, E., Wilk, P., 2008. The effects of health status, distress, alcohol and medicinal drug use on subsequent motor vehicle injuries. Accident Analysis and Prevention 40 (6), 1901–1907. Wilmot, C.G., Khanal, M., 1999. Effect of speed limits on speed and safety: a review. Transport Reviews 19 (4), 315–329. World Health Organization, 2013. Global Status Report on Road Safety 2013: Supporting a Decade of Action. World Health Organization, Geneva, Switzerland. Yagil, D., 1998. Gender and age-related differences in attitudes towards traffic laws and traffic violations. Transportation Research Part F: Traffic Psychology and Behaviour 1 (2), 123–135. Yannis, G., Papadimitriou, E., Antoniou, C., 2007. Multilevel modelling for the regional effect of enforcement on road accidents. Accident Analysis and Prevention 39 (4), 818–825.