1. Introduction

Teenagers are over-represented in a variety of risky behaviors (Arnett, 1996; Jonah, 1990), and peer presence and influence have often been associated with higher involvement in these behaviors. For example, smoking (Simons-Morton and Farhat, 2010) and drinking (Borsari and Carey, 2001) are more common among teenagers and young adults when their friends engage in these behaviors. Also, teenage drivers accompanied by teenage passengers are more frequently involved in fatal crashes (Chen et al., 2000; Ouimet et al., 2010). Traffic crashes represent the main cause of morbidity and mortality for teenagers (World Health Organization, 2009) and are one of the most serious consequences of negative peer influences. A better understanding of teenage passenger and driver interactions could help guide development of traffic injury prevention efforts.

Risky driving and inattention (due to distraction), which can be affected by teenage passengers, are two of the main behaviors preceding crashes among young drivers (Curry et al., 2012; McKnight and McKnight, 2003). While peer presence is commonly associated with risky behavior, its effect on inattention is probably unique to driving compared to other types of risky behavior. Most observational and experimental studies have shown that the presence of teenage passengers is associated with higher involvement in (i) risky driving practices such as speeding (Shepherd et al., 2011; Simons-Morton et al., 2005; Toxopeus et al., 2011), shorter headway (Simons-Morton et al., 2005), lower safety belt use (Williams et al., 2003); and (ii) inattention such as higher “looked-but-failed-to-see” driving errors (White and Caird, 2010), lower identification of and reaction time to hazardous situations (Gugerty et al., 2004), and more driving errors (Rivardo et al., 2008; Toxopeus et al., 2011). Some epidemiological and observational studies have found stronger negative effects when both the driver and passenger were males (Chen et al., 2000; Ouimet et al., 2010; Simons-Morton et al., 2005). Other studies, however, have indicated beneficial or no effects of passengers. Lower risky driving practices were found for teenage drivers traveling with peer passengers in a naturalistic study (Simons-Morton et al., 2011) and faster reaction times were observed during simulated driving (Toxopeus et al., 2011). Some experimental studies focusing on distraction found no difference between driving with a passenger and driving alone (Drews et al., 2008) or performance deterioration when young drivers were accompanied by a passenger who was talking to them, but not when the passenger was silent (Rivardo et al., 2008; Toxopeus et al., 2011). While many studies demonstrate elevated risk in the presence of teenage passengers, the existence of mixed results suggests that there might be specific conditions under which teenage driving risk is increased or decreased in the presence of passengers.

Another important question is whether both peer influence and distraction affect teenage driving while in the presence of teenage passengers. Most experimental studies have assumed that either peer influence leads to risky driving practices (Oei and Kerschbaumer, 1990; Shepherd et al., 2011) or that distraction results in inattention (Gugerty et al., 2004; Rivardo et al., 2008; Toxopeus et al., 2011; White and Caird, 2010). Both peer influence and distraction could occur under the same circumstances, however. For example, in research on the effects of distraction on driver performance during which study confederates engaged participants in a conversation (Gugerty et al., 2004; Rivardo et al., 2008; Toxopeus et al., 2011; White and Caird, 2010), effects on driver behavior could also be attributed to perceived social norms conveyed by the passengers. Similarly, if the confederate applies direct peer pressure on the teenager to engage in a risky behavior (Shepherd et al., 2011), there could also be an effect of distraction. It appears that both peer influence and distraction can play a role and their effects could be addressed in the same research design.

Finally, most experimental studies have focused on the effect of peer pressure, though other modes of peer influence may be involved. In general, peers can influence behavior by directly encouraging or discouraging it or by modeling the behavior. They can also influence behavior through their general attitude, expectations, and judgments by suggesting how normative and acceptable a behavior is (or perceptions of peer social norms) and thereby promoting or discouraging a particular behavior. The intensity of these modes of peer influence can also vary. Influences can be direct, such as through verbal and overt non-verbal expressions of encouragement (or discouragement). Influences can also operate indirectly through social norms, which can be transmitted through modeling and verbal and non-verbal actions. In the adoption of smoking, the indirect effect of peers appears to be more important than direct pressure (Nichter et al., 1997), but both processes have been demonstrated in other studies (Simons-Morton and Farhat, 2010). With respect to driving, a pilot study conducted by our research group indicated that 70% of the participants reported that the presence of a teenage passenger (who is in the vehicle, but not talking) impacted their driving performance compared to 25% who reported that direct peer pressure (comments made to the driver) played a role (Ouimet, 2009). These results contrast with those of experimental studies showing no effect of a silent passenger (Rivardo et al., 2008; Toxopeus et al., 2011) but significant effects on increased risky driving behavior of a confederate expressing peer pressure verbally before the driving session (e.g., Oei and Kerschbaumer, 1990). If most of the influence of peer presence leading to increased risk is through perceived social norms and less through direct peer pressure, more research is needed on the role of subtler modes of peer influence on increased risky driving practices and inattention, such as social norms conveyed by a silent and stationary passenger.

Some models, such as peer influence and social facilitation, can help explain the required conditions for increased or decreased risky driving practices and inattention in the presence of a passenger and reconcile some of the discrepant findings in the literature. The conceptual model of peer influence proposed by Brown and colleagues (2008) suggests that four elements are involved in peer influence in teenagers: an event, activation of peer influence, a response, and generation of a measurable outcome. Thus, when applied to the passenger effect, the presence of a teenage passenger in a vehicle driven by a teenager would activate peer influence that can be accepted/acceded, rejected/ignored, or countered by the driver. The selected response may affect risky driving and inattention. The effect of peers on risky driving practices and inattention can also be explained by social facilitation models, which posit that, in the presence of others, the performance of an easy and familiar task is enhanced whereas the performance of a complex and unfamiliar task is degraded (Zajonc, 1965). A meta-analysis of the evidence in support of this perspective concluded that the presence of others is associated with higher speed and accuracy for the completion of simple tasks and lower speed and accuracy for complex tasks (Bond and Titus, 1983). The effect of the presence of others is postulated to be due to increased arousal (Guerin, 1986; Zajonc, 1965) that can be generated by apprehension of being evaluated by others (Cottrell et al., 1968) or increased cognitive load and distraction if the participant's attention is divided between the demands of the task and the social situation (Baron and Kenny, 1986; Sanders and Baron, 1975). The presence of others can also have an effect if a person's attention is focused on reducing the difference between their behavior and perceived conformity to norms (Carver and Scheier, 1981). Applied to driving, it could mean that if the presence of a passenger is meaningful to the driver, it could affect risky behavior and attention as a function of the difficulty and complexity of the task. In summary, the effect of the presence and influence of peer passengers might be more complex than previously known.

Both peer influence and social facilitation models have often ignored individual factors in their explanation of the effects of peers (Brown et al., 2008; Uziel, 2007). Nevertheless, a systematic review of social facilitation models found that participants’ characteristics were more important moderators of the effects of the presence of others on performance than the complexity of the task (Uziel, 2007). This review also pointed out performance improvement for people with positive-self-assured attributes (e.g., high self-esteem) and performance impairment for those with negative-apprehensive attributes (e.g., low self-esteem). From the peer influence purview, individual factors in both drivers and passengers are considered to moderate the response to peer influence (e.g., accepting, refusing). These factors include male sex (Cooper et al., 1979; Simons-Morton et al., 2005), high sensation seeking (Slater, 2003), openness or susceptibility to peer influence (Brown et al., 2008), peer characteristics such as sociability (Collins et al., 1985), and having friends who are involved in risky behavior (Shope et al., 2003; Simons-Morton et al., 2011). It is then important to investigate the influence of driver and passenger individual characteristics on driving performance.

In summary, a better understanding of how peers affect teenage driver behavior, the nature of the processes underlying the effects of peers, and the role of teenagers’ characteristics could help guide the development of traffic injury prevention efforts. Given the results of studies showing stronger effects when both drivers and passengers are males (Chen et al., 2000; Ouimet et al., 2010; Simons-Morton et al., 2005), the purpose of this study was to determine the effect on male teenage driving behavior of (i) the presence of a male teenage passenger; and (ii) passenger characteristics intended to influence the driver's perceived social norms (i.e., risk-accepting vs. risk-averse passenger). It was hypothesized that teenagers would engage in more risky driving practices and be less attentive in the presence of a passenger (vs. driving alone) and with a risk-accepting (vs. risk-averse) passenger in a driving simulator. An exploratory goal of the study was to examine the relationship between driver individual characteristics and driving performance with a passenger.

