A field operational trial evaluating a feedback–reward system on speeding and tailgating behaviors

https://doi.org/10.1016/j.trf.2014.09.002Get rights and content

Highlights

  • Real-time feedback combined with rewards was evaluated in a field trial.

  • Two driver groups were identified based on baseline data: driving styles A and B.

  • Drivers in A were less speed and headway compliant during the baseline than B.

  • The intervention closed the gap between the two groups.

Abstract

Objective: This paper investigates the effect of a feedback–reward system on speeding and tailgating behaviors. Background: Inappropriate speed choice and headway increase crash risks. A wide range of countermeasures such as law enforcement and educational messages are in effect to limit speeding and tailgating behaviors. However, these countermeasures are not tailored to the behavior of the individual, and may be absent in some situations. Emerging technology can circumvent the limits of current countermeasures by providing real-time feedback to alert the drivers to these inappropriate behaviors. Further, incentives can be utilized to increase drivers’ motivation to adapt their driving behavior according to feedback. Previous research provides encouraging findings for both real-time feedback and incentives. Method: A field trial was conducted with 37 participants with the system installed in participants’ own vehicles. Real-time in-vehicle feedback was provided based on speed limit compliance and safe headway maintenance. Participants also accumulated reward points and could view related information on a secured website. The trial consisted of three phases: baseline (two weeks), intervention with the system (twelve weeks), and post-intervention without the system (two weeks). Results: A cluster analysis based on baseline data revealed two driving styles: more speed and headway compliant (Cluster A), less speed and headway compliant (Cluster B). Overall, the intervention closed the gap between the two groups. For speed limit compliance, both groups became more compliant in the intervention phase with a slight decline in the post-intervention phase, which was still better than the baseline. For headway compliance, Cluster A did not have any changes throughout the trial, whereas for Cluster B, the headway compliance rate increased with the intervention and then decreased slightly in the post-intervention phase, which again was still better than the baseline. Conclusion: The results are promising, in particular as the group of drivers who were less speed and headway compliant during the baseline and needed the intervention more, benefited more from it.

Introduction

Traffic crashes result in approximately 1.2 million deaths every year (Peden et al., 2004). Human error is estimated to be the sole cause in 57% of all traffic crashes and a contributing factor in over 90% of them (Treat et al., 1979). Inappropriate speed choice and headway are two human errors which increase crash risks (Kloeden et al., 1997, Kloeden et al., 2001, Knipling et al., 1993, Taylor et al., 2000). For example, the risk of involvement in a fatal crash doubles with each 5 km/h increase in traveling speed above a 60 km/h speed limit (Kloeden et al., 1997). In the U.S., speeding played a role in about 31% of all fatal crashes in 2010 (NHTSA, 2012), while it was listed as a contributing factor in approximately 25% of fatalities and 20% of injuries in Canada (Transport Canada., 2008). As for inappropriate headway, it plays a significant role when coupled with inattention, particularly in rear-end crashes, which constitute approximately 30% of all U.S. crashes (Singh, 2003). Knipling et al. (1993) found that headway that is too short to react appropriately to a lead vehicle’s sudden braking was the primary cause in 7% and a contributing factor in 19% of the rear-end crashes they examined.

Speeding and tailgating are complex behaviors and drivers exhibit them for a variety of reasons. Drivers may speed because they feel time pressure, because they want to reach their destination fast, due to fast flow of traffic, or due to emotional reasons such as anger or seeking thrill (Åberg et al., 1997, Paris and Broucke, 2008, Tarko, 2009). Drivers may also speed under the influence of other people, including role models, other drivers, and passengers who speed or have a favorable attitude towards speeding (Åberg et al., 1997, Forward, 2009, Haglund and Åberg, 2000). In fact, perceived social norms, in particular, perceived expectations of other drivers have a significant effect on speeding (Haglund & Åberg, 2000). In addition to these intentional speeding behaviors, drivers may also speed inadvertently by failing to realize the posted speed limit or at which speed they are traveling. According to the results of a survey conducted by Transport Canada (2007), 51% of drivers who admitted to speeding at least occasionally, declared that in general they did not pay attention to the speed at which they were driving.

As for tailgating, drivers may maintain short headway times as they may believe that a sudden deceleration by a lead vehicle occurs rarely or as their past experiences may reinforce that driving at a short headway is fairly safe (Evans, 2004). There might also be an intentional component here, with drivers choosing shorter headways to keep others from cutting in front of their vehicles. Another possible reason is the inability of drivers to accurately estimate headway times and thus in turn fail to follow headways suggested for safe driving (Ben-Yaacov, Maltz, & Shinar, 2002). In previous studies, headway time estimation errors have been reported with drivers greatly overestimating headway times (at times twice the actual value), in particular for higher speeds (Ben-Yaacov et al., 2002, McLeod and Ross, 1983, Taieb-Maimon and Shinar, 2001). Given that drivers may speed inadvertently and they may not accurately estimate headway or assess if it is unsafe, technological devices which provide feedback to drivers based on unsafe speed or headways can potentially help reduce crashes.

