Older driver distraction: A naturalistic study of behaviour at intersections

Accident Analysis and Prevention 58 (2013) 271–278

Contents lists available at ScienceDirect

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

Older driver distraction: A naturalistic study of behaviour at intersections Judith L. Charlton ∗ , Matthew Catchlove, Michelle Scully, Sjaan Koppel, Stuart Newstead Monash University Accident Research Centre (MUARC), Building 70, Clayton, VIC 3800, Australia

a r t i c l e

i n f o

Article history: Received 28 October 2011 Received in revised form 5 November 2012 Accepted 20 December 2012 Keywords: Older drivers Naturalistic driving Intersections Driver distraction Self-regulation

a b s t r a c t This study examined older driver engagement in distracting behaviours (secondary activities) at intersections using naturalistic driving data from a larger study based in Melbourne, Australia. Of interest was whether engagement in secondary activities at intersections was influenced by factors such as driver gender and situational variables, in particular, those relating to the complexity of the driving environment. Specifically we expected that when making left/right turns, older drivers would reduce the proportion of time engaged in secondary behaviours at intersections which required gap judgements (partly controlled or uncontrolled) compared with intersections that were fully controlled by traffic signals. Consideration was given to engagement in secondary activity with hands off the wheel and when the vehicle was moving versus stationary. Older drivers aged between 65 and 83 years drove an instrumented vehicle (IV) on their regular trips for approximately two weeks. The IV was equipped with a video camera system, enabling recording of the road environment and driver and a data acquisition unit, enabling recording of trip distance, vehicle speed, braking, accelerating, steering and indicator use. Driving experience and demographics were collected and functional abilities were assessed using the Useful Field of View (UFOV), Trail Making Test B, Mini Mental Status Examination (MMSE), visual acuity and contrast sensitivity. The study yielded a total of 371 trips with 4493 km (99.8 h) of naturalistic driving data including 1396 left and right turns. Trips were randomly selected from the dataset and in-depth analysis was conducted on 200 intersection manoeuvres (approximately 50% left turns, 50% right turns). The most frequently observed secondary activities were scratching/grooming (42.5%), talking/singing (30.2%) and manipulating the vehicle control panel (12.2%). Glances “off road” 2 s or longer were associated with reading, reaching and manipulation of the vehicle control panel. Hands off the wheel was associated with reading. Key parameters associated with the percent of intersection time that drivers engaged in secondary activities were intersection complexity, vehicle status (moving vs. stationary) and traffic density. In conclusion, older drivers appeared to engage selectively in secondary activities according to roadway/driving situations, supporting the notion that drivers self-regulate by engaging in secondary tasks less frequently when the driving task is more challenging compared with less challenging manoeuvres. © 2012 Elsevier Ltd. All rights reserved.

1. Introduction Intersections feature prominently in the crash statistics for older drivers. Specific problem areas include failure-to-yield, looked-butfail-to-see, and inaccurate gap selection (Benekohal et al., 1994; Stamatiadis et al., 1991; Staplin et al., 2001; Mayhew et al., 2006). Despite this widely acknowledged problem, there is limited understanding about the real-world driving behaviours of older drivers. In particular, relatively little is known about the role of inattention and the propensity for older drivers to engage in distracting behaviours whilst undertaking turning manoeuvres at intersections.

∗ Corresponding author. Tel.: +61 3 9905 1903; fax: +61 3 9905 4363. E-mail address: [email protected] (J.L. Charlton). 0001-4575/$ – see front matter © 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.aap.2012.12.027

Previous research suggests that older drivers are more likely to be involved in crashes when turning across traffic (Griffin, 2004; Staplin et al., 1998; Chandraratna et al., 2002; Chandraratna and Stamatiadis, 2003; Mayhew et al., 2006) or when evaluating the gaps between their vehicles and other vehicles (Chandraratna and Stamatiadis, 2003; Oxley et al., 2006) compared to younger age groups. Recent literature has also demonstrated that a greater percentage of older drivers’ intersection crashes occur at stop sign–controlled intersections than at signalised intersections (Oxley et al., 2006; Braitman et al., 2007; Preusser et al., 1998; Viano and Ridella, 1996). Given the differences in older and younger driver crash types, researchers have speculated that the behaviours that lead to older driver crashes may be more related to inattention or slowed perception and responses than to deliberate unsafe actions that are more common in younger drivers (for a review, see Koppel et al., 2009). For example, Stutts et al. (2001) conducted an analysis of the