2. Methods

2.1 Participants

Male teenage drivers were recruited from high schools and driving schools in Amherst, Massachusetts, United States. In that state, drivers can receive their learner's permit at age 16 and their provisional license (called a junior operator's license; allowing unsupervised driving) at age 16 and a half. An information sheet was distributed to adolescents at high schools and driving schools. Candidates who contacted the research team were pre-screened for inclusion and exclusion criteria over the phone. Inclusion criteria were: (i) male sex; (ii) being 16 or 17 years old; (iii) possessing a provisional driving license; (iv) having driven in the past 3 months; and v) both parental consent and teenage assent. Exclusion criteria were: (i) frequently experiencing motion sickness, either as a passenger or driver, or in other modes of transport; (ii) non-normal or uncorrected vision (i.e., contacts lenses or eyeglasses were accepted). The study required about 75 minutes for which participants received a $25.00 USD compensation. The study protocol received approval from the Institutional Review Board of the University of Massachusetts Amherst.

2.2 Study design, random assignment, and concealment

This study employed a 2 X (2) mixed design, i.e., a between Passenger Risk Group condition (accepting vs. averse) and a within Passenger Presence condition (passenger vs. no passengers). Hence, each participant had two drives, one with no passengers and the other with a randomly assigned risk-accepting or risk-averse passenger. Restricted randomization (i.e., the random allocation rule) was used to assign participants to a Passenger Risk Group by an investigator not involved in the testing procedure. The order of the drives (driving alone first vs. driving with the passenger first) was also randomized to account for potential learning effects from repeated exposure to a similar task. Participants and data analysts were blind to the assignment. Passenger type was concealed from the research assistants until the confederate entered the testing room and was introduced to participants.

2.3 Driving simulation and equipment

There were two drive sets comprised of three continuous drives simulating a rural, a suburban, and an urban environment. The drives contained a lead vehicle in front of the participant. A Realtime Technologies fixed-base driving simulator and an eye tracker from ASL Mobile Eye collected data on risky driving practices and inattention measures. The simulator uses a 1995 Saturn sedan equipped to respond to simulated roadways displayed on three large screens in the front and sides of the vehicle providing a 135-degree horizontal field of view. When the driver accelerates, turns, decelerates, etc., the vehicle appears to progress on the simulated road as it would on an actual road. The eye tracker monitors eye glance and horizontal eye movement patterns by way of cameras and a reflector attached to the safety glasses worn by participants. It superimposes cross hairs indicating the driver's fixation point on a video.

2.4 Dependent Variables

Risky driving measures included the percentage of time above the speed limit, headway, and gap acceptance. Inattention measures were eye glances at hazards and horizontal eye movement. The percentage above speed limit was measured at several sections in the rural drive for 15-30 seconds. These sections had posted speed limits of 30 or 45 mph (approximately 50 km/h or 80 km/h). As there was a lead vehicle, the variable was not measured under free speed conditions; however, the locations were chosen such that the lead vehicle was not visible to participants during the measurements. The distance headway in meters, i.e., the average distance between the participant's vehicle and the lead vehicle, was measured in one section of the rural drive for 20-25 seconds. Gap acceptance for left turns at four signalized intersections was measured in the urban and suburban drives. Participants, who followed the lead vehicle, had to make a left turn at the intersection while there was a steady stream of 10 oncoming vehicles from the opposing lane. The number of oncoming vehicles that the participant waited for before making the left turn was recorded, and the mean was calculated for the four intersections. Eye glances at hazards were measured during the rural drives that contained five scenarios, derived from scenarios tested in prior studies (e.g., Pollatsek et al., 2006; Pradhan et al., 2005), where a potential hazard could be obscured from the driver's line of sight (e.g., a pedestrian in a crosswalk obscured from the driver's view by a large vehicle stopped in the parking lane). The mean number of eye glances at predetermined “target zones” (on the forward view) was calculated. Horizontal eye movement refer to the average standard deviation of the horizontal gaze position ( i.e., the spread of horizontal scanning); it was measured in pixels at different points in the rural drive when no other variables were collected.

2.5 Questionnaires

Demographic information and data on driving exposure in terms of miles driven in the past week, time since licensing, driving experience with peer passengers, citations and crashes were gathered. Risky driving behavior in the past week was measured with a 19-item questionnaire on a 1 “never” to 5 “always” scale with good internal consistency (Cronbach's alpha > 0.88) (Simons-Morton et al., 2006b). It includes behaviors such as driving under the influence of alcohol or drugs and driving unbelted. The risk perception of crash/injury for newly licensed teenagers was measured with a 14-item questionnaire on a 1 “lowest risk” to 5 “highest risk” scale with good internal consistency (Cronbach's alpha > 0.88) (Simons-Morton et al., 2006a). The Brief Sensation Seeking Scale was designed to measure sensation seeking in adolescents (Hoyle et al., 2002; Stephenson et al., 2003) and contains eight items derived from the 40-item Sensation Seeking Scale Form V (Zuckerman, 1994). Responses are on a 5-item scale from “strongly disagree” to “strongly agree.” Internal consistency was acceptable (Cronbach's alpha = 0.76) (Hoyle et al., 2002). Self-esteem was measured with the 10-item Rosenberg Self-Esteem Scale with ratings ranging from 1 “strongly disagree” to 5 “strongly agree” (Rosenberg, 1989) and good internal consistency (Cronbach's alpha > 0.80) (Schmitt and Allik, 2005). The Tolerance of Deviance measure contains 5 items on a 4-point scale from “very wrong” to “not wrong” and has good internal consistency (Cronbach's alpha > 0.80) (Rachal et al., 1975; Shope et al., 2003). The Susceptibility to Peer Pressure measure has 7 questions on a 4-point scale (i.e., “no,” “probably not,” “probably,” and “yes”) (Dielman et al., 1987) and has good internal consistency (Cronbach's alpha > 0.85) (Shope et al., 2003). The success of the social norm manipulation (or perception of passenger) was evaluated using 10 questions related to descriptive and injunctive norms, inspired by previous work from Parker and colleagues (1992). Participants were asked the following questions: “As best you can tell, the passenger seemed like someone who probably: is quiet, likes to party, would rather stay home doing homework than going out with friends, seeks thrilling experiences, drives conservatively, switches lanes to weave through slower traffic, drives after drinking alcohol or using illegal drugs, plays the radio so loudly he wouldn't be able to hear other vehicles’ horns or sirens, would approve of you following a slow vehicle very closely, and would approve of you driving after drinking two glasses of alcohol in the hour before driving”. Participants had to report their impression of the passenger on a 5-point scale from “very unlikely” to “very likely”. Internal consistency was good (Cronbach's alpha = 0.84).

2.6 Procedure

Participants had a 5-minute practice drive and then drove the simulator, once without the passenger and once with the passenger (randomized order), who played either the risk-accepting or risk-averse role, according to the randomized schedule. The passenger was a male confederate (the same throughout the entire study) who portrayed both types of passengers differing only in terms of clothing and non-verbal attitude (i.e., risk-accepting passenger wearing casual dark clothes vs. risk-averse passenger wearing a polo shirt tucked in his pants). Table 1 presents the instructions given to the confederate in both Passenger Risk Group conditions. To minimize variability, the confederate was instructed not to look at or speak to the driver again after saying hello and entering the vehicle. Drivers were also instructed not to talk to the passenger. Passengers were introduced to participants with a few lines explaining why they were interested in the study. The risk-accepting passenger was introduced as someone interested in participating in a future study examining the driving performance of students who applied for an emergency vehicle operator license; the risk-averse passenger was introduced as someone interested in participating in a future study looking at the driving performance of mathematics honors students. Before leaving the laboratory, participants completed questionnaires and then received their compensation. Participants were explained the goals of the study and the role of the passenger in a letter sent at the end of the study.

Table 1

Instructions for the confederate by Passenger Risk Group (accepting, averse)

2.7 Analyses

To test the two main hypotheses, main effects of repeated measure analyses of variance were examined for each of the dependent variables. For the exploratory goal, partial correlations, controlling for outcomes with no passengers, were used to examine the relationship between driving outcomes significantly affected by passenger presence and drivers’ individual characteristics (i.e., risk perception, sensation seeking, self-esteem, tolerance of deviance, and susceptibility to peer pressure). Pearson's correlations between outcomes are provided for descriptive purposes. As results of non-parametric and parametric tests were similar for all analyses, only the results of parametric tests are reported. Success of social norm manipulation was examined with a t-test.

3. Results

3.1 General description of the sample

A total of 36 male drivers with a provisional license were recruited for the study. Table 2 presents means and standard deviations (or percentages) for age, driving experience and exposure, and other individual characteristics by Passenger Risk Group. Table 3 presents Pearson's correlations between driving simulation variables, with and without passengers. The results indicate that measures of risky driving practices (e.g., higher speed and shorter headway and gap acceptance) were significantly correlated with each other. Eye glances at hazards with and without passengers were significantly correlated, as were measures of horizontal eye movement. There were no relationships, however, between measures of eye glances at hazards and measures of horizontal eye movement. Apart from one exception (for higher speed and less horizontal eye movement when driving alone), measures of risky driving practices and inattention were not significantly correlated.

