Graduated driver licensing and differential deterrence: The effect of license type on intentions to violate road rules

Accident Analysis and Prevention 110 (2018) 62–70

Contents lists available at ScienceDirect

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

Graduated driver licensing and differential deterrence: The effect of license type on intentions to violate road rules

MARK

⁎

Brigitte Poiriera, Etienne Blaisb, , Camille Fauberta a

School of Criminology, Université de Montréal, 3150, rue Jean-Brillant, Montreal, Quebec, H3T 1N8, Canada School of Criminology and International Centre for Comparative Criminology, Université de Montréal 3150, rue Jean-Brillant Room C-4121, Montreal, Quebec, H3T 1N8, Canada b

A R T I C L E I N F O

A B S T R A C T

Keywords: Graduated driver license Deterrence Social control Delinquent peers Traffic violations

In keeping with the differential deterrence theory, this article assesses the moderating effect of license type on the relationship between social control and intention to violate road rules. More precisely, the article has two objectives: (1) to assess the effect of license type on intentions to infringe road rules; and (2) to pinpoint mechanisms of social control affecting intentions to violate road rules based on one’s type of driver license (a restricted license or a full license). This effect is examined among a sample of 392 young drivers in the province of Quebec, Canada. Drivers taking part in the Graduated Driver Licensing (GDL) program have limited demerit points and there is zero tolerance for drinking-and-driving. Propensity score matching techniques were used to assess the effect of the license type on intentions to violate road rules and on various mechanisms of social control. Regression analyses were then conducted to estimate the moderating effect of license type. Average treatment effects from propensity score matching analyses indicate that respondents with a restricted license have lower levels of intention to infringe road rules. While moral commitment and, to a lesser extent, the perceived risk of arrest are both negatively associated with intentions to violate road rules, the license type moderates the relationship between delinquent peers and intentions to violate road rules. The effect of delinquent peers is reduced among respondents with a restricted driver license. Finally, a diminished capability to resist peer pressure could explain the increased crash risk in months following full licensing.

1. Introduction In high-income countries, young drivers are generally overrepresented in traffic fatalities (Elvik, 2010). The Province of Quebec, Canada, is no exception. In 2015, 20% of drivers involved in road crashes and 14% of fatally-injured drivers were under 25 years old, while only 9% of all license holders fell into this age category (Société de l’assurance and automobile, 2016). Their inexperience, immaturity, and reckless behaviors all increase their risk of fatal crash (Hedlund et al., 2003). Many jurisdictions have consequently adopted Graduated Driver Licensing (GDL) programs to reduce crashes among young and novice drivers (Shope, 2007). GDL introduces a learning stage and maintains a low-risk, supervised learning environment. While deterrent mechanisms are central components of GDL, few studies have investigated how GDL influences the effect of mechanisms of social control on driving violations (Simpson, 2003; Williams, 2007). Given the fact that crash risk significantly increases once drivers obtain their full license (Williams, 2007; Curry et al., 2015), examining this influence is of paramount importance. ⁎

Corresponding author. E-mail address: [email protected] (B. Poirier).

http://dx.doi.org/10.1016/j.aap.2017.10.001 Received 25 July 2017; Received in revised form 1 October 2017; Accepted 2 October 2017 0001-4575/ © 2017 Elsevier Ltd. All rights reserved.

In keeping with the recent scholarship on differential deterrence (Piquero et al., 2011), this study compares the effect of mechanisms of formal and informal social control on non-compliance of road rules among young Quebec drivers with a restricted driver license (GDL) and those with a full license. This article has two objectives: (1) to estimate the effect of license type on intentions to violate road rules; and (2) to identify formal and informal mechanisms of social control affecting intentions to violate road rules based on one’s type of driver license (a restricted license or a full license). 2. Literature review The literature review is divided in three sections. The first section describes GDL components and objectives. The second presents findings of studies that have investigated the effect of formal and informal mechanisms of social control on intentions to violate road rules among young drivers. Finally, the third section introduces the differential deterrence theory and demonstrates its relevance to the investigation of factors associated with the intentions to violate road rules among young

Accident Analysis and Prevention 110 (2018) 62–70

B. Poirier et al.

informal deterrence mechanisms were collected through an online survey. Their results indicate that P2 holders have higher rates of noncompliance than P1 holders with GDL restrictions. The latter have a higher perceived risk of being punished by their parents if they disobey restrictions. Accordingly, informal mechanisms of social control increase compliance (Bates et al., 2015). Another study shows similar findings (Allen et al., 2015). Among a sample of 151 young Australian drivers, “informal deterrence” (such as feelings of guilt) elevates compliance with GDL restrictions.

drivers. This last section also highlights the potential moderating effect of the license type on the relationship between mechanisms of social control and intentions to infringe road rules. 2.1. Graduated driver licensing: components and objectives GDL aims to provide a safe learning environment to novice drivers by extending their learning phase and by incorporating driving restrictions (Vanlaar et al., 2009). GDL is usually divided into three stages. Drivers in the first stage (learner license) can drive only when accompanied by a full license holder. Drivers must successfully pass a driving test to access the next stage. In the second stage (provisional license), drivers can drive unsupervised but under some restrictions (e.g. limiting the number of young passengers and prohibiting nighttime driving) (Fell et al., 2011; Lin and Fearn, 2003; Chen et al., 2001; Cooper et al., 2005). There is generally zero tolerance for drinking-anddriving in the first two stages and drivers have a limited number of demerit points. If they respect all restrictions imposed in the second stage, after a predetermined period of time (e.g. two years), drivers earn a full permit in the third stage. Previous research shows that GDL reduces the crash risk among young drivers (Russell et al., 2011; Shope, 2007; Williams and Shults, 2010). Shope (2007) reviewed 27 studies published since 2002 and concluded that GDL reduces crash risk by 20% to 40%. Effects of GDL are enhanced when it includes: (1) nighttime driving restrictions, (2) limitations on teenage passengers, (3) a six-month learning period, (4) a minimum age for full licensing, and (5) mandatory driving lessons (Dee et al., 2005; Masten et al., 2011; Vanlaar et al., 2009; Williams, 2007; Williams and Shults, 2010). Despite the benefits of GDL, young drivers still display increased crash risk (Conner and Smith, 2017; Gregersen et al., 2003). Curry et al., (2015) studied 410 230 drivers aged 17–20 years old, and report that « (…) independent of age and experience, teen drivers’ crash risk increased substantially at the point of transition to a full license, while drivers of a similar age who remained in the intermediate phase continued to experience a decline in crash rates » (p. 243). Although novice drivers acquire additional skills in the learning phase, the long-term benefits of GDL remain unclear (Conner and Smith, 2017; Curry et al., 2015; Gregersen et al., 2003). Others argue that lower crash risk is not attributable to GDL and the development of new skills but rather to limited risk exposure (Karaca-Mandic and Ridgeway, 2010). The factors responsible for the lower crash risk during GDL need to be further explored (Simpson, 2003; Williams, 2007).