272

J.L. Charlton et al. / Accident Analysis and Prevention 58 (2013) 271–278

1995–1999 Crashworthiness Data System (CDS) data to determine the role of driver distraction in police reported crashes in the United States (where at least one vehicle was towed away) and the specific sources of this distraction. Younger drivers (less than 20 years) were more likely than drivers 65 years and older to be identified as distracted at the time of their crash: 11.7% of younger drivers were found to be distracted, compared to 7.9% of older drivers. In contrast, older drivers were more than three times more likely to have “looked but didn’t see” (16.5%) listed as a contributing factor in crashes compared to younger drivers (5.4%). However, Stutts et al. (2001) reported that these differences were not statistically significant. The findings of Stutts et al. (2001) are consistent with the findings of European research by Hakamies-Blomqvist (1994) who found that older drivers (65 years and older) were significantly less likely to be distracted by a non-driving activity (such as eating, drinking, smoking, listening to the radio, and conversing.) immediately preceding a crash (42%) than younger drivers (aged 26–40 years; 57%). Similar findings were reported in an Australian-based study. McEvoy et al. (2007) examined the prevalence and type of distracting activities involved in serious injury crashes. Interviews were conducted with hospitalised drivers within hours of their crash. Crashes involving a distracting activity were more likely to be reported by younger drivers (17–29 years) compared with drivers aged 50 years and older (39.1% vs. 21.9%). In a recent study focusing on intersection crashes, Braitman et al. (2007) confirmed that failure-to-yield the right-of-way crashes increased with age and noted that the reasons for failure-toyield crashes tended to vary by age. Using information from police reports and follow-up telephone interviews with drivers, the authors found that compared with drivers aged 35–54 years and 80 years and older, drivers aged 70–79 years made more evaluation errors – seeing another vehicle but misjudging whether there was adequate time to proceed. In contrast, drivers aged 80 years and older predominantly failed to see or detect the other vehicle. Drivers aged 35–54 years also tended to make search errors which were more often attributed to distraction. A study of older driver ‘blackspot’ crash sites in Australasia noted that the principal problem for older drivers at intersections was selecting safe gaps (Fildes et al., 2000). The authors noted that the difficulty in gap selection was often exacerbated by factors such as high task complexity and the presence of other road users. It is also widely acknowledged that the increased complexity of intersections could produce very high momentary cognitive overload which would result in driving errors of the kind discussed above (Hakamies-Blomqvist et al., 1999; Hancock et al., 1990). While the over-involvement of older drivers in intersection crashes is well documented, it is possible that some older drivers modify this risk at least in part, through self-regulatory behaviour (Charlton et al., 2006). A number of older drivers reportedly use an extensive range of adaptive strategies including driving at slower speeds, avoiding adverse weather conditions, avoiding busy traffic, limiting driving to the daytime and travelling shorter distances (Baldock et al., 2006; Charlton et al., 2003, 2006; Smiley, 2004). Despite the successful use of self-regulation in a range of contexts, it is less well known whether older drivers modify their driving at intersections and in particular, whether their selfregulation is associated with reducing secondary (non-driving related) activity at times of high cognitive load. A small number of published studies have surveyed drivers’ self-reported engagement in or intention to engage in secondary activities, albeit predominantly with younger and middle-aged drivers. Findings suggest that older drivers are also less likely to report eating or drinking in the car compared to other age groups. Further, approximately 40% of older drivers reported using stops in traffic to engage in