Table 2

Means and standard deviations (or percentages) of individual characteristics by Passenger Risk Group (accepting, averse) (n = 36)

Open in a separate window Note. Scale ranges from: ^a1 “never” to 5 “always”; ^b1 “lowest risk” to 5 “highest risk”; ^c1 “strongly disagree” to 5 “strongly agree; ^d1 “very wrong” to 4 “not wrong”; ^e0 “no”, 1 “probably not”, 2 “probably”, and 3 “yes”.

Table 3

Pearson's correlations between driving simulation variables with passenger and no passengers (n = 36). Click here to view table.

3.2 Effects of Passenger Presence and Passenger Risk Group

Table 4 shows means and standard deviations of risky driving practices and inattention by Passenger Presence (passenger vs. no passengers) and Passenger Risk Group (accepting vs. averse). A significant main effect of Passenger Presence was found for gap acceptance with the passenger condition associated with waiting for a greater number of vehicles before initiating a left turn in a steady stream of vehicles than the no passenger condition (F(1, 32) = 5.71, p = 0.023, η² = 0.15). Results also showed a main effect of Passenger Presence on eye glances at hazards; participants glanced less often in the passenger than in the no passenger condition (F(1, 32) = 4.82, p = 0.035, η² = 0.13). A trend was also found for participants displaying less horizontal eye movement with the passenger than when driving alone (F(1, 32) = 3.76, p = 0.061, η² = 0.11). There were no effects of Passenger Presence for the percentage above the speed limit and headway.

Table 4

Means and standard deviations of risky driving practices and inattention by Passenger Presence and Passenger Risk Group (accepting, averse) (n = 36)

Notes: ^†p < 0.10;^*p < 0.05. ^aMean number of vehicles before initiating left turns. ^b Spread of horizontal eye movement in pixels.

The main effects of the Passenger Risk Group were found for headway and eye glances at hazards. Participants in the Passenger Risk Group [accepting] had longer headway (F(1, 32) = 4.40, p = 0.044, η² = 0.12) and more eye glances at hazards (F(1, 32) = 5.96, p = 0.020, η² = 0.16) than those in the Passenger Risk Group [averse]. No effects were found for the percentage above the speed limit, gap acceptance, and horizontal eye movement. While results of Passenger Risk Group were contrary to hypothesis, the success of the social norm manipulation was confirmed as participants who drove with the risk-accepting passenger reported that they found him to be more risk oriented than those who drove with the risk-averse passenger (M = 2.76, SD = 0.63 vs. M = 2.32, SD = 0.45; t₍₃₄₎ = 2.42, p = 0.011). Further analysis (not shown) indicated that the best predictors of the participants’ perception of passenger risk acceptance, explaining 52% of the variance, were self-reported past risky driving behavior and passenger random assignment (risk accepting or averse).

3.3 Relationship between drivers’ individual characteristics and driving performance

An exploratory goal of the study was to examine the relationship between drivers’ individual characteristics, which include risk perception, sensation seeking, self-esteem, tolerance of deviance, susceptibility to peer pressure (see Table 2), and measures of driving performance significantly affected by passenger presence (i.e., headway, gap acceptance, eye glances at hazards). Results of partial correlations, controlling for outcomes while driving alone, indicated that earlier initiation of a left turn in a steady stream of oncoming vehicles was significantly correlated with higher scores for tolerance of deviance (pr = −0.38, p = 0.026) and susceptibility to peer pressure (pr = −0.38, p = 0.024). Lower eye glances at hazards was significantly correlated with lower self-esteem (pr = 0.37, p = 0.03). No significant results were found for headway.

4. Discussion

In contrast to previous studies focusing on peer pressure, the current study aimed to better understand if perceived social norms can trigger risky behavior and inattention in the presence of a passenger. This is important because social norms appear to be a common mode of peer influence leading to increased risky driving (Ouimet, 2009). This study also attempted to reduce to a minimum the potentially distracting effect of passengers. The mode of activation of peer influence was based on the passenger's non-verbal portrayal, before data collection, of social norms regarding risk. The passenger was instructed to behave in a specific way to indicate his attitude about risk acceptance, but not to talk or provide non-verbal cues during data collection.

4.1 Effects of Passenger Presence and Passenger Risk Group

One of the first goals of the study was to examine if passenger presence increases inattention and risky driving. Results showed that attention decreased in the presence of the passenger, but risky driving was not affected. The former finding is consistent with previous results from experiments in which drivers interacted verbally with the passenger (Gugerty et al., 2004; Toxopeus et al., 2011; White and Caird, 2010). However, the effect of passenger presence on inattention was not found in other studies when the passenger was silent (Rivardo et al., 2008; Toxopeus et al., 2011) and could not see the drivers’ performance (Rivardo et al., 2008). The results of the current study indicate that the presence of a “not-neutral” passenger had an impact on teenage driving performance. The nature of the effect of the passenger, which could be due to different factors including increased arousal or cognitive load (Baron and Kenny, 1986; Sanders and Baron, 1975; Zajonc, 1965), should be examined in future studies as previous studies have shown that increased cognitive load was associated with less road scanning (Sodhi et al., 2002).

In contrast to measures of inattention, some measures of risky driving, such as speed and headway, were not affected by passenger presence; a finding similar to that reported by Drews et al. (2008). Higher speed, however, was found in other studies using a more direct mode of peer influence and a higher intensity of influence (Oei and Kerschbaumer, 1990; Shepherd et al., 2011). In the current study, the measurement of speed when following a lead vehicle, as opposed to free speed, most certainly affected the speed and correlations between speed and headway, which could explain the discrepant results between studies.

Finally, participants took more time before initiating a left turn when with a passenger than when alone. To our knowledge, this is the first experimental study to measure gap acceptance in a driving simulator with and without a passenger. With fixed-based simulation, breaking and turning are often reported by participants as the main elements differing from real driving conditions. This situation, and the results reported in an observational study (Ebbesen and Haney, 1973), where drivers who had to wait before turning left took more risks when alone than when with passengers might reflect a concern by participants to not expose the passenger to a potentially negative outcome. It might also suggest that principles of social facilitation models (Zajonc, 1965) apply here. Gap acceptance tasks may have been considered by participants to be complex and unfamiliar. The task coupled with the possible increased arousal associated with passenger presence could have increased workload and contributed to the slower execution (Baron and Kenny, 1986; Sanders and Baron, 1975). It is also possible that inattention is the first driving feature impacted by passenger presence and that risky driving practices are only affected when stronger modes of peer influence or increased arousal occur, or when potential distractions are more prominent.

Another goal of the study was to examine if risky driving and inattention vary according to the Passenger Risk Group condition. Comparisons between the risk-accepting and risk-averse passenger conditions were not significantly different for speed, gap acceptance, and horizontal eye movement. Moreover, while increased risky driving practices and inattention were expected with the risk-accepting passenger, in fact, participants with the risk-accepting passenger maintained longer headway with the lead vehicle and glanced at more hazards than participants with the risk-averse passenger. One plausible explanation for these unexpected results is that some features of the risk-accepting and risk-averse passengers were perceived unpredictably by participants. The success of the social norm manipulation was supported as participants who drove with the risk-accepting passenger reported that they found him to be more risk oriented compared to those who drove with the risk-averse passenger. It is possible, however, that the risk-averse passenger, who was introduced as a mathematics honor student and instructed to be more tentative and withdrawn in his attitude than the risk-accepting passenger, was also perceived as less accepting of the participant than the risk-accepting passenger. Future studies are needed to clarify if other characteristics, including perceived social rank (Cohen and Prinstein, 2006) and feelings of acceptance, might also play a role or interact with perceived social norms. While the results were contrary to our expectations, this study shows that very small changes in the characteristics of the passenger affected driver behavior.

4.2 Relationship between drivers’ individual characteristics and driving performance

An exploratory goal of the study was to examine whether there was a relationship between driver individual characteristics and measures of driving performance with a passenger. For risky driving practices, the results showed significant relationships between earlier initiation of a left turn in a steady stream of oncoming vehicles and higher tolerance of deviance susceptibility to peer pressure. A relationship between self-report risky practices while driving alone and tolerance of deviance and susceptibility to peer pressure has been demonstrated in the past (Shope et al., 2003). For inattention, the results indicated a significant association between higher glances at hazards and higher self-esteem. In a review of the literature on social facilitation models, better performance in the presence of others was found for people with higher self-esteem than for those with lower self-esteem (Uziel, 2007). It is possible that lower self-esteem in combination with passenger presence might increase participants’ arousal and workload, leading to lower performance. The respective mediating effect of tolerance of deviance, susceptibility to peer pressure, and self-esteem on risky driving practices and inattention in the presence of passengers warrants more detailed investigation.