2.3. The present study: GDL and differential deterrence Previous studies show that informal social control increases compliance with GDL restrictions (Allen et al., 2015; Bates et al., 2015). Prior studies, however, use samples of restricted license holders only. Consequently, their findings do not explain the increase in the crash risk right after full licensing among young drivers. Two central questions about the effectiveness of restricted licensing are not addressed by previous studies: (1) what is the relationship between license type and non-compliance; and (2) how does license type moderate the effect of formal and informal mechanisms of social control on non-compliance? Differential deterrence theory can address both questions. Classical deterrence theory posits that in order to prevent an offense, a sanction must be sufficiently certain, severe and swiftly applied. Studies in differential deterrence identify contextual and individual characteristics that moderate the effect of sanctions (Andenaes, 1974; Geerken and Gove, 1975; Piquero et al., 2011). Social capital, moral beliefs, selfcontrol, emotional arousal and drug/alcohol consumption have been shown to affect perceptions and reactions to the threat of sanction (Piquero et al., 2011). A neglected aspect of conditions likely to affect compliance is a person’s status in a sanction system. Indeed, the type and severity of sanction vary according to a person’s status in a sanction system. A radical example is the “three strikes” law applied in several American states. Felons who have already received “two strikes” (i.e. have already been convicted twice for a felony) are exposed to longer imprisonment penalties than those with a clean record (Zimring et al., 2001). In traffic safety, demerit point systems are implemented in several jurisdictions. A demerit point system aims at deterring drivers from committing traffic infringements (Castillo-Manzano and Castro-Nuño, 2012). The “system” threatens to suspend the license of drivers exceeding a predetermined level of demerit points (Basili and Nicita, 2005). According to the number of accumulated points, the driver’s status changes. A driver approaching the limit of demerit points may see sanctions as more threatening than a driver without any demerit points. In the former situation, a license suspension is added to the fine. In comparison to fully licensed drivers, drivers enrolled in GDL are exposed to a zero-tolerance policy for drinking-and-driving and to harsher demerit point systems.3 Hence, drivers in GDL expose themselves to increased consequences when they violate road rules (Simpson, 2003). Holding a restricted license is therefore likely to activate or enhance mechanisms of formal and informal social control in at least three ways. First, in keeping with the deterrence theory, GDL restrictions can increase levels of perceived certainty and severity of sanction (Homel, 1988; Zimring et al., 1973). Several studies emphasize that the sanction threat not only deters drivers but it also changes the social norm (i.e. moral commitment) and educates drivers about the crash risk associated with traffic infringements (Blais and Ouimet, 2005; Kennedy, 2009).4

2.2. GDL restrictions and the prevention of traffic violations In order to pinpoint factors associated with compliance among young drivers with a restricted license, some studies have investigated the role of formal and informal social control (Allen et al., 2015; Bates et al., 2015).1 For Simpson (2003), « (t)he probationary scheme is anchored in the concept of deterrence. It is assumed that safe driving habits, at least when the initial risk of collision is much higher, will be encouraged by the threat of punishment and its application » (p. 26). Deterrent mechanisms are, however, seldom addressed in studies on GDL. One exception is Bates et al. (2015) who used a sample of 236 young drivers holding P1 and P2 licenses in Queensland, Australia (P1 and P2 correspond to two types of restricted licenses).2 Data on formal and 1 From now on, we will refer to drivers participating in GDL as restricted license holders. 2 In Queensland (Australia), new drivers under 25 years old first hold a Provisional 1 (P1) license after completing the “apprentice” stage. To obtain a P1 license, they must be at least 17 years old, hold an apprentice license for at least 12 months, and complete 100 h of supervised driving. P1 license plates are identified with a red P. After holding the P1 license for a year and successfully passing a driving exam, drivers get a Provisional 2 license. A full license can be obtained after holding a Provisional license for three years.

3 In Quebec, drivers under 25 years old or participating in GDL have a limited number of demerit points. GDL drivers get their license suspended once they accumulate four demerit points in comparison to 15 for drivers over 25 with a full license. 4 Morally committed individuals have internalized the social norms − that is, that violating road rules is likely to cause a prejudice to other road users − and such individuals will not infringe road rules because their self-concept will not allow them to do so, regardless of possible sanctions (Andenaes, 1974; Grasmick and Green, 1981).

63

Accident Analysis and Prevention 110 (2018) 62–70

B. Poirier et al.

Table 1 Descriptive Statistics. n

x

Md

SD

Min

Max

Cronbach alpha

License type (1 = restricted license)

390

0.25

0

0.43

0

1.00

–

Age Gender (1 = female) Previously involved in a crash (1 = yes) Self-control Exposure to police enforcement activities Delinquent peers Driving frequency (1 = driving daily) Car ownership (1 = yes) Driving experience (in months) Demerit points (1 = yes)

392 392 382 380 386 381 391 391 387 392

20.98 0.54 0.25 2.60 0.45 2.11 0.56 0.66 40.88 0.26

21.00 1.00 0 2.63 0.40 2.14 1.00 1.00 37.00 0

1.78 0.50 0.44 0.64 0.26 0.62 0.50 0.48 25.48 0.44

18.00 0 0 1.0 0 1.00 0 0 0 0

25.00 1.00 1.00 5.54 1.00 5.00 1.00 1.00 142.00 1.00

– – – 0.89 0.54 0.83 – – – –

Perceived risk of arrest Perceived severity of sanction Threat of social disapproval Perceived seriousness of traffic violations Guilt feelings

386 382 389 382 372

4.16 2.02 2.42 3.57 2.60

4.57 2.00 2.43 3.57 2.67

1.55 0.32 0.67 0.60 0.68

1.00 1.00 1.00 1.00 1.00

6.00 3.00 4.00 5.00 4.00

0.94 0.64 0.86 0.78 0.91

Intentions to violate road rules

380

2.12

2.14

0.70

1.00

5.00

0.81

3.2. Measures

Second, shame and guilt can act as informal deterrence mechanisms (Bates et al., 2015; Grasmick and Bursik, 1990; Grasmick et al., 1993). Friends and relatives can exert social control by reminding drivers of GDL restrictions, which can generate feelings of guilt and shame when a novice driver infringes the rules (Allen et al., 2015). Parents can also restrict vehicle access (Bates et al., 2015). Finally, several studies show that teenage passengers enhance the crash risk among young drivers (Chen et al., 2000; Doherty et al., 1998; Simons-Morton et al., 2005). GDL can reduce this risk by increasing resistance to peer pressure. In a study on random breath testing in New South Wales (Australia), Homel (1988) indicates that “(…) legal actions reduce peer pressure to drive after drinking by providing an exculpatory or legitimate excuse for actions taken to avoid the offense” (p. 26). Hence, this study seeks to improve scholarship on the relationship between the type of license (restricted vs. full license) and intents to violate road rules. More precisely, this article aims to: (1) assess the effect of the license type on perceptions about informal and formal social control and intentions to violate road rules and (2) investigate the relationship between mechanisms of social control and intentions to violate road rules among a sample of young drivers with a full or a restricted license.