A wide range of countermeasures such as law enforcement, variable message signs, educational messages, and supervising drivers are in effect to limit speeding and tailgating behaviors. Each medium or person provides some feedback on what the driver should be doing, or has already done in the hopes that they will correct their behavior in the future. However, such feedback is dependent on the environment, is not tailored to the behavior of the individual, and may be absent in some situations (Donmez, Boyle, & Lee, 2009). Therefore, such feedback may have little influence on drivers’ behavior. Emerging technology can circumvent the limits of current feedback and may provide an effective means by which to alert the driver to inappropriate behaviors. For example, Brookhuis and de Waard (1999) showed that the amount of time during which their participants drove 10% over the speed limit significantly decreased when they were provided with real-time visual and auditory feedback. Similarly, Adell, Várhelyi, and Hjälmdahl (2008) reported positive results for both auditory and haptic systems which provided feedback on speed. Further, Ben-Yaacov et al. (2002) showed that auditory warnings based on headway decreased the amount of time drivers spent maintaining short headways. However, these studies either did not assess behavioral change effects when the system was removed (Brookhuis & de Waard, 1999), or found no long-lasting positive effects (Adell et al., 2008), or found a lasting effect during a limited duration assessment in an experimental setting rather than a more naturalistic one (Ben-Yaacov et al., 2002). Although providing feedback consistently appears to provide benefits, whether the effects would sustain when feedback is no longer available is not well understood.

One approach to increase drivers’ motivation to adapt their driving behavior according to feedback is to utilize incentives. Incentives may be particularly effective if the behavior is intentional rather than inadvertent, such as speeding for thrill as opposed to speeding due to a failure of recognizing the speed limit. Incentives can significantly influence behavior, and rewarding desirable behaviors is usually more effective than penalizing undesirable behaviors, and can result in lasting behavioral modification (Brehm, 1966, Skinner, 1953). In Lahrmann, Agerholm, Tradisauskas, Berthelsen, and Harms (2012), an intelligent speed adaptation system was compared to the introduction of a monetary penalty for speeding. The former was found to be more effective than the latter. Further, when the system and the penalty were provided together, the results were similar to the case where the intelligent adaptation system was used alone. In contrast, Reagan, Bliss, Houten, and Hilton (2012) showed that a monetary incentive was the main driver of behavioral change regarding speeding. The reductions in speeding were similar when drivers had incentives only and when they had incentives combined with feedback. In general, it appears that incentive-based strategies to motivate behavioral change has not yet received much attention from the traffic safety research community despite the fact that insurance companies have already introduced vehicle-telemetrics-based premium incentives.

In the current paper, the effects of a dynamic feedback-reward system on enhancing speed compliance and promoting safe headway times are reported for when the intervention was available and when it was removed. An on-road field trial, SafeMiles, was conducted in Winnipeg, Manitoba, and consisted of three phases: baseline (two weeks), intervention (twelve weeks), and post-intervention (two weeks). During the intervention phase, real-time visual feedback was provided on an in-vehicle display. Participants also accumulated reward points and could view related information on a secured website. A similar feedback-reward system was initially designed and evaluated in the Netherlands, through the Belonitor Trial (Mazureck & van Hattem, 2006). In the Belonitor Trial, percentage of kilometers covered at a safe speed and at a safe headway time improved with the intervention. However, the positive effects did not sustain when the intervention was no longer provided. Different than the Belonitor Trial, in this paper, we first investigate natural groupings in drivers based on their speeding and tailgating behaviors as collected through the baseline period. We then evaluate if differences exist in these groups in terms of how they responded during the intervention and post-intervention phases. As will be discussed in more detail, grouping drivers reveals significant effects which could not be determined otherwise.

Section snippets

Method

The SafeMiles Trial was commissioned by Transport Canada, and was conducted in Winnipeg, Manitoba, over a four-month period from mid-August to mid-November in 2009. The driving environment was a mixture of rural, urban, and suburban roads. The speed map covered an approximate 100 km zone and included the rural bedroom communities around Winnipeg to capture rural commuting into the city. Speed limit compliance and safe headway maintenance were assessed continuously through the use of GPS/GIS, a

Variables

Speeding behavior was measured through the speed limit compliance rate which was defined as the ratio of the compliant time (GPS based speed  posted speed limit + 2 km/h) over the total time spent driving within each experimental phase (baseline, intervention, and post-intervention) and speed limit (50, 60, 70, 80, 90, 100 km/h) combination. Tailgating behavior was measured through the headway time compliance rate which was defined as the ratio of the compliant time (headway time > 1.2 s) over the

Discussion

This paper presents findings from a field operational trial conducted to evaluate the effects of a feedback-reward system in enhancing speed limit compliance and promoting safe headway times. A cluster analysis based on baseline data revealed two driving styles: more speed and head compliant (Cluster A), less speed and headway compliant (Cluster B). Grouping the participants made it possible to investigate how different groups responded to the system. When individual differences were not

Acknowledgements

Funding for this research was provided by the Natural Sciences and Engineering Research Council (NSERC) of Canada and the University of Toronto Connaught New Researcher Award. We gratefully acknowledge Dr. Jing Feng for the valuable feedback she provided for this research. We would also like to acknowledge the Dutch Government for loaning us equipment used in the trial, as well as Gord Taylor and Frank Franczyk for their assistance with the trial planning, data collection, and equipment

References (39)

  • P. Wasielewski

    Speed as a measure of driver risk: Observed speeds versus driver and vehicle characteristics

    Accident Analysis and Prevention

    (1984)
  • L. Åberg et al.

    Observed vehicle speed and drivers’ perceived speed of others

    Applied Psychology

    (1997)
  • I. Ajzen et al.

    Understanding attitudes and predicting social behavior

    (1980)
  • A. Ben-Yaacov et al.

    Effects of an in-vehicle collision avoidance warning system on short- and long-term driving performance

    Human Factors

    (2002)
  • J.W. Brehm

    A theory of psychological reactance

    (1966)
  • B. Donmez et al.

    Designing feedback to mitigate distraction

  • Evans, L. (2004). Traffic safety. Science Serving...
  • M. Halkidi et al.

    On clustering validation techniques

    Journal of Intelligent Information Systems

    (2001)
  • Harms, L., Klarborg, B., Lahrmann, H., Agerholm, N., Jensen, E., & Tradisauskas, N. (2007). Effects of ISA on the...
  • Cited by (0)

    View full text