4.3 Strengths

This randomized controlled experiment has several strengths including a design allowing causal inferences, the use of a subtler mode of peer influence than in previous studies (e.g., Shepherd et al., 2011), and the study of peer influence in a dynamic social context (e.g., Cohen and Prinstein, 2006; Gardner and Steinberg, 2005) using driving simulation rather than self-reported and retrospective measures. Several experimental studies have demonstrated the effect of peer influence on risk behavior in other research areas (Bauman and Fisher, 1986; Borsari and Carey, 2001, for a review; Cohen and Prinstein, 2006; Gardner and Steinberg, 2005). With few exceptions (e.g., Gugerty et al., 2004; Shepherd et al., 2011), the evidence on the effect of peer passengers on increased risk comes mainly from national databases (e.g., Chen et al., 2000; Ouimet et al., 2010) and field observations (e.g., Simons-Morton et al., 2005; Williams et al., 2003). These two methods are suitable for understanding the scope of the problem, but inappropriate for understanding the causes underlying behavior. Fortunately, the availability of driving simulators allows for the experimental randomized manipulation of driving conditions, including passenger presence and characteristics, and the careful measurement of driving behavior, without actual risk (Ouimet et al., 2011). The external validity of driving simulators has been demonstrated in a number of studies (e.g., Blaauw, 1982; Fisher et al., 2007; Riemersma et al., 1990).

4.4 Limitations

Participants in this study were volunteers and, by definition, they might not be representative of the population of teenage drivers. Also, results might not be generalizable to females or older male teenagers and young adults. The sample size was small, but it is comparable to those of studies using a stronger mode of peer influence or discussions between drivers and passengers or between passengers (e.g., Gugerty et al., 2004; Shepherd et al., 2011). This study did not examine the effects of multiple passengers, which are associated with a higher fatal crash risk than one passenger (Chen et al., 2000), nor the influence of multiple peers at the same time or as a reciprocal and transactional process (Brown et al., 2008). While teenagers are more likely to drive with friends in real life than with strangers, confederates unfamiliar to participants are often used in laboratory experiments in order to increase internal validity and decrease sources of confound. While many studies with confederates have shown the negative direct effect of peer pressure on behavior, friends in a vehicle might influence behavior in a more subtle way than by direct peer pressure (Ouimet, 2009). The effect of perceived norms conveyed by a friend has been demonstrated in other situations (Chein et al., 2011). This study examining the effect of a subtle mode of peer influence was a carefully designed experiment to limit the effects of confounds. While the results suggest that a subtle mode of influence from a stranger can affect some aspects of teenage driving, the attempt to manipulate perceived norms might have been too subtle to elicit risky behavior in the presence of a confederate; it is possible that some other unmeasured aspects linked to the manipulation might have played a role. Inspired by the results obtained in the current study, another study examining the effect of more salient norms was designed and the results showed effects of passenger type in the expected direction (Bingham et al., submitted for publication).

5. Conclusions

This study showed that passenger presence had a detrimental effect on measures of driving attention. Contrary to hypothesis, risk was actually lower in the presence of a risk-accepting passenger than in the presence of a risk-averse passenger. The results of this study suggest that relatively small differences in the mainly non-verbal behavior and the appearance of the passenger had an impact on teenage driver behavior. The impact on teenage driver behavior of the presence of a non-interacting teenage passenger and driver characteristics need to be further considered.

Link to Article

1. Introduction

Adolescents exhibit a disproportionately high involvement in risky behaviors (Arnett, 1996; Jonah, 1990), with peer presence and influence often implicated as contributing factors. This association is evident in behaviors like smoking (Simons-Morton & Farhat, 2010) and alcohol consumption (Borsari & Carey, 2001), which are more prevalent among adolescents and young adults when their peers engage in these activities. Furthermore, the presence of teenage passengers is linked to an increased incidence of fatal crashes involving teenage drivers (Chen et al., 2000; Ouimet et al., 2010). Given that traffic crashes represent the leading cause of morbidity and mortality among adolescents (World Health Organization, 2009), and are significantly impacted by negative peer influence, understanding the dynamics between teenage drivers and their passengers is crucial for developing effective traffic injury prevention strategies.

Risky driving and inattention, often stemming from distractions, are primary factors contributing to crashes among young drivers (Curry et al., 2012; McKnight & McKnight, 2003). While peer presence is generally associated with risky behavior, its impact on inattention appears particularly relevant to driving. Observational and experimental studies consistently demonstrate that teenage drivers with teenage passengers engage in riskier driving behaviors, including speeding (Shepherd et al., 2011; Simons-Morton et al., 2005; Toxopeus et al., 2011), reduced headway (Simons-Morton et al., 2005), and lower seatbelt usage (Williams et al., 2003). These passengers also contribute to inattention, as evidenced by increased "looked-but-failed-to-see" errors (White & Caird, 2010), delayed identification and reaction times to hazards (Gugerty et al., 2004), and an overall higher frequency of driving errors (Rivardo et al., 2008; Toxopeus et al., 2011).

Interestingly, some epidemiological and observational studies highlight a more pronounced negative effect when both the driver and passenger are male (Chen et al., 2000; Ouimet et al., 2010; Simons-Morton et al., 2005), while other studies suggest a beneficial or negligible effect of passengers. For instance, Simons-Morton et al. (2011) observed safer driving practices among teenage drivers accompanied by peer passengers in a naturalistic study, and Toxopeus et al. (2011) reported faster reaction times during simulated driving with passengers. Similarly, some experimental studies focusing on distraction found no significant difference in performance between driving with a passenger and driving alone (Drews et al., 2008), or observed performance decline only when the passenger engaged in conversation (Rivardo et al., 2008; Toxopeus et al., 2011).

These conflicting findings underscore the complexity of peer passenger influence on teenage driving risk, suggesting the presence of specific conditions under which risk might increase or decrease. Furthermore, the interplay between peer influence and distraction in this context warrants further investigation. Most experimental studies have explored either the impact of peer influence on risky driving practices (Oei & Kerschbaumer, 1990; Shepherd et al., 2011) or the role of distraction in driving inattention (Gugerty et al., 2004; Rivardo et al., 2008; Toxopeus et al., 2011; White & Caird, 2010). However, these factors could operate concurrently. For example, studies examining the impact of distraction on driving performance often involve confederates engaging participants in conversation (Gugerty et al., 2004; Rivardo et al., 2008; Toxopeus et al., 2011; White & Caird, 2010), potentially introducing the influence of perceived social norms conveyed by the passenger. Conversely, studies exploring peer pressure by having confederates encourage risky behavior (Shepherd et al., 2011) might also inadvertently introduce an element of distraction. Therefore, it is crucial to investigate the combined influence of peer pressure and distraction within the same research design.

Beyond direct peer pressure, other modes of peer influence might be at play. Peers can influence behavior through direct encouragement or discouragement, modeling, conveying attitudes and expectations, and shaping perceptions of social norms (Brown et al., 2008). The relative importance of these modes can vary depending on the behavior in question. While the indirect influence of peers seems more crucial in smoking initiation (Nichter et al., 1997), both direct and indirect influences have been documented in other studies (Simons-Morton & Farhat, 2010). In the context of driving, Ouimet (2009) found that 70% of participants in a pilot study reported that the mere presence of a silent, non-interacting teenage passenger influenced their driving, compared to 25% who reported being influenced by direct peer pressure. This finding contrasts with experimental studies demonstrating no effect of a silent passenger (Rivardo et al., 2008; Toxopeus et al., 2011), but significant effects of verbal peer pressure from a confederate prior to driving (Oei & Kerschbaumer, 1990). If perceived social norms play a more significant role than direct pressure in influencing risky driving, further research is needed to understand the impact of subtle peer influence cues, such as those conveyed by a silent and passive passenger.

Theoretical frameworks like peer influence and social facilitation models offer potential explanations for the observed variations in risky driving and inattention in the presence of passengers. Brown et al.'s (2008) conceptual model of peer influence suggests that an event, such as the presence of a teenage passenger, can activate peer influence, prompting the driver to accept, reject, or counter this influence. The chosen response can then affect driving behavior. Similarly, social facilitation models posit that the presence of others can enhance performance on simple, familiar tasks, but impair performance on complex, unfamiliar tasks (Zajonc, 1965). This effect is attributed to heightened arousal (Guerin, 1986; Zajonc, 1965), potentially stemming from evaluation apprehension (Cottrell et al., 1968) or increased cognitive load due to divided attention (Baron & Kenny, 1986; Sanders & Baron, 1975). Additionally, the presence of others can influence behavior through a desire to conform to perceived social norms (Carver & Scheier, 1981). Applied to driving, these models suggest that the presence of a passenger considered meaningful by the driver could affect risky behavior and attention depending on the perceived complexity of the driving task.