3.2.1. License type The main independent variable of this study is license type. Because of their low frequency, learner and probationary license holders were combined in one category. The variable is dichotomous (0 = full license; 1 = restricted license). When the questionnaire was filled out, learner license holders had to be accompanied by a driving monitor or by someone holding a full license for at least two years; probationary license holders could drive unsupervised. Both types of restricted license come with four demerit points and zero-tolerance for drinkingand-driving. About 25% of respondents have a restricted license (see Table 1).

3.2.2. Perceptions about formal and informal social control Five composite scales were used to measure perceptions about social control. This refers to mechanisms likely to inhibit intentions to violate road rules. Some variables correspond to mechanisms of formal social control (such as perceived risk of arrest and perceived severity of sanctions), while others correspond to informal sources of social control (such as threat of disapproval by relatives and feeling of guilt) (Bates et al., 2015). Perceived seriousness of traffic violations refers to the internalization of the norm or moral commitment (see Grasmick and Green (1981) for a similar scale). In all cases, composite scales were created by averaging scores of their respective items.

3. Methods 3.1. Data Data come from a questionnaire addressing several traffic safety issues such as: (1) traffic violations and their consequences, (2) risk perceptions (ex: perceived risk of arrest or accident), (3) vicarious and personal experiences with sanctions; (4) perceptions about the seriousness of traffic violations; and (5) license type. Participants are all undergraduate university students and were recruited either by a polling firm or via in-class advertisement. They respectively filled out an electronic or paper-pencil version of the questionnaire. In order to have a homogenous sample, only those aged 18–25 years old (n = 392) were included in this study. On average, participants were 20.98 years old (SD = 1.78) and 54% were female. About 25% were previously involved in a crash as a driver or a passenger. Most of the participants were driving daily (56%) and owning a car (66%). At last, participants’ driving experience averaged 40.88 months and 26% had demerit points on their driving record (Table 1).

3.2.3. Perceived risk of arrest This variable measures the perceived risk of being pulled over and fined if the respondent continues to adopt the same behaviors (such as running a red light or violating speed limits). The perceived certainty of arrest was created by averaging the scores of seven Likert-type items with values ranging between 1 (probability less than 0.0001%) and 6 (probability of 100%).

3.2.4. Perceived severity of sanctions This scale measures the perceived severity of sanctions (such as fines and demerit points). It was created by averaging the scores of six Likert-type items ranging from 1 (not severe at all) to 3 (very severe). Respondents were asked to assess the severity of various sanctions (such as three demerit points and 100–200 dollars fine for failing to stop at a red light).

64

Accident Analysis and Prevention 110 (2018) 62–70

B. Poirier et al.

Seventh, age is also included as a control variable to disentangle the effect of age from driving experience. Respondents were on average 21.98 years old. Age and driving experience are, however, strongly correlated (r = 0.67; p ≤ 0.001). [See Table A1 in Appendix A for the correlation matrix] Eighth, owning a car usually increases driving frequency and the opportunities to commit driving violations. Respondents were asked if they own a car (1 = yes; 0 = no). Ninth, demerit points are used to control for previous experiences with punishment. The initial distribution varied between 1 and 15, but it was highly skewed since 74.2% of all respondents did not have any demerit points. The variable was therefore dichotomized (1 = demerit points; 0 = no demerit points). Finally, a variable measures driving frequency. It is an important control, as some authors claim that the lower exposure to risk explains the low crash rates among restricted license holders (Karaca-Mandic and Ridgeway, 2010). The original variable contains six values but it was recoded since 55.8% of participants drive daily (1 = drive daily; 0 = less than one time per month up to 2–4 times per week).

3.2.5. Threat of social disapproval To measure the threat of social disapproval, participants were asked if their relatives would change their opinion of them if they knew the participants had committed a traffic violation. This scale is composed of seven Likert-type items ranging from 1 (their opinion about me would not change) to 4 (they would have a bad opinion about me). 3.2.6. Perceived seriousness of traffic violations This scale is inspired by a question on road safety issues found in the Road Safety Monitor (Brown et al., 2016). Averaging the scores of seven Likert-type items created the perceived seriousness of road violations scale. Respondents were asked about the seriousness of seven traffic violations, with answers varying between 1 (not a problem at all) to 5 (extremely serious problem). 3.2.7. Feelings of guilt Feelings of guilt are conceptualized as a self-imposed punishment (Grasmick and Scott, 1982). Respondents were asked how guilty they feel when they commit traffic violations. This scale is composed of seven Likert-type items ranging from 1 (not guilty at all) to 4 (totally guilty). As for the previous scales, an average was computed.

3.3. Statistical analysis

3.2.8. Dependent variable: intentions to violate road rules This variable estimates self-reported intentions to violate road rules. Respondents were asked about the likelihood that they would commit various violations in the next three months. This scale is composed of seven Likert-type items ranging from 1 (certainly not) to 5 (certainly). As for the other scales, an average score was computed. The mean value ( x = 2.12; S.D. = 0.70) indicates that there is a low reporting of intentions to violate traffic rules.

Four separate sets of statistical analyses were conducted to evaluate the effect of the license type and social control on intentions to violate road rules. First, propensity score matching was used to evaluate the effect of license type on social control mechanisms and intentions to infringe road rules. Propensity score matching seeks to reproduce experimental conditions and reduce possible biases (Apel and Sweeten, 2010; Rosenbaum and Rubin, 1983). In order to simulate a randomized distribution of participants, individuals are matched with those who share similar propensity scores (individual characteristics such as driving experience, sex and age) (Becker and Ichino, 2002; Shadish et al., 2002). Propensity score matching was conducted with STATA 12. Second, correlation coefficients were computed to assess bivariate associations between all variables. Third, two regression models (with and without control variables) estimate the effect of the license type and social control mechanisms on intentions to violate road rules. Fourth, interaction terms were introduced in regression models to examine the possible moderating effect of license type. Interaction terms were created by multiplying the license type by each of the social control variables (Aiken et al., 1991). To avoid multicollinearity issues, each composite scale was previously standardized. Also, quadratic terms were created with each mechanism of social control to account for non-linear relationships. Correlation and regression analyses were conducted with SPSS 21.