secondary activities (Young and Lenne, 2010). Similarly, surveys conducted by Lansdown (2012) have found that age was negatively predictive of engagement in distracting behaviours. Lerner et al. (2008) also reported age group differences in drivers’ propensity for distraction, including use of mobile telephone or navigational system and eating, when driving. A consistent finding was that older drivers were less willing to engage in these behaviours than younger drivers and perceived the risk of engaging in secondary activities to be higher than the younger drivers. While the findings reviewed above suggest that older drivers may self-regulate distracting behaviours, they are derived from drivers’ self-report data, which may or may not differ from realworld driving behaviours. With developments of covert in-vehicle technologies and naturalistic driving methods for monitoring driver behaviour, there is a growing body of evidence on drivers’ frequency of engagement in distracting activities and their role in crash causation (Dingus et al., 2006; Klauer et al., 2005; Sayer et al., 2005; Stutts et al., 2003). Research from the 100-Car Study showed that almost 80% of all crashes and 65% of near crashes involved the driver looking away from the road prior to the conflict (Dingus et al., 2006). U.S. naturalistic data also shows that the most common distracting behaviours are manipulating audio controls, conversing, eating or drinking, grooming, reading or writing and using a mobile (Sayer et al., 2005; Stutts et al., 2003). Observational studies using naturalistic methods have also identified some negative consequences of distraction on driver performance. For example, Stutts et al. (2003) reported that reading, mobile telephone use and reaching for an object were associated with an increased likelihood of the driver taking both hands off the steering wheel. Intuitively, this action can result in the vehicle wandering within the lane or crossing into adjacent lanes. Furthermore, the secondary activities of reaching, reading and using a mobile telephone were associated with the diversion of drivers’ eyes off the roadway. What remains unclear is whether these findings generalise to the older driver population. The sample sizes were small and unrepresentative and no specific figures were given for older drivers as a group. An advantage of using naturalistic methods to study driving behaviour is that this affords the opportunity to investigate driving patterns across a range of driving situations and conditions. For instance, Sayer et al. (2005) found that drivers of all ages were less likely to engage in some distracting behaviours when braking, on wet roads, travelling around bends, or during night driving. This study suggested that drivers choose to perform secondary activities at what might be perceived as safer times. Stutts et al. (2003) also found that drivers tended to engage in distractions more frequently when they were stationary than when they were moving. To date, this approach has not been applied in the study of older driver distraction. This study examined older driver engagement in distracting behaviours (secondary activities) at intersections using naturalistic driving data from a larger study based in Melbourne, Australia. Of interest was whether engagement in secondary activities at intersections was influenced by driver characteristics and situational variables, in particular, those relating to the complexity of the driving environment. The primary hypothesis was that older drivers would exercise a greater level of self-regulation by reducing potential distractions during complex driving manoeuvres. Specifically we expected that when making left/right turns, older drivers would reduce the proportion of time engaged in secondary activities at intersections which required gap judgements (partly controlled or uncontrolled) compared with intersections that were fully controlled by traffic signals. In addition, consideration was also given to level of engagement in secondary activities when the vehicle was moving versus stationary. It was expected that the driver might perceive that driving task is more demanding when the vehicle

J.L. Charlton et al. / Accident Analysis and Prevention 58 (2013) 271–278 Table 1 Summary of participant functional measures.

MMSE score (>23 normal functioning) Trail Making Test B (s) UFOV Divided Attention (ms) Visual acuity (LogMAR, Binocular) Contrast sensitivity (Pelli–Robson, Binocular)

273

Table 2 Summary of instrumentation of study vehicle. Mean

Range

28.20 81.90 90.00 0.00 1.70

26–30 36–130 16–323 0.10 to +0.10 1.65–1.80

is moving and modify their engagement in secondary activities accordingly. 2. Method In order to examine whether older drivers adjust their engagement in secondary activities while driving, a naturalistic, observational approach was implemented. Participants drove a study vehicle during the collection of observational data. The observational data was supplemented by functional measures of cognitive and visual abilities and a questionnaire designed to gather demographic and driving information. 2.1. Participants Participants were ten older drivers (six males and four females) aged between 65 and 83 years (Mean = 74.1, S.D. = 5.76), and who held a driving licence for an average of 50.6 years (S.D. = 8.21). To ensure a high level of compatibility between drivers’ own vehicle and the study vehicle, it was also a requirement that participants were familiar with driving a large sedan of similar size to the study vehicle. Drivers were recruited from a pool of older drivers who had previously participated in other older driver studies conducted by the authors and had agreed to be contacted for future studies. Eight of the ten participating drivers, were married couples whose driving trips were recorded during the same observation period. Participants’ (corrected) binocular visual acuity (LogMAR; Bailey and Lovie, 1976) and binocular contrast sensitivity (Pelli et al., 1988) were within normal range. The mean binocular visual acuity was 0.00 (range −0.10 to +0.10). The mean binocular contrast sensitivity (Pelli–Robson) was 1.66 (range 1.65–1.80). Useful Field of View (UFOV) scores for Subtest 2: Divided Attention (ms) were below impairment threshold associated with higher crash risk, suggesting adequate visual attention function on this measure (Ball et al., 2006). All participants were screened for presence of cognitive impairment based on a cut-off score of 23 on the Mini Mental Status Examination (MMSE; Folstein et al., 1975). No participant reported a history or current evidence of any neurological or psychiatric impairment. A summary of participants’ functional abilities is presented in Table 1. 2.2. Study vehicle The study vehicle was a large, luxury model family sedan with automatic transmission. The vehicle was fitted with a data acquisition unit and a camera/recording system, providing images of the driver and front seat passenger, the road and traffic ahead, laterally and to the rear. A summary of the instrumentation fitted to the vehicle is presented in Table 2. The camera/recording system comprised seven cameras, positioned in such a way as to gain an overall view of the forward and lateral road scene at intersections and the interior of the cabin with minimal disruption to the driver’s view and concealed so as not to be obvious to the vehicle occupants. Small CCD cameras were located on the vehicle as described below:

Video recording system Camera system Data acquisition unit a

Appro® DVR with 250 GB hard disk – Mobile Vehicle Video Recorder DVR-3064; 25 frames/s Small CCD cameras with lenses 90◦ /150◦ viewing angle Motec ACL/ADL3a data recorder

http://www.motec.com/adl3/adl3overview/ (accessed 13.07.11).

• Cameras 1 and 2 were located within a watertight dome mounted on a roof rack. Both cameras had wide angle lenses and viewed the road ahead and oblique side views (one viewed the left side and one viewed the right side, with a central overlap area). Together, these cameras provided a forward viewing angle of approximately 240◦ . • Camera 3 was embedded within the internal rear-view mirror (behind a pinhole, approximately 10 mm in diameter), providing a view of the driver’s face. • Camera 4 was located behind the centre internal rear-view mirror, providing a view of the forward road and traffic scene. • Camera 5 was located in a light cavity inserted in the roof of the vehicle between the front seats. Its purpose was to view the instrument panel to record driver/passenger use of the HMI/HVAC unit. • Camera 6 was located under the steering column in the driver’s foot well. The purpose of this camera was to view braking and acceleration behaviour. • Camera 7 was located on the rear window just beneath the internal rear stop lamp providing a view of the road and traffic scene to the rear of the driver’s vehicle. • All cameras were connected via cables concealed within the vehicle trim to one of two digital 4-channel video recorders (DVR) mounted in the trunk of the vehicle. Cameras 1–4 provided the ‘primary’ camera views and were connected to one DVR. Cameras 5–7 were connected to a second DVR and the output of a second DVR. The vehicle instrumentation was triggered automatically with the vehicle ignition. When the driver started the vehicle a 12 V signal was sent to the video recording and the vehicle data logging equipment, activating the system creating new files for recording. The duration of the start-up process took approximately 15 s. When the ignition was switched off, the system automatically shut down and data files were saved by date and time. The recording system could also be de-activated by the driver by means of pressing a (red) button on the dash behind the steering wheel. This feature was necessary to satisfy ethics requirements and allowed participants to opt out of the study temporarily by shutting down the recording system for any reason at the start of a trip or whilst driving. The status of the recording system was indicated by a dim red light (when recording) so as not to distract the driver. If the red button was pressed, the system stopped recording for the remainder of the trip. Vehicle data was captured using a data acquisition unit with capacity for 1 GB of logging memory. The system recorded vehicle information (10 Hz) including speed, acceleration, braking, steering wheel angle and indicator use and had the capacity to synchronise vehicle data with the video files. 2.3. Procedure Participants were asked to drive the instrumented ‘study vehicle’ on their regular trips for two weeks. Prior to the commencement of each two-week observation period, drivers’ informed consent was obtained in accordance with Institutional Ethics Committee requirements. Participants received $80 as partial