While insightful, these models often overlook the role of individual factors (Brown et al., 2008; Uziel, 2007). A systematic review by Uziel (2007) revealed that individual characteristics had a greater moderating effect on performance in the presence of others than task complexity. Individuals with positive self-perceptions (e.g., high self-esteem) tended to perform better, while those with negative self-perceptions (e.g., low self-esteem) showed performance impairment. Similarly, individual factors in both drivers and passengers can influence their response to peer influence (Brown et al., 2008). Factors such as male sex (Cooper et al., 1979; Simons-Morton et al., 2005), high sensation seeking (Slater, 2003), openness to peer influence (Brown et al., 2008), passenger characteristics like sociability (Collins et al., 1985), and having friends involved in risky behavior (Shope et al., 2003; Simons-Morton et al., 2011) can all play a role.

In conclusion, a nuanced understanding of how peers, particularly through subtle cues and individual characteristics, affect teenage driver behavior is critical for developing effective traffic safety interventions. Given the observed stronger effects when both drivers and passengers are male (Chen et al., 2000; Ouimet et al., 2010; Simons-Morton et al., 2005), this study aimed to investigate: (i) the effect of a male teenage passenger's presence on male teenage driving behavior; and (ii) the influence of passenger characteristics intended to manipulate the driver's perceived social norms (i.e., risk-accepting vs. risk-averse passenger). We hypothesized that teenagers would exhibit riskier driving practices and decreased attention when driving with a passenger compared to driving alone, and that this effect would be more pronounced with a risk-accepting passenger compared to a risk-averse passenger. Additionally, we explored the relationship between driver individual characteristics and driving performance with a passenger.

2. Methods

2.1 Participants

Male adolescents aged 16-17 years old, holding a provisional driving license and having driven in the past three months, were recruited from high schools and driving schools in Amherst, Massachusetts. Participants were required to have parental consent and provide assent. Exclusion criteria included frequent motion sickness and uncorrected vision. A total of 36 participants were compensated $25.00 USD for their participation in the 75-minute study. The University of Massachusetts Amherst Institutional Review Board approved the study protocol.

2.2 Study Design, Random Assignment, and Concealment

A 2 x 2 mixed design was employed, with Passenger Risk Group (accepting vs. averse) as a between-subjects factor and Passenger Presence (passenger vs. no passenger) as a within-subjects factor. Participants completed two drives: one alone and one with a randomly assigned risk-accepting or risk-averse passenger. Restricted randomization, conducted by an investigator not involved in testing, was used to assign participants to Passenger Risk Group and driving order (alone first vs. with passenger first) to control for potential learning effects. Both participants and data analysts were blinded to passenger type, which was revealed to research assistants only upon the confederate's entry into the testing room.

2.3 Driving Simulation and Equipment

Two drive sets, each comprising three continuous drives simulating rural, suburban, and urban environments, were utilized. The drives included a lead vehicle. Risky driving practices and inattention were measured using a Realtime Technologies fixed-base driving simulator and an ASL Mobile Eye eye tracker, respectively. The simulator, resembling a 1995 Saturn sedan, responded to simulated roadways displayed on three screens, providing a 135-degree horizontal field of view. The eye tracker, using cameras and a reflector attached to the participant's safety glasses, monitored eye movements and superimposed a crosshair indicating the driver's fixation point on a video recording.

2.4 Dependent Variables

Risky driving was assessed using percentage of time exceeding the speed limit, headway distance, and gap acceptance. Inattention was measured using eye glances at hazards and horizontal eye movement. Percentage above the speed limit was measured for 15-30 second intervals at designated sections of the rural drive with posted speed limits of 30 or 45 mph, ensuring the lead vehicle was not visible to participants during measurement. Headway distance, defined as the average distance between the participant's vehicle and the lead vehicle, was measured for 20-25 seconds in a specific section of the rural drive. Gap acceptance, measured at four signalized intersections in the urban and suburban drives, involved recording the number of oncoming vehicles the participant allowed to pass before making a left turn in a steady stream of 10 vehicles. Eye glances at hazards were measured in the rural drives, which included five scenarios with potentially obscured hazards (e.g., a pedestrian hidden by a large vehicle). The average number of glances towards predetermined target zones representing these hazards was calculated. Horizontal eye movement, representing the spread of horizontal scanning, was measured in pixels at various points in the rural drive when no other variables were being collected.

2.5 Questionnaires

Questionnaires assessed demographic information, driving exposure (miles driven in the past week, time since licensing, experience with peer passengers, citations, and crashes), past risky driving behavior, crash/injury risk perception, sensation seeking, self-esteem, tolerance of deviance, and susceptibility to peer pressure. Past risky driving behavior was measured using a 19-item questionnaire (Simons-Morton et al., 2006b) with a 5-point Likert scale (1 = "never" to 5 = "always"; Cronbach's α > 0.88). Crash/injury risk perception for newly licensed teenagers was assessed using a 14-item questionnaire (Simons-Morton et al., 2006a) with a 5-point Likert scale (1 = "lowest risk" to 5 = "highest risk"; Cronbach's α > 0.88). The Brief Sensation Seeking Scale (Hoyle et al., 2002; Stephenson et al., 2003), consisting of eight items derived from the Sensation Seeking Scale Form V (Zuckerman, 1994), used a 5-point Likert scale ("strongly disagree" to "strongly agree"; Cronbach's α = 0.76). Self-esteem was measured using the 10-item Rosenberg Self-Esteem Scale (Rosenberg, 1989) with a 5-point Likert scale (1 = "strongly disagree" to 5 = "strongly agree"; Cronbach's α > 0.80) (Schmitt & Allik, 2005). Tolerance of deviance was assessed using a 5-item measure with a 4-point Likert scale (from "very wrong" to "not wrong"; Cronbach's α > 0.80) (Rachal et al., 1975; Shope et al., 2003). Susceptibility to peer pressure was measured using a 7-item measure with a 4-point response scale ("no," "probably not," "probably," and "yes"; Cronbach's α > 0.85) (Dielman et al., 1987; Shope et al., 2003). The effectiveness of the social norm manipulation was evaluated using 10 questions inspired by Parker et al. (1992), assessing participants' perceptions of the passenger's descriptive and injunctive norms related to risk-taking behaviors (Cronbach's α = 0.84).

2.6 Procedure

Participants completed a 5-minute practice drive followed by two experimental drives: one without a passenger and one with a passenger (randomized order). The passenger, a male confederate blind to participant assignment and driving order, adopted either a risk-accepting or risk-averse persona, differing only in clothing and nonverbal behavior (casual dark clothing for risk-accepting vs. collared shirt tucked into pants for risk-averse). Table 1 outlines the instructions provided to the confederate for each Passenger Risk Group condition. To minimize variability, the confederate, after greeting the participant upon entering the vehicle, refrained from interacting with them. Participants were also instructed to avoid interacting with the confederate. Prior to each drive, the passenger was introduced to participants with a brief explanation of their purported interest in the study (emergency vehicle operator license for risk-accepting, mathematics honors student for risk-averse). After completing both drives, participants completed questionnaires and received compensation. A debriefing letter explaining the study's objectives and the passenger's role was sent to participants afterward.

2.7 Analyses

Repeated measures analyses of variance (ANOVAs) were used to examine main effects of Passenger Presence on each dependent variable. Partial correlations, controlling for outcomes without passengers, were used to explore the relationship between driving outcomes significantly affected by passenger presence and driver individual characteristics (risk perception, sensation seeking, self-esteem, tolerance of deviance, and susceptibility to peer pressure). Pearson's correlations between outcomes are reported for descriptive purposes. As results of parametric and non-parametric tests were similar, only parametric test results are presented. A t-test was used to assess the success of the social norm manipulation.

3. Results

3.1 General Description of the Sample

Thirty-six male drivers with a provisional license participated in the study. Table 2 presents descriptive statistics for age, driving experience and exposure, and individual characteristics by Passenger Risk Group. Table 3 displays Pearson's correlations between driving simulation variables with and without passengers. Results indicate significant correlations between risky driving practices (e.g., higher speed, shorter headway and gap acceptance). Eye glances at hazards with and without passengers were significantly correlated, as were measures of horizontal eye movement. However, no significant correlations were found between measures of eye glances at hazards and horizontal eye movement. Except for one instance (higher speed and less horizontal eye movement when driving alone), no significant correlations were observed between risky driving practices and inattention.