3.2.9. Control variables Ten control variables are used to account for the potential influence of other factors on respondents’ inclination to commit traffic offences. First, gender (0 = male; 1 = female) controls for the overrepresentation of males in traffic offences (Harré et al., 1996; McKnight and McKnight, 2000). Second, several studies show that low self-control is associated with high levels of prohibited behaviors (Gottfredson and Hirschi, 1990; Pratt and Cullen, 2000; Vazsonyi et al., 2017). The self-control scale was created by averaging the scores of 24 Likert-type items varying between 0 and 6. High scores are associated with low self-control (Grasmick et al., 1993). Third, prior involvement in a car crash is a dichotomous variable (1 = yes; 0 = no). Being involved in a car crash can have a deterrent effect on drivers’ inclination to adopt prohibited behaviors (Homel, 1988). Fourth, a variable controls for exposure to police enforcement activities. Such activities effectively prevent traffic violations and crashes (Blais and Dupont, 2005). Exposure to enforcement activities was computed by averaging scores of five items. Respondents were asked if they have been exposed (1 = yes; 0 = no) to specific traffic enforcement activities (such as sobriety checkpoints or speed enforcement activities). The scale ranges from 0 to 1. Fifth, a variable measures the respondent’s association with delinquent peers (that is, peers infringing road rules). Such peers can put pressure on the driver to adopt reckless behaviors and they can serve as models (Homel, 1988; Piquero and Pogarsky, 2002). Respondents were asked to estimate the proportion of their peers who had committed specific traffic violations. The scores of seven Likert-type items, ranging from 1 (none) to 6 (all of them), were averaged to create this composite scale. Sixth, driving experience is often associated with careful driving (Hedlund et al., 2003). Driving experience was estimated by differencing two dates: (1) the acquisition of the driver license; and (2) the date of the survey.

4. Results 4.1. Influence of license type on social control and intentions to violate road rules Propensity scores range from 0 to 1, illustrating the likelihood of being assigned to the experimental group (that is, restricted license) (Apel and Sweeten, 2010). Control variables previously presented were entered in a logistic regression model to assess the balance between groups. The logistic regression also permits the calculation of the propensity score, indicating the likelihood of being assigned to the experimental group. Table 2 shows that both groups are unbalanced. Females are more likely to have a restricted license (OR = 2.09; p ≤ 0.05). Respondents driving daily (OR = 0.41; p ≤ 0.05) and those with more driving experience (OR = 0.92; p ≤ 0.001) are less likely to have a restricted license. The kernel approach was used to match comparable respondents and then to estimate an average treatment effect (ATE) of the license type on social control mechanisms and intentions to violate road rules. The kernel approach generally permits a greater balance between 65

Accident Analysis and Prevention 110 (2018) 62–70

B. Poirier et al.

Table 2 Logistic Regression of License Type on Control Variables.

Table 4 Results of Multiple Regression Analysis: Effects of the License Type and Mechanisms of Social Control on Intentions to Violate Road Rules.

Variables

Odds ratio (OR)

95% CI

Age Gender (1 = male) Crash involvement (1 = yes) Self-control Exposure to police enforcement activities Delinquent peers Driving frequency (1 = driving daily) Car ownership (1 = yes) Driving experience Demerit points (1 = yes) Pseudo R2 Log likelihood

0.99 2.09* 0.65 0.61 0.33 1.08 0.41* 0.83 0.92*** 0.35 0.43*** −116.88

0.76; 1.01; 0.29; 0.36; 0.08; 0.61; 0.18; 0.37; 0.89; 0.11;

1.29 4.36 1.42 1.06 1.28 1.90 0.92 1.85 0.94 1.17

* = p ≤ 0.05; ** = p ≤ 0.01; *** = p ≤ 0.001. Because of pairwise deletion, 368 respondents are included in the model.

experimental and control groups in comparison to the other techniques (Apel and Sweeten, 2010; Becker and Ichino, 2002). Results for the propensity score matching are presented in Table 3. Two main findings come out of these analyses. First, the license type does not significantly influence mechanisms of social control. Second, results indicate that drivers with a restricted license have lower intentions to violate road rules (between 36 and 42 fewer percent points than those with a full license).

Model 1 Beta

Model 2 Beta

License type (1 = restricted)

–

−0.13**

Perceptual variables Perceived risk of arrest Perceived severity of sanctions Threat of social disapproval Perceived seriousness of traffic violations Guilt feelings

0.10* 0.06 −0.08 −0.35*** −0.06

0.01 0.03 −0.08 −0.28*** −0.09*

Control variables Age Sex (1 = female) Previously involved in crash (1 = yes) Self-control Exposure to police enforcement activities Delinquent peers Driving frequency (1 = driving daily) Car ownership (1 = yes) Driving experience Demerit points (1 = yes)

– – – – – – – – – –

−0.07 0.07 0.01 0.22*** 0.07 0.30*** 0.12* 0.03 −0.01 0.01

Constant

2.16***

1.14*

R-squared F-statistic

0.19*** 16.34

0.46*** 17.30

* p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001.

4.2. Results of the correlation and regression analyses

control (b = 0.22; p ≤ 0.001), delinquent peers (b = 0.30; p ≤ 0.001), and driving daily (b = 0.12; p ≤ 0.05) are all positively associated with non-compliance. Models 3–8 (Tables 5–7) integrate interaction terms. These models investigate the moderating effect of license type on the relationships between social control mechanisms and intentions to violate road rules. Since the sample is relatively small, each model integrates one series of interaction terms (that is, interaction terms associated with one mechanism of social control). According to Model 3, there is a negative relationship between the perceived risk of arrest and intentions to infringe road rules (b = −0.14; p ≤ 0.05). The effect of the perceived risk of arrest on intentions exponentially grows, as indicated by its quadratic term (b = −0.27; p ≤ 0.01). Interaction terms are not significant, suggesting that the perceived risk of arrest has the same effect among all respondents. Model 4 shows that the perceived severity of sanctions does not affect non-compliance. Model 5 indicates that threat of social disapproval has a non-linear influence on non-compliance; its quadratic term is statistically significant (b = −0.11; p ≤ 0.05). The effect is similar for both groups of license holders. Results in Model 6 indicate that the perceived seriousness of traffic violations is negatively associated with intentions to violate road rules (b = −0.31; p ≤ 0.001). Interaction terms are not significant in Models 5 and 6. Model 7 (Table 7) indicates that feelings of guilt are negatively associated with intentions to commit traffic violations (b = −0.11; p ≤ 0.05). The quadratic term is not statistically significant. Results in Model 8 show that delinquent peers positively influence intentions to