274

J.L. Charlton et al. / Accident Analysis and Prevention 58 (2013) 271–278

recompense for their commitment of time and for petrol costs associated with trips to the University for assessment sessions. Handover (and return) of the study vehicle occurred at participants’ homes during a briefing session conducted by members of the research team. Participants also completed legal documents that were necessary to comply with the loan of the study vehicle. Following each participant study period, the vehicle was serviced, cleaned, filled with petrol, photographed (standard inspection photos), and data downloaded to a laptop. During handover participants were briefed about the operation of the vehicle and the recording equipment including how to turn cameras on/off, should they so choose. A written summary of this information was provided in the vehicle. Participants were instructed to drive the vehicle as they would normally drive their own vehicle (including safely and responsibly). An orientation drive was included at the time of vehicle handover whereby the participant drove the vehicle with a researcher as a passenger. In addition to their regular driving trips during the two-week study period, drivers were asked to drive to the University twice, following a pre-defined route (entering and exiting the University). These pre-defined route trips were not considered part of the participants everyday driving and therefore were not included in the analyses for this paper. Participants completed the functional ability assessment during one of the sessions. 2.4. Data coding and analysis Video recordings were viewed, coded and analysed to examine driver behaviour, intersection type, and driving manoeuvre. The study yielded a total of 371 trips with 4493 km (99.8 h) of naturalistic driving data including 1396 left and right turns. Trips were randomly selected from the dataset and in-depth analysis was conducted on 200 intersection manoeuvres (approximately 50% left turns and 50% right turns). The total observation time across all intersections was 78.27 min. Selection of intersections was determined by several factors. Firstly, to ensure that the participants had adapted to driving the instrumented vehicle, only video footage collected after the first hour was analysed. Dingus et al. (2006) reported that after 1 h of driving an instrumented vehicle participants adapted to being monitored and drove more naturally. Secondly, to minimise bias, trip files were randomly selected without replacement from each participant’s data set, and every intersection in the video file was coded until 20 intersections with left or right turn manoeuvres had been analysed. In total 200 intersections were selected for analysis (i.e., 20 intersections per participant). Key variables were: Secondary activities: The proportion of intersection manoeuvre time engaged in secondary activities was the primary dependent variable. Drawing from key distraction studies (Dingus et al., 2006; Sayer et al., 2005; Stutts et al., 2003), nine secondary activities were identified for coding: Talking/singing, scratching/grooming, mobile telephone use, adjusting vehicle audio and climate controls, reaching for objects, reading, eating/drinking, gesturing, and adjusting glasses. Attention to the roadway: For each secondary activity, driver’s looking behaviour was logged. This variable was recorded dichotomously, as ‘yes’ if the driver paid attention to the roadway (determined by glances to the forward roadway and to the left/right to check cross-flow traffic) or ‘no’ if the driver looked inside the car or glanced elsewhere outside the vehicle. If the driver’s gaze was directed away from the roadway for a period of two or more seconds at a time, this was classified as inattentive. This definition was based on findings reported by Klauer et al.

(2006) showing that inattention to the roadway for two or more seconds almost doubles the odds of a crash or near crash. Hands on the wheel: The status of drivers’ grasp of the steering wheel was recorded while engaging in a secondary activity (i.e., none, one, two hands). This variable was of interest as it provides a measure of vehicle control. Research by Stutts et al. (2003) showed significant lane deviation associated with having both hands off the wheel. Intersection type: Intersections were classified as: fully controlled, with traffic lights with an arrow for right turns; partially controlled, with traffic lights but without a turning arrow; roundabouts; and uncontrolled intersections, with stop signs or give-way signs only. For the purpose of this study, the intersection classification scheme reflected ascending levels of complexity of decision-making required and therefore was a proxy for cognitive load associated with the driving manoeuvre. At fully controlled intersections, reflecting lowest complexity, on-coming and cross flow traffic are restricted by lights and drivers have right-of-way to complete their turn when the turning arrow is green. Partly controlled intersections by definition have some traffic flow which is not restricted by traffic light/arrow, thus representing moderate levels of complexity. Similarly, roundabouts have partial control of traffic flow which is determined by the road geometry. Uncontrolled intersections arguably have the highest demands on decision making since traffic flow is determined by signage and relies heavily on driver decisions about safe gaps in traffic. Manoeuvre type: Left and right turns were studied. For the purpose of this study, the start/end of an intersection turn manoeuvre was defined as the time that the vehicle indicator was switched on/off, as determined from the vehicle data recorder. Road type: Roadways were coded as divided or undivided. Vehicle status: The state of motion of the vehicle was coded as stationary/moving by inspection of the videotape. It was expected that the driver might perceive that driving task is more demanding when the vehicle is moving and modify their engagement in secondary activities accordingly. Snapper software (Copyright, Webbsoft Technologies, 2008) was used as the viewing platform to facilitate the logging of events into a database. The duration of activities (h:min:s), including secondary activities was recorded using the video frame count (25 frames/s) and expressed as a proportion of the intersection manoeuvre time. A second, independent rater coded a random selection of 10% of the intersections. Inter-rater reliability was 85%. The statistical software program SPSS version 16 was used to analyse the data. Secondary activity events at intersection were treated as independent units and were pooled across participants. Descriptive statistics were used to describe frequency and duration of secondary activities. Logistic General Estimating Equation modelling was used to model factors influencing the outcome measure of interest: engagement in secondary activities, expressed as the percentage of total intersection manoeuvre time. This approach was appropriate given the binary nature of the outcome variable and to adjust for potential correlations among the multiple observations per participant (Hardin and Hilbe, 2003; Liang and Zeger, 1986). 3. Results For the 200 intersection manoeuvres selected for observation (78.27 min of driving), drivers were observed to negotiate turns at each of the four designated types of intersections: fully controlled (traffic light/with turn arrow), partly controlled (traffic light/no turn arrow), roundabout and uncontrolled (stop/give way sign). Turn manoeuvres at uncontrolled intersections were the most