3.2 Effects of Passenger Presence and Passenger Risk Group

Table 4 presents means and standard deviations of risky driving practices and inattention by Passenger Presence and Passenger Risk Group. A significant main effect of Passenger Presence was observed for gap acceptance, F(1, 32) = 5.71, p = 0.023, η**2 = 0.15, with participants waiting for more vehicles before turning left when a passenger was present. A significant main effect of Passenger Presence was also found for eye glances at hazards, F(1, 32) = 4.82, p = 0.035, η**2 = 0.13, with fewer glances observed when a passenger was present. A trend towards less horizontal eye movement in the passenger condition was also noted, F(1, 32) = 3.76, p = 0.061, η**2 = 0.11. No significant effects of Passenger Presence were found for percentage above the speed limit and headway.

Main effects of Passenger Risk Group were found for headway, F(1, 32) = 4.40, p = 0.044, η**2 = 0.12, and eye glances at hazards, F(1, 32) = 5.96, p = 0.020, η**2 = 0.16. Participants in the risk-accepting passenger condition maintained longer headway distances and made more eye glances at hazards compared to those in the risk-averse passenger condition. No significant effects of Passenger Risk Group were observed for percentage above the speed limit, gap acceptance, and horizontal eye movement. While these results contradicted our hypothesis, the manipulation of social norms was successful, as evidenced by participants in the risk-accepting condition perceiving their passenger as significantly more risk-oriented than those in the risk-averse condition, t(34) = 2.42, p = 0.011. Further analysis revealed that self-reported past risky driving behavior and passenger random assignment (risk-accepting or risk-averse) were the strongest predictors of perceived passenger risk acceptance, explaining 52% of the variance.

3.3 Relationship Between Drivers' Individual Characteristics and Driving Performance

Exploratory analyses using partial correlations, controlling for outcomes while driving alone, were conducted to examine the relationship between driver characteristics and driving performance measures significantly affected by passenger presence. Earlier left turn initiation in the presence of oncoming traffic was significantly correlated with higher scores on tolerance of deviance, pr = -0.38, p = 0.026, and susceptibility to peer pressure, pr = -0.38, p = 0.024. Fewer eye glances at hazards were significantly correlated with lower self-esteem, pr = 0.37, p = 0.03. No significant relationships were found for headway.

4. Discussion

This study sought to understand whether perceived social norms, rather than direct peer pressure, could influence risky driving and inattention in the presence of a passenger, particularly as social norms appear to be a significant factor contributing to increased risky driving (Ouimet, 2009). To minimize potential distractions from the passenger, the study employed subtle manipulations of perceived social norms through the passenger's nonverbal behavior and appearance, while minimizing verbal interaction.

4.1 Effects of Passenger Presence and Passenger Risk Group

Our findings partially supported the hypothesis that passenger presence increases inattention. Consistent with previous research involving verbal interaction between drivers and passengers (Gugerty et al., 2004; Toxopeus et al., 2011; White & Caird, 2010), we found that attention decreased in the presence of a passenger. However, this finding contrasts with studies where silent passengers had no effect on inattention (Rivardo et al., 2008; Toxopeus et al., 2011), particularly when the passenger could not observe the driver's performance (Rivardo et al., 2008). This discrepancy suggests that the "non-neutral" nature of the passenger in our study, even without direct interaction, may have impacted teenage driving performance. Further research is needed to explore the specific mechanisms underlying this effect, such as increased arousal or cognitive load (Baron & Kenny, 1986; Sanders & Baron, 1975; Zajonc, 1965), particularly as previous studies have linked increased cognitive load to reduced road scanning (Sodhi et al., 2002).

Contrary to our hypothesis, risky driving measures like speed and headway were not significantly affected by passenger presence. This finding aligns with Drews et al. (2008) but contrasts with studies employing more direct and intense modes of peer influence, which reported increased speeding (Oei & Kerschbaumer, 1990; Shepherd et al., 2011). The use of a lead vehicle in our study, as opposed to free speed conditions, may explain this discrepancy. Furthermore, our finding that participants took longer to initiate left turns when a passenger was present suggests a potential influence of social facilitation models (Zajonc, 1965), where the perceived complexity of the gap acceptance task, coupled with increased arousal from the passenger's presence, may have led to slower responses (Baron & Kenny, 1986; Sanders & Baron, 1975). It is plausible that inattention precedes risky driving behaviors in response to passenger presence, with the latter manifesting only under more potent peer influence, heightened arousal, or more salient distractions.

Regarding the influence of passenger risk-taking propensity, no significant differences were found between the risk-accepting and risk-averse passenger conditions for speed, gap acceptance, and horizontal eye movement. Surprisingly, drivers in the risk-accepting condition exhibited safer driving behaviors, maintaining longer headway distances and making more eye glances at hazards compared to those in the risk-averse condition. While unexpected, this finding underscores the potential impact of subtle passenger characteristics on driving behavior. One possible explanation is the inadvertent manipulation of perceived social acceptance alongside risk perception. The risk-averse passenger, portrayed as a mathematics honors student instructed to appear more reserved and withdrawn, might have been perceived as less accepting of the participant compared to the more casually dressed, risk-accepting passenger. Future research should investigate the interplay between perceived social norms, social rank (Cohen & Prinstein, 2006), and feelings of acceptance in influencing driving behavior.

4.2 Relationship Between Drivers' Individual Characteristics and Driving Performance

Exploratory analyses revealed significant associations between driver characteristics and driving performance in the presence of a passenger. Earlier left turn initiation in the presence of oncoming traffic was linked to higher tolerance of deviance and susceptibility to peer pressure, mirroring previous findings on the relationship between self-reported risky driving and these traits (Shope et al., 2003). Furthermore, decreased eye glances at hazards were associated with lower self-esteem. This finding aligns with Uziel's (2007) review, which highlighted better performance in the presence of others among individuals with higher self-esteem. It is possible that lower self-esteem, coupled with the presence of a passenger, amplifies arousal and cognitive load, leading to poorer performance. Further investigation is warranted to understand the mediating roles of tolerance of deviance, susceptibility to peer pressure, and self-esteem on risky driving practices and inattention in the presence of passengers.

4.3 Strengths

This study utilized a randomized controlled experimental design, enabling causal inferences about the impact of passenger presence and perceived social norms on driving behavior. Unlike previous studies employing more overt manipulations of peer influence (e.g., Shepherd et al., 2011), this study explored the impact of subtle nonverbal cues, enhancing ecological validity. The use of a driving simulator provided a controlled environment to manipulate passenger characteristics and measure driving behavior without exposing participants to real-world risks. Driving simulators offer good external validity, as demonstrated in various studies (Blaauw, 1982; Fisher et al., 2007; Riemersma et al., 1990), making them valuable tools for studying driving behavior.

4.4 Limitations

The reliance on volunteer participants might limit the generalizability of findings to the broader population of teenage drivers. Furthermore, the study's focus on male participants aged 16-17 restricts generalizability to females and older demographics. While the sample size was comparable to other studies using similar methodologies (Gugerty et al., 2004; Shepherd et al., 2011), a larger sample size would enhance statistical power. The study did not examine the effects of multiple passengers, a known risk factor for fatal crashes (Chen et al., 2000), nor did it explore the reciprocal and transactional nature of peer influence (Brown et al., 2008). The use of confederates, while common in experimental research to control for confounding variables, might not fully capture the nuances of peer influence within existing friendships. While our findings suggest that subtle cues from a stranger can impact driving behavior, the manipulation of perceived norms might have been too subtle to elicit risky driving in this context. Future studies could benefit from exploring more salient manipulations of perceived norms. Indeed, a subsequent study by Bingham et al. (submitted for publication), inspired by our findings, employed more salient manipulations and found significant effects of passenger risk-taking propensity on driving behavior.

5. Conclusions

Our study provides evidence for the detrimental effect of passenger presence on teenage driver attention. Contrary to our hypothesis, the presence of a risk-accepting passenger, compared to a risk-averse passenger, was associated with safer driving behaviors. These findings highlight the sensitivity of teenage driving behavior to subtle variations in passenger characteristics and underscore the need for further investigation into the complex interplay between passenger presence, perceived social norms, individual characteristics, and driving performance. A nuanced understanding of these factors is crucial for developing effective interventions to mitigate the risks associated with teenage driving.

Link to Article

1. Introduction

Teenagers frequently engage in risky behaviors (Arnett, 1996; Jonah, 1990), and the presence and influence of their peers often contribute to this pattern. This is evident in behaviors like smoking (Simons-Morton and Farhat, 2010) and drinking (Borsari and Carey, 2001), where teen participation increases when friends are also involved. A significant example is driving, where teens with teen passengers are more likely to be in fatal car accidents (Chen et al., 2000; Ouimet et al., 2010). Traffic accidents are a leading cause of death and injury for this age group (World Health Organization, 2009), highlighting the serious consequences of negative peer pressure in driving. Therefore, understanding how teen drivers and passengers interact is crucial for developing effective accident prevention strategies.