The correlation matrix is presented in Table A1 in Appendix A. Having a restricted license is negatively associated with intentions to violate road rules (r = −0.28; p ≤ 0.001). The threat of social disapproval (r = −0.22; p ≤ 0.001), the perceived seriousness of traffic violations (r = −0.37; p ≤ 0.001) and guilt feelings (r = −0.18; p ≤ 0.001) are all negatively associated with intentions to infringe road rules. The perceived risk of arrest is not significantly associated with intentions to break the rules (r = 0.06; p > 0.05) while the perceived severity of sanctions is positively related to intentions (r = 0.18; p ≤ 0.001). Table A1 in Appendix A also shows that the license type is associated with several mechanisms of social control. Results suggest that the perceived seriousness of traffic violations (r = 0.13; p ≤ 0.01) is lower among drivers with a restricted license. The latest, however, display higher levels of delinquent peers (r = −0.11; p ≤ 0.05) and perceived risk of arrest (r = −0.11; p ≤ 0.05). At last, correlation coefficients suggest that driving experience is positively associated with demerit points (r = 0.28; p ≤ 0.001) and intentions to infringe road rules (r = 0.11; p ≤ 0.05). Results of multiple regression analyses are presented in Tables 4–7 (Models 1–8). Multiple regression models control for covariance and for other potential confounding factors. According to Model 2 (Table 4), license type (b = −0.13; p ≤ 0.01), perceived seriousness of traffic violations (b = −0.28; p ≤ 0.001) and feelings of guilt (b = −0.09; p ≤ 0.05) are negatively associated with intention to commit traffic violations. In Model 2, three control variables also predict intentions. Self-

Table 3 Results of Propensity Score Matching: Effect of License Type on Social Control and Intentions to Violate Road Rules. Model

Risk of arrest

Severity of sanctions

Threat of social disapproval

Seriousness of traffic violations

Guilt feelings

Intentions to violate road rules

Uniform kernel Gaussian kernel Epanechnikov kernel

0.30 0.21 0.28

−0.09 −0.08 −0.09*

0.11 0.10 0.12

0.30 0.23 0.30

0.26 0.20 0.28

−0.40** −0.36** −0.42**

* = p ≤ 0.05; ** = p ≤ 0.01; *** = p ≤ 0.001 Note: After the matching, the mean bias is inferior to 20% in every model.

66

Accident Analysis and Prevention 110 (2018) 62–70

B. Poirier et al.

Table 5 Effects of the Perceived Risk of Arrest and of the Perceived Severity of Sanctions on Intentions to Violate Road Rules.

License type (1 = restricted)

Model 3 Beta

Model 4 Beta

−0.15*

−0.13*

Mechanisms of formal and informal social control Perceived risk of arrest (Perceived risk of arrest)2 Perceived severity of sanctions (Perceived severity of sanctions)2 Threat of social disapproval Perceived seriousness of traffic violations Guilt feelings

−0.14* −0.27*** 0.04 – −0.07 −0.25*** −0.06

0.01 – 0.06 0.02 −0.08 0.28*** −0.09*

Control variables Age Sex (1 = female) Previously involved in crash (1 = yes) Self-control Exposure to police enforcement activities Delinquent peers Frequency of driving (1 = driving daily) Car ownership (1 = yes) Driving experience Demerit points (1 = yes)

−0.08 0.05 0.02 0.19*** 0.06 0.26*** 0.10* 0.04 0.02 0.01

−0.06 0.07 0.01 0.22*** 0.07 0.29*** 0.12* 0.02 −0.01 0.01

Interaction terms Type of license * Perceived risk of arrest Type of license * (Perceived risk of arrest)2 Type of license * Perceived severity of sanctions Type of license * (Perceived severity of sanctions)2 Constant

0.01 0.07 – – 2.20***

– – −0.06 −0.01 2.03***

R-squared F-statistic

0.49*** 16.57

0.46*** 14.61

Table 6 Effects of the Threat of Social Disapproval and of the Perceived Seriousness of Traffic Violations on Intentions to Violate Road Rules. Model 5 Beta

Model 6 Beta

Type of license (1 = restricted)

−0.16**

−0.10

Perceptual variables Perceived risk of arrest Perceived seriousness of sanctions Threat of social disapproval (Threat of social disapproval)2 Perceived seriousness of traffic violations (Perceived seriousness of traffic violations)2 Guilt feelings

0.01 0.02 −0.09 −0.11* −0.28*** – −0.09*

0.02 0.03 −0.08 – −0.31*** 0.06 −0.08

−0.05 0.06 0.01 0.21*** 0.07 0.30*** 0.12** 0.02 −0.01 0.01

−0.07 0.06 0.02 0.20*** 0.07 0.29*** 0.13** 0.02 −0.01 0.01

0.03 0.07 –

– – 0.07

–

−0.09

2.20***

2.01***

0.46*** 14.89

0.47*** 15.13

Control variables Age Gender (1 = female) Previously involved in crash (1 = yes) Self-control Exposure to police traffic enforcement Delinquent peers Frequent driving (1 = driving daily) Car ownership (1 = yes) Driving experience Demerit points (1 = yes) Interaction terms Type of license * Threat of social disapproval Type of license * (Threat of social disapproval)2 Type of license * Perceived seriousness of traffic violations Type of license * (Perceived seriousness of traffic violations)2 Constant R-squared F-statistic

* p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001. * p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001.

violate road rules (b = 0.28; p ≤ 0.001). This relationship is also linear. However, the influence of delinquent peers is weaker for respondents with a restricted license. The interaction term (Restricted license * (Delinquent peers)2) is significant (b = −0.21; p ≤ 0.001).

Table 7 The Effect of Guilt Feelings and Delinquent Peers on Intentions to Violate Road Rules. Model 7 Beta

Model 8 Beta

Type of license (1 = restricted)

−0.12*

−0.05

Mechanisms of informal and formal social control Perceived risk of arrest Perceived severity of sanctions Threat of social disapproval Perceived seriousness of traffic violations Guilt feelings (Guilt feelings)2

0.01 0.04 −0.07 −0.27*** −0.11* −0.10

0.01 0.03 −0.07 −0.29*** −0.10* –

−0.07 0.06 0.01 0.21*** 0.06 0.29*** – 0.10* 0.05 0.01 0.01

−0.07 0.07 −0.01 0.20*** 0.08 0.28*** 0.07 0.14** 0.02 0.01 −0.01

0.01 −0.01 – – 2.07***

– – 0.06 −0.21*** 2.01***

0.46*** 14.95

0.48*** 15.74

5. Discussion The purpose of this study was to scrutinize: (1) the effect of license type on mechanisms of social control and on intentions to violate road rules; and (2) the possible moderating effect of license type on formal and informal mechanisms of social control. In sum, the present findings indicate that driving restrictions do not elevate the level of mechanisms of social control but they do lower intentions to violate road rules. The restricted license also moderates the effect of delinquent peers on intentions to infringe road rules. The effect of delinquent peers on noncompliance was weaker among respondents with a restricted license. These findings can be further discussed to address issues related to the effectiveness of GDL and the relevance of the differential deterrence theory. The present findings indicate that GDL really works. While several systematic reviews report that GDL reduces crash risk among novice drivers (Russell et al., 2011; Shope, 2007; Williams and Shults, 2010), few studies have investigated mechanisms responsible for its effectiveness (Simpson, 2003). Some researchers consequently argue that the lower exposure to at-risk situations (such as nighttime driving) and restricted driving conditions (such as supervised driving) were responsible for the lower crash rates observed among drivers with a restricted license (Curry et al., 2015; Groeger and Banks, 2007; KaracaMandic and Ridgeway, 2010). Results from the propensity score matching analysis address this issue. Respondents with a restricted license do have less experience and drive on an occasional basis in comparison to respondents with a full license. Yet, when respondents