J.L. Charlton et al. / Accident Analysis and Prevention 58 (2013) 271–278

275

Table 3 Mean (S.D.) duration, frequency (%) of secondary activities. Secondary task

Car stationary (s)

S.D.

N (%)

Car moving (s)

S.D.

N (%)

Scratch/groom Talk/sing Control panel Read Reach Adjust glasses Gesture Mobile telephone

7.39 7.00 7.07 9.20 3.67 5.00 2.60 12.0

11.03 8.17 6.04 8.07 2.88

41 (29.50%) 15 (10.79%) 14 (10.07%) 5 (3.60%) 6 (4.32%) 1 (0.72%) 5 (3.60%) 2 (1.44%)

2.56 4.19 2.67 0 0 1.50 0 0

2.12 2.83 1.53

18 (12.95%) 27 (19.42%) 3 (2.16%) 0 (0%) 0 (0%) 2 (1.44%) 0 (0%) 0 (0%)

89(64%)

3.4

2.59

All tasks

1.34 11.31

6.93

8.9

frequently observed (71.7%), fully controlled intersections made up 20.8% and roundabouts and partially controlled intersections together made up 7.6% of all intersections observed.

0.71

50 (36%)

Total N (%) 59 (42.45%) 42 (30.22%) 17 (12.23%) 5 (3.60%) 6 (4.32%) 3 (2.16%) 5 (3.60%) 2 (1.44%) 139 (100%)

total time for these events. Lowest rates of eyes off the road were observed for scratching/grooming, talking/singing and gesturing and similarly low rates were observed with hands off the wheel for talking/singing (less than 8%).

3.1. Frequency and duration of secondary activities 3.3. Predictors of engagement in secondary activities Secondary activities observed while participants were negotiating intersections, included: scratching/grooming (42.45%), talking/singing (30.22%), manipulating the control panel (12.23%), reading (3.60%), reaching for objects (4.32%), adjusting glasses (2.16%), gesturing (3.60%) and mobile telephone use (1.44%). Across the 200 intersections, drivers engaged in a single secondary activity on 41 occasions (20.5%). On 25 occasions (12.5%), drivers engaged in multiple secondary activities during the intersection manoeuvre, with the maximum number of secondary activities observed being 11 (at n = 2 intersections). Table 3 shows the mean duration of secondary activities and the frequency and percentage occurrence of these events, both whilst the vehicle was stationary and whilst moving. Overall, the most frequently observed secondary activities were scratching/grooming (42.5%), talking/singing (30.2%) and manipulating the vehicle control panel (12.2%). Notably, use of mobile telephones comprised less than 2% of the total time for intersection manoeuvres. Drivers engaged in secondary tasks such as reading, reaching for an object, gesturing and using a mobile telephone, only while the vehicle was stationary. In contrast, drivers were more likely to engage in activities such as talking/singing when the vehicle was in motion (19.42% vs. 10.79%). 3.2. Secondary activities and driving performance

4. Discussion

Table 4 shows the proportion of time for each secondary activity for which drivers had their hands off the wheel and their eyes off the roadway. Reading and mobile telephone use were associated with the highest levels of hands off the wheel and eyes diverting from the roadway (100%). Reaching for objects, gesturing, adjusting glasses and scratching/grooming were also associated with moderate levels of hands off the wheel (18.5–50%) and adjusting spectacles required eyes off the road on around one third of the Table 4 Secondary activities and driving performance. Secondary activity

% Time both hands off wheela

% Time eyes not on roadwaya