Two major factors contributing to accidents among young drivers are risky driving and inattention, often caused by distractions (Curry et al., 2012; McKnight and McKnight, 2003). While peer presence is generally linked to risky behavior, its influence on inattention is particularly significant in driving. Studies have consistently shown that teen drivers with teen passengers engage in riskier driving practices, including speeding (Shepherd et al., 2011; Simons-Morton et al., 2005; Toxopeus et al., 2011), following too closely (Simons-Morton et al., 2005), and lower seat belt use (Williams et al., 2003). Additionally, they exhibit more inattention, demonstrated by an increase in "looked-but-failed-to-see" errors (White and Caird, 2010), slower reactions to hazards (Gugerty et al., 2004), and a higher number of overall driving errors (Rivardo et al., 2008; Toxopeus et al., 2011).

Interestingly, some studies suggest that passengers can have positive or neutral effects. For instance, naturalistic studies have observed safer driving practices among teens with passengers (Simons-Morton et al., 2011), and simulations have shown faster reaction times (Toxopeus et al., 2011). Some experiments focused on distraction found no difference in driving performance with or without a passenger (Drews et al., 2008). Others noted decreased performance only when passengers actively engaged in conversations (Rivardo et al., 2008; Toxopeus et al., 2011).

These conflicting findings highlight the complexity of peer influence on teen driving. It suggests that specific conditions may either increase or decrease risk. This raises the question of whether both peer influence and distraction impact driving when teen passengers are present. Many studies tend to focus on either peer pressure leading to risky driving (Oei and Kerschbaumer, 1990; Shepherd et al., 2011) or distraction leading to inattention (Gugerty et al., 2004; Rivardo et al., 2008; Toxopeus et al., 2011; White and Caird, 2010). However, it's likely that both are at play simultaneously. For example, studies using confederates to engage participants in conversations to measure distraction (Gugerty et al., 2004; Rivardo et al., 2008; Toxopeus et al., 2011; White and Caird, 2010) might also be inadvertently influencing driver behavior through social cues from the passenger. Similarly, when confederates exert direct peer pressure to engage in risky driving (Shepherd et al., 2011), they may also cause distraction. This suggests that a comprehensive approach considering both peer influence and distraction is necessary.

Furthermore, most experimental studies emphasize direct peer pressure, but there are subtler ways peers exert influence. These include modeling behavior, expressing attitudes and expectations, and shaping perceptions of social norms, all of which can either encourage or discourage certain actions. Research on smoking suggests that these indirect influences are more impactful than direct pressure (Nichter et al., 1997), and similar findings have been observed in other studies (Simons-Morton and Farhat, 2010). A pilot study on driving found that 70% of teens reported that the mere presence of a silent passenger impacted their driving, while only 25% attributed their behavior to direct peer pressure (Ouimet, 2009). This contrasts with experimental studies showing no effect of silent passengers (Rivardo et al., 2008; Toxopeus et al., 2011) but significant influence from verbal pressure exerted before driving (Oei and Kerschbaumer, 1990). This highlights the need for more research on how subtle cues, like perceived social norms conveyed by a silent passenger, contribute to risky driving and inattention.

Models like peer influence and social facilitation can help explain these varying effects. Brown et al. (2008) proposed a model where peer influence involves an event (like a passenger's presence), activation of influence, the driver's response (accepting, rejecting, or countering the influence), and a measurable outcome. Social facilitation models suggest that the presence of others can either enhance or hinder performance depending on the complexity of the task (Zajonc, 1965). Simple tasks benefit from the presence of others, while complex tasks are negatively affected (Bond and Titus, 1983). This effect is attributed to increased arousal (Guerin, 1986; Zajonc, 1965), potentially caused by evaluation apprehension (Cottrell et al., 1968) or divided attention between the task and the social situation (Baron and Kenny, 1986; Sanders and Baron, 1975). Additionally, individuals might adjust their behavior to conform to perceived norms (Carver and Scheier, 1981). In driving, this means a passenger's presence can impact risk-taking and attention depending on the driving situation and how meaningful the passenger is to the driver.

However, both models often overlook individual factors (Brown et al., 2008; Uziel, 2007). A review of social facilitation models found that personal characteristics, like self-esteem, are more influential than task complexity (Uziel, 2007), with positive traits leading to improved performance and negative traits leading to impairment. Similarly, individual factors in both the driver and passenger (e.g., gender, sensation seeking, openness to peer influence, sociability) can moderate responses to peer influence (Brown et al., 2008; Cooper et al., 1979; Simons-Morton et al., 2005; Slater, 2003; Shope et al., 2003; Simons-Morton et al., 2011). Therefore, exploring the influence of these individual characteristics is crucial.

Given the increased risk when both drivers and passengers are male (Chen et al., 2000; Ouimet et al., 2010; Simons-Morton et al., 2005), this study aims to examine how the presence of a male teen passenger and the passenger's perceived attitude towards risk (risk-accepting vs. risk-averse) impact male teen driving behavior in a driving simulator. It's hypothesized that teens will drive more riskily and be less attentive with a passenger and even more so with a risk-accepting passenger. Additionally, the study explores the relationship between driver characteristics and driving performance with a passenger.

2. Methods

2.1. Participants

Male teen drivers aged 16-17 with provisional licenses were recruited from high schools and driving schools in Amherst, Massachusetts. Participants had to have driven in the past three months, have parental consent, and provide assent. They were excluded if they experienced motion sickness or had uncorrected vision. Participants received $25.00 USD for their participation in the 75-minute study, which received ethical approval from the University of Massachusetts Amherst's Institutional Review Board.

2.2. Study Design, Random Assignment, and Concealment

A 2 x 2 mixed design was used with between-subjects factor Passenger Risk Group (accepting vs. averse) and within-subjects factor Passenger Presence (present vs. absent). Participants completed two drives: one alone and one with a randomly assigned risk-accepting or risk-averse passenger. The order of the drives was also randomized to control for learning effects. Both the participants and data analysts were blinded to the passenger type assignment.

2.3. Driving Simulation and Equipment

Two drive sets, each with three continuous drives simulating rural, suburban, and urban environments, were used. A lead vehicle was present in all drives. A Realtime Technologies fixed-base driving simulator and an ASL Mobile Eye eye tracker were used to collect data on driving behavior and eye movements.

2.4. Dependent Variables

Risky driving measures included:

• Percentage of time above the speed limit: Measured in specific sections of the rural drive with posted speed limits.

• Headway: Average distance from the participant's vehicle to the lead vehicle, measured in a specific section of the rural drive.

• Gap acceptance: Number of oncoming vehicles allowed to pass before making a left turn at four intersections in the urban and suburban drives.

Inattention measures included:

• Eye glances at hazards: Number of times participants glanced at predetermined "target zones" where potential hazards were obscured from direct view in the rural drives.

• Horizontal eye movement: The average standard deviation of the driver's horizontal gaze position, measured in pixels, indicating the breadth of their visual scanning.

2.5. Questionnaires

Questionnaires were used to collect demographic information, driving exposure (miles driven, time since licensing, experience with passengers, citations, and crashes), self-reported risky driving behaviors in the past week (19-item scale, Cronbach's alpha > 0.88), perceived crash risk (14-item scale, Cronbach's alpha > 0.88), sensation seeking (Brief Sensation Seeking Scale, 8 items, Cronbach's alpha = 0.76), self-esteem (Rosenberg Self-Esteem Scale, 10 items, Cronbach's alpha > 0.80), tolerance of deviance (5 items, Cronbach's alpha > 0.80), and susceptibility to peer pressure (7 items, Cronbach's alpha > 0.85).

To assess the success of the social norm manipulation, a 10-item questionnaire (Cronbach's alpha = 0.84) measured participants' perceptions of the passenger on descriptive and injunctive norms, including traits like being quiet, liking to party, driving conservatively, and approving of risky driving behaviors.

2.6. Procedure

After a 5-minute practice drive, participants completed two drives: one alone and one with a risk-accepting or risk-averse passenger (randomly assigned). The same male confederate played both passenger roles, differentiated only by clothing and non-verbal behavior (risk-accepting: casual dark clothes; risk-averse: collared shirt tucked into pants). The confederate was instructed to greet the participant upon entering the vehicle and remain silent and avoid eye contact afterwards.

The risk-accepting passenger was introduced as a potential participant for a study on emergency vehicle operators, while the risk-averse passenger was presented as a potential participant for a study on math honors students. After completing the drives, participants answered questionnaires and received compensation. The study's purpose and the passenger's role were fully explained to participants in a letter sent after the study.

2.7. Analyses

Repeated measures ANOVAs were used to examine the main effects of Passenger Presence and Passenger Risk Group on the dependent variables. Partial correlations, controlling for driving performance without a passenger, were used to explore the relationship between drivers' individual characteristics (risk perception, sensation seeking, self-esteem, tolerance of deviance, and susceptibility to peer pressure) and driving outcomes significantly affected by passenger presence (headway, gap acceptance, eye glances at hazards). A t-test was used to analyze the success of the social norm manipulation.