Control variables Age Sex (1 = female) Previously involved in crash (1 = yes) Self-control Exposure to police traffic enforcement Delinquent peers (Delinquent peers)2 Driving frequency (1 = driving daily) Car ownership (1 = yes) Driving experience Demerit points (1 = yes) Interaction terms Type of license * Guilt feelings Type of license * (Guilt feelings)2 Type of license * Delinquent peers Type of license * (Delinquent peers)2 Constant R-squared F-statistic

67

Accident Analysis and Prevention 110 (2018) 62–70

B. Poirier et al.

du Québec (the traffic safety code) since data collection. Since 2010, new drivers must successfully complete a driving course to hold a probationary license. Since June 2011, zero-tolerance for drinking-anddriving applies to all drivers under 21. In April 2012, the demerit point system was amended for drivers younger than 25. Drivers under 23 have a maximum of eight demerit points while drivers between 23 and 24 have 12 points. The 15 points are only awarded to drivers 25 and older. Further research is necessary to appraise the effect of these changes on the behaviors of young drivers in Quebec. Third, any generalization to all 18–25 years old drivers remains tenuous. Our respondents are not representative of the young driver population of the Province of Quebec. Our sample mainly consists of undergraduate university students living in the greater Montreal area. Again, other studies should be conducted with drivers from other jurisdictions and with different GDL programs.

with a restricted license are matched with fully-licensed respondents with a similar propensity score, the former still display lower levels of non-compliance. Intentions to violate the rules among drivers with a restricted license are not associated with their experience level nor with their driving frequency. Differential deterrence seeks to pinpoint factors moderating the relationship between sanction threat and offending. In a demerit point system, restrictions are likely to moderate the effect of formal and informal mechanisms of social control on traffic violations. Some mechanisms of social control influence all young drivers while some are more effective among participants with a restricted license. As reported in other studies (Freeman and Watson, 2006; Snortum et al., 1986), moral commitment–or the perceived seriousness of traffic violations − is associated with lower inclination to infringe road rules. It is the strongest mechanism of social control. Our findings also support the deterrent impact of the perceived risk of arrest (Freeman et al., 2016; Freeman and Watson, 2009), but the effect is quadratic (Kane, 2006; Yu and Liska, 1993). Threat of social disapproval and guilt feelings did not seem to significantly affect non-compliance; relationships were either weak or non-significant across models regardless of the license type. Results also show that a restricted license exerts additional control on the respondents’ intentions to violate road rules. In all regression models, delinquent peers are associated with greater levels of noncompliance (see Piquero and Paternoster, 1998; Simons-Morton et al., 2012 for similar findings). The coefficient of the interaction term indicates that the effect of peers is limited among respondents with a restricted driver license. A similar finding is reported by Homel (1988) in his seminal work on random breath testing (RBT) in New South Wales, Australia. The introduction of RBT provided drivers with legitimate excuses and other arguments to resist peer pressure to drink and drive. Other studies also show that young drivers could use probable informal consequences (e.g. my parents will forbid me from using their car) to resist social pressure (Allen et al., 2015; Bates et al., 2015). Conversely, our findings indicate that delinquent peers produce greater effects among respondents with a full license. This could explain the sudden increase in the crash risk among young drivers in the months following the acquisition of a full licensure. For instance, a study shows that driver deaths per 1000 crashes increase for 16 and 17 years old transporting passengers younger than 30 years old (Chen et al., 2000). Another study indicates that the increased risk of fatal and non-fatal crash persists up to 29 years old (Braver and Trempel, 2004). Since the raised crash risk persists up to the late 20s, GDL could be extended and, accordingly, restrictions could be lifted more gradually.

7. Conclusions and future research directions Our findings indicate that respondents with a restricted license are less inclined to infringe road rules than those with a full license, indicating that the preventive effect of some mechanisms of social control appears to be conditional upon the license type. It appears that restrictions − namely a limited number of demerit points and zero-tolerance for drinking-and-driving − increase one’s ability to resist peer pressure to infringe road rules. Other mechanisms of social control are not contingent upon the license type. Moral commitment (that is, the internalization of the norm) and, to a lesser extent, the perceived risk of arrest are both negatively associated with intentions to infringe road rules among all respondents. The present findings call for additional studies on the relationship between driving experience and road rules violation. In our study, a positive relationship was observed between driving experience and demerit points (a proxy of previous violations). One has to admit that the concept of experience is relatively broad in GDL research; it needs to be further refined in order to understand its association with driving behaviors. For instance, vicarious and direct experiences with punishment are likely to shape perceptions about the risk of being arrested for an offense (Piquero and Paternoster, 1998; Stafford and Warr, 1993). Hence, experiences with punishment and punishment avoidance both influence one’s perception about the threat of punishment. Accordingly, experience can either exacerbate or prevent prohibited driving behaviors (Useche et al., 2015). As last, our findings also call for additional research on young drivers enrolled in GDL, their perceptions, and their intentions to or their actual violations of road rules (see also Ivers et al. (2009) for similar suggestions regarding future research avenues). Such concerns reiterate the need to conduct research on the policy-to-perception-to-behavior link (Ivers et al., 2009; Nagin, 2013) to have an improved understanding about the effect of GDL on traffic safety.

6. Limitations of the study Besides the potential social desirability bias inherent to all self-report surveys (Grimm, 2010), the present findings need to be appreciated in light of specific limitations. First, we had to combine respondents with a learner and a probationary license. Some studies, however, report variations in non-compliance rates for the various stages of GDL (Allen et al., 2015; Bates et al., 2015). If possible, future studies should investigate the preventive role of social control mechanisms for all stages of GDL. Second, changes have been made to the Code de la Sécurité Routière

Acknowledgements The Social Sciences and Humanities Research Council of Canada funded part of this study (grant number: 410-2011-2555). The authors also thank Katherine Pendakis for editing this article.