3. Results

3.1. General Description of the Sample

Thirty-six male drivers participated in the study. Risky driving measures were significantly correlated with each other, as were the inattention measures. However, there were generally no significant correlations between risky driving and inattention measures, with one exception: higher speed was correlated with less horizontal eye movement when driving alone.

3.2. Effects of Passenger Presence and Passenger Risk Group

A significant main effect of Passenger Presence was found for gap acceptance and eye glances at hazards. Participants waited for more vehicles before making left turns and glanced at hazards less often when a passenger was present. A trend towards less horizontal eye movement was also observed with a passenger.

There were no significant main effects of Passenger Presence on the percentage of time speeding or headway. However, main effects of Passenger Risk Group were found for these variables, with participants driving with a risk-accepting passenger maintaining longer headways and glancing at hazards more frequently than those with a risk-averse passenger. No effects of Passenger Risk Group were found for the other driving measures.

The social norm manipulation was successful, as participants perceived the risk-accepting passenger as more risk-oriented than the risk-averse passenger.

3.3. Relationship Between Drivers' Individual Characteristics and Driving Performance

Partial correlations showed that earlier left turn initiation (indicating riskier driving) was significantly correlated with higher tolerance of deviance and susceptibility to peer pressure. Lower self-esteem was significantly correlated with fewer eye glances at hazards in the presence of a passenger. No significant relationships were found for headway.

4. Discussion

4.1. Effects of Passenger Presence and Passenger Risk Group

This study aimed to investigate whether perceived social norms, rather than direct pressure, could influence risky driving and inattention in teen drivers. The results showed that while passenger presence increased inattention, it did not affect risky driving behaviors. This aligns with previous research on inattention (Gugerty et al., 2004; Toxopeus et al., 2011; White and Caird, 2010) but contradicts studies finding no effect of silent passengers (Rivardo et al., 2008; Toxopeus et al., 2011). This suggests that even a non-interacting passenger can influence driving, potentially due to increased arousal or cognitive load (Baron and Kenny, 1986; Sanders and Baron, 1975; Zajonc, 1965). Future studies should explore these factors in more detail.

The lack of effect on risky driving measures like speed and headway is consistent with some previous work (Drews et al., 2008) but contrasts with studies using more direct forms of peer influence (Oei and Kerschbaumer, 1990; Shepherd et al., 2011). This discrepancy could be attributed to the use of a lead vehicle in this study, potentially limiting opportunities for speeding and influencing the relationship between speed and headway. The finding that participants took longer to initiate left turns with a passenger might reflect a desire to avoid negative outcomes in front of the passenger, aligning with observational data (Ebbesen and Haney, 1973). It could also be explained by social facilitation models, where increased arousal from the passenger's presence hinders performance on complex tasks like gap acceptance (Baron and Kenny, 1986; Sanders and Baron, 1975; Zajonc, 1965).

Contrary to expectations, the risk-accepting passenger was associated with safer driving (longer headways) and increased attention to hazards compared to the risk-averse passenger. While the social norm manipulation was successful, it's possible that the risk-averse passenger's demeanor (tentative and withdrawn) was unintentionally perceived as less accepting of the driver, highlighting the potential influence of factors like social rank and acceptance (Cohen and Prinstein, 2006).

4.2. Relationship Between Drivers' Individual Characteristics and Driving Performance

Exploratory analyses revealed significant relationships between driver characteristics and driving performance. Earlier left turn initiation was associated with higher tolerance of deviance and susceptibility to peer pressure, consistent with previous findings (Shope et al., 2003). Interestingly, lower self-esteem was linked to fewer eye glances at hazards when a passenger was present, suggesting that lower self-esteem might exacerbate the negative effects of passenger presence on attention, potentially through increased arousal and cognitive load (Uziel, 2007).

4.3. Strengths

This study utilized a randomized controlled design, allowing for causal inferences about the effects of passenger presence and perceived social norms on driving. It employed a subtler manipulation of peer influence than previous studies, focusing on perceived norms rather than direct pressure, and used a driving simulator to create a dynamic social context while ensuring safety.

4.4. Limitations

The study's limitations include a small, self-selected sample, potentially limiting generalizability. Future research should include female participants and older drivers. Additionally, the study focused on single, unfamiliar passengers, while real-world driving often involves friends or multiple passengers, which might exert different influences. While the subtle manipulation of perceived norms minimized confounds, it might have been insufficiently salient to elicit risky driving behavior. Future research could explore more overt manipulations.

5. Conclusions

This study highlights that even a non-interacting passenger can negatively impact teen driver attention. Interestingly, a risk-accepting passenger did not lead to riskier driving, suggesting that the relationship between perceived norms and driving behavior is complex and potentially influenced by other social factors. Further research is needed to understand these nuances and develop effective prevention strategies tailored to the complexities of teen driving behavior.

Link to Article

Teen Passengers and Risky Driving: Is It Just Distraction or Something More?

Introduction

We all know teenagers sometimes make risky choices, and having friends around can make things riskier. Think about it: teens are more likely to smoke or drink if their friends do. Even worse, car crashes, a leading cause of death for teens, are more common when teen drivers have teen passengers. So, what's going on? Is it just distraction, or are teens trying to impress their friends?

What's going on?

Many studies show that teen passengers often lead to riskier driving: speeding, tailgating, not wearing seatbelts, and missing hazards . Some studies even point to guys being the biggest culprits.

But here's the catch: other studies show passengers can actually make teens drive better or have no effect at all . This tells us there's more to the story.

Peer Pressure and Distraction Studied

It seems like both peer pressure and distraction are at play. Peer pressure is when friends try to get a person to do something, and distraction is when something takes a person's attention away from driving. Most studies have looked at these separately, but what if they're both happening at the same time?

In a study where a passenger distracts the driver by talking, the passenger might also be accidentally sending messages about what's "cool" (even if they don't mean to). That's the peer pressure part. This means we need to figure out how much of the risky driving is from peer pressure and how much is from distraction.

Here's another thing: there are different types of peer pressure. Friends can pressure a person directly by telling them to do something, or indirectly by just acting a certain way. In fact, one study found that teens said just having a friend in the car, even if they weren't talking, affected their driving more than direct pressure. This means we need to understand the ways friends can influence each other, even without saying a word.

Conclusion

So, how can we make sense of all this? There are some theories that might help. One is the idea of social facilitation, which basically says that having people around makes a person do better at easy things and worse at hard things. So, a passenger could make a teen drive better in easy situations and worse in difficult ones.

Here's the bottom line: teen driving is complicated. To make it safer, we need to look at both peer pressure and distraction, the different ways friends influence each other, and even how comfortable a teen is with driving. This study focused on figuring out if the type of passenger a teen has (one who seems to like risk vs. one who seems cautious) affects their driving. Since guys seem to be most affected by passengers, we only looked at teen guys driving with other teen guys. We wanted to see if they drove differently depending on whether they thought their passenger was a "risk-taker" or a "safe driver."

Link to Article

Driving with Friends: Are Teenagers Safer When Driving Alone?

Introduction

Teenagers sometimes make risky choices, even while driving. Having friends in the car can make things even riskier. It's a big problem because car accidents are one of the main dangers for teenagers.

Scientists wanted to understand why teenagers take more risks when driving with friends. Do friends pressure them to drive dangerously, or are they simply distracted by their friends? To find out, they did a study using a driving simulator, kind of like a video game but for driving.

The Experiment

The scientists invited teenage boys with driver's licenses to try out the driving simulator. They wanted to see how the teenagers drove when they were alone and when they had a passenger. But here's the interesting part: they used an actor to pretend to be a passenger!

The actor played two different roles: a "risk-taking" passenger who dressed casually and acted cool, and a "risk-averse" passenger who dressed neatly and acted more serious. The actor didn't talk to the drivers during the experiment; they just sat in the passenger seat.

The Results

So, what happened? The scientists found that teenagers were actually less attentive when they had a passenger, even though the passenger wasn't talking! They didn't look at hazards as often, which means they might not see something dangerous on the road.

Surprisingly, teenagers drove more carefully when the passenger seemed risk-taking. They kept a safer distance from the car in front of them. The scientists think this might be because the teenagers didn't want to seem scared in front of the "cool" passenger.

The study also looked at how different personality types affected driving. For example, teenagers who thought breaking rules wasn't a big deal were more likely to take risks when turning left. This means that a teenager's personality can play a big role in how they drive with friends.

Conclusion

This study showed us that even having a quiet passenger can make teenagers less attentive while driving. More research is needed to understand exactly how our friends influence our driving and how to stay safe on the road.

Link to Article