Appendix A

68

Accident Analysis and Prevention 110 (2018) 62–70 1 −0.18***

References

1 −0.26*** −0.14** −0.04 0.05 −0.04 0.07 .11* .67*** .22*** −0.01 −0.02 −0.09 −0.06 0.01 −0.02

1 0.06 −0.14** 0.03 −0.05 0.04 −0.04 −0.20*** −0.15** −0.07 0.01 0.01 .20*** .16** −0.02

1 0.01 0.06 0.07 0.11* 0.08 −0.03 0.05 0.01 −0.04 0.04 0.03 0.09 0.05

1 0.02 .24*** 0.07 0.01 −0.01 0.08 .16** .15** 0.08 −0.15** −0.08 .33***

1 .19*** 0.24*** .30*** .15** .21*** .11* −0.05 0.03 .13** 0.03 .18***

1 0.17*** .13** 0.06 .17*** 0.07 0.09 −0.17*** −0.17*** −0.04 .46***

1 .57*** .25*** .35*** 0.09 −0.04 −0.07 −0.11* −0.01 .31***

.30*** .24*** .10* −0.07 −0.03 −0.12* −0.03 .23***

1 .28*** 0.04 −0.03 −0.06 −0.07 −0.01 .11*

1 0.07 0.06 −0.14** −0.16** −0.03 .21***

1 0.06 0.03 0.08 0.02 0.06

1 −0.22*** −0.27*** −0.29*** .18***

1 .36*** .29*** −0.22***

1 .26*** −0.37***

Aiken, L.S., West, S.G., Reno, R.R., 1991. Multiple Regression: Testing and Interpreting Interactions. Sage. Allen, S., Murphy, K., Bates, L., 2015. What drives compliance? The effect of deterrence and shame emotions on young drivers’ compliance with road laws. Policing Soc. 1–15. Andenaes, J., 1974. Punishment and Deterrence. University of Michigan Press, Ann Arbor. Apel, R.J., Sweeten, G., 2010. Propensity score matching in criminology and criminal justice. In: Piquero, A.R., Weisburd, D. (Eds.), Handbook of Quantitative Criminology. Springer, New York, pp. 543–562. Basili, M., Nicita, A., 2005. Deterrence and Compliance in a Demerit Point System. Dipartimento Di Economia Politica, Universita Degli Studi Di Siena. Bates, L., Darvell, M.J., Watson, B., 2015. Young and unaffected by road policing strategies: using deterrence theory to explain provisional drivers’ (non)compliance. Aust. N. Z. J. Criminol. 50 (1), 23–38. Becker, S.O., Ichino, A., 2002. Estimation of average treatment effects based on propensity scores. Stata J. 2 (4), 358–377. Blais, É., Dupont, B., 2005. Assessing the capability of intensive police programmes to prevent severe road accidents a systematic review. Br. J. Criminol. 45 (6), 914–937. Blais, É., Ouimet, M., 2005. The effect of legal interventions on fatal accidents and associated to driving while impaired by alcohol (DWI) in Quebec between 1980 and 2001. Can. J. Criminol. Criminal Justice 47, 545–578. Braver, E.R., Trempel, R.E., 2004. Are older drivers actually at higher risk of involvement in collisions resulting in deaths or non-fatal injuries among their passengers and other road users? Inj. Prev. 10 (1), 27–32. Brown, S.W., Hing, M.M., Vanlaar, W.G.M., Robertson, R.D., 2016. Road safety monitor 2016: drinking and driving in Canada. Traffic Inj. Res. Found. Castillo-Manzano, J.I., Castro-Nuño, M., 2012. Driving licenses based on points systems: efficient road safety strategy or latest fashion in global transport policy? A worldwide meta-analysis. Transp. Policy 21, 191–201. Chen, L.-H., Baker, S.P., Braver, E.R., Li, G., 2000. Carrying passengers as a risk factor for crashes fatal to 16-and 17-year-old drivers. J. Am. Med. Assoc. 283 (12), 1578–1582. Chen, L.H., Braver, E.R., Baker, S.P., Li, G., 2001. Potential benefits of restrictions on the transport of teenage passengers by 16 and 17 year old drivers. Inj. Prev. 7 (2), 129–134. Conner, K.A., Smith, G.A., 2017. An evaluation of the effect of Ohio’s graduated driver licensing law on motor vehicle crashes and crash outcomes involving drivers 16–20 years of age? Traffic Inj. Prev. 18 (4), 344–350. Cooper, D., Atkins, F., Gillen, D., 2005. Measuring the impact of passenger restrictions on new teenage drivers. Accid. Anal. Prev. 37 (1), 19–23. Curry, A.E., Pfeiffer, M.R., Durbin, D.R., Elliott, M.R.1, 2015. Young driver crash rates by licensing age, driving experience, and license phase. Accid. Anal. Prev. 80, 243–250. Dee, T.S., Grabowski, D.C., Morrisey, M.A., 2005. Graduated driver licensing and teen traffic fatalities. J. Health Econ. 24 (3), 571–589. Doherty, S.T., Andrey, J.C., MacGregor, C., 1998. The situational risks of young drivers: the influence of passengers: time of day and day of week on accident rates. Accid. Anal. Prev. 30 (1), 45–52. Elvik, R., 2010. Why some road safety problems are more difficult to solve than others. Accid. Anal. Prev. 42 (4), 1089–1096. Fell, J.C., Jones, K., Romano, E., Voas, R., 2011. An evaluation of the graduated driver licensing effects on fatal crash involvements of young drivers in the United States. Traffic Injury Prevention 12, 423–431. Freeman, J., Watson, B., 2006. An application of Stafford and Warr’s reconceptualisation of deterrence to a group of recidivist drink drivers. Accid. Anal. Prev. 38 (3), 462–471. Freeman, J., Watson, B., 2009. Drink driving deterrents and self-reported offending behaviours among a sample of Queensland motorists. J. Safety Res. 40 (2), 113–120. Freeman, J., Szogi, E., Truelove, V., Vingilis, E., 2016. The law isn’t everything: the impact of legal and non-legal sanctions on motorists’ drink driving behaviors. J. Safety Res. 59, 53–60. Geerken, M.R., Gove, W.R., 1975. Deterrence: some theoretical considerations. Law Soc. Rev. 497–513. Gottfredson, M.R., Hirschi, T., 1990. A General Theory of Crime. Stanford University Press, Stanford. Grasmick, H.G., Bursik, R.J., 1990. Conscience significant others, and rational choice: extending the deterrence model. Law Soc. Rev. 24 (3), 837–861. Grasmick, H.G., Green, D.E., 1981. Deterrence and the morally committed. Sociological Q. 22 (1), 1–14. Grasmick, H.G., Scott, W.J., 1982. Tax evasion and mechanisms of social control: a comparison with grand and petty theft. J. Econ. Psychol. 2 (3), 213–230. Grasmick, H.G., Tittle, C.R., Bursik, R.J., Arneklev, B.J., 1993. Testing the core empirical implications of gottfredson and hirschi’s general theory of crime. J. Res. Crime Delinquency 30 (1), 5–29. Gregersen, N.P., Nyberg, A., Berg, H.-Y., 2003. Accident involvement among learner drivers—an analysis of the consequences of supervised practice. Accid. Anal. Prev. 35 (5), 725–730. Grimm, P., 2010. Social Desirability Bias. Wiley International Encyclopedia of Marketing. Groeger, J.A., Banks, A.P., 2007. Anticipating the content and circumstances of skill transfer: unrealistic expectations of driver training and graduated licensing? Ergonomics 50 (8), 1250–1263. Harré, N., Field, J., Kirkwood, B., 1996. Gender differences and areas of common concern in the driving behaviors and attitudes of adolescents. J. Safety Res. 27 (3), 163–173.

*p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001.

1 −0.30*** .16** −0.10* −0.09 −0.23*** −0.11* −0.35*** −0.34*** −0.52*** −0.27*** −0.11* 0.01 0.05 .13** 0.01 −0.28*** 1. License type 2. Age 3. Gender 4. Previous crash 5. Self-control 6. Exposure to traffic enforcement 7. Delinquent peers 8. Driving frequency 9. Car ownership 10. Driving experience 11. Demerit points 12. Perceived risk of arrest 13. Perceived severity of sanctions 14. Threat of social disapproval 15. Perceived seriousness of traffic violations 16. Guilt feelings 17. Intentions to violate road rules

Table A1 Correlation Matrix.

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

B. Poirier et al.

69

Accident Analysis and Prevention 110 (2018) 62–70

B. Poirier et al.

internal validity. Experimental and Quasi-Experimental Designs for Generalized Causal Inference. pp. 45–48. Shope, J.T., 2007. Graduated driver licensing: review of evaluation results since 2002. J. Safety Res. 38 (2), 165–175. Simons-Morton, B., Lerner, N., Singer, J., 2005. The observed effects of teenage passengers on the risky driving behavior of teenage drivers. Accid. Anal. Prev. 37 (6), 973–982. Simons-Morton, B.G., Ouimet, M.C., Chen, R., Klauer, S.G., Lee, S.E., Wang, J., Dingus, T.A., 2012. Peer influence predicts speeding prevalence among teenage drivers? J. Safety Res. 43 (5), 397–403. Simpson, H.M., 2003. The evolution and effectiveness of graduated licensing. J. Safety Res. 34 (1), 25–34. Snortum, J.R., Hauge, R., Berger, D.E.1, 1986. Deterring alcohol-impaired driving: a comparative analysis of compliance in Norway and the United States. Justice Q. 3 (2), 139–165. Société de l’assurance automobile du Québec, 2016. Données et statistiques 2015. (Retrieved from). https://saaq.gouv.qc.ca/fileadmin/documents/publications/ donnees-statistiques-2015.pdf. Stafford, M.C., Warr, M., 1993. A reconceptualization of general and specific deterrence. J. Res. Crime Delinquency 30 (2), 123–135. Useche, S., Serge, A., Alonso, F., 2015. Risky behaviors and stress indicators between novice and experienced drivers. Am. J. Appl. Psychol. 3 (1), 11–14. Vanlaar, W., Mayhew, D., Marcoux, K., Wets, G., Brijs, T., Shope, J., 2009. An evaluation of graduated driver licensing programs in North America using a meta-analytic approach. Accid. Anal. Prev. 41 (5), 1104–1111. Vazsonyi, A.T., Mikuška, J., Kelley, E.L., 2017. It’s time: a meta-Analysis on the selfControl-Deviance link. J. Criminal Justice 48, 48–63. Williams, A.F., Shults, R.A., 2010. Graduated driver licensing research, 2007-present: a review and commentary. J. Safety Res. 41 (2), 77–84. Williams, 2007. Contribution of the components of graduated licensing to crash reductions. J. Safety Res. 38 (2), 177–184. Yu, J., Liska, A.E., 1993. The certainty of punishment: a reference group effect and its functional form. Criminology 31, 447. Zimring, F.E., Hawkins, G., Vorenberg, J., 1973. Deterrence: The Legal Threat in Crime Control. University of Chicago Press, Chicago. Zimring, F.E., Hawkins, G., Kamin, S., 2001. Punishment and Democracy: Three Strikes and You’re Out in California. Oxford University Press on Demand.

Hedlund, J., Shults, R.A., Compton, R., 2003. What we know, what we don’t know, and what we need to know about graduated driver licensing. J. Safety Res. 34 (1), 107–115. Homel, R.J., 1988. Policing and punishing the drinking driver. A Study of Specific and General Deterrence. Springer-Verlag, New York. 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. Am. J. Public Health 99 (9), 1638–1644. Kane, R.J., 2006. On the limits of social control: structural deterrence and the policing of suppressible crimes. Justice Q. 23 (02), 186–213. Karaca-Mandic, P., Ridgeway, G., 2010. Behavioral impact of graduated driver licensing on teenage driving risk and exposure. J. Health Econ. 29 (1), 48–61. Kennedy, D.M., 2009. Deterrence and Crime Prevention: Reconsidering the Prospect of Sanction. Routledge, New York, NY. Masten, S.V., Foss, R.D., Marshall, S.W., 2011. Graduated driver licensing and fatal crashes involving 16-to 19-year-old drivers. J. Am. Med. Assoc. 306 (10), 1098–1103. McKnight, A.J., McKnight, A.S., 2000. The behavioral contributors to highway crashes of youthful drivers. Annual Proceedings/Association for the Advancement of Automotive Medicine, vol. 44, 321 (Association for the Advancement of Automotive Medicine). Nagin, D.S., 2013. Deterrence in the twenty-First century. Crime Justice 42 (1), 199–263. Piquero, A.R., Paternoster, R., 1998. An application of stafford and warr’s reconceptualization of deterrence to drinking and driving. J. Res. Crime Delinquency 35 (1), 3–39. Piquero, A.R., Pogarsky, G., 2002. Beyond stafford and warr’s reconceptualization of deterrence: personal and vicarious experiences impulsivity, and offending behavior. J. Res. Crime Delinquency 39 (2), 153–186. Piquero, A.R., Paternoster, R., Pogarsky, G., Loughran, T., 2011. Elaborating the individual difference component in deterrence theory. Ann. Rev. Law Soc. Sci. 7 (1), 335–360. Pratt, T.C., Cullen, F.T., 2000. The empirical status of Gottfredson and Hirschi’s general theory of crime: a meta-analysis. Criminology 38 (3), 931–964. Rosenbaum, P.R., Rubin, D.B., 1983. The central role of the propensity score in observational studies for causal effects? Biometrika 70 (1), 41–55. Russell, K.F., Vandermeer, B., Hartling, L., 2011. Graduated driver licensing for reducing motor vehicle crashes among young drivers. Cochrane Database of Systematic Reviews. (10). Shadish, W.R., Cook, T.D., Campbell, D.T., 2002. Statistical conclusion validity and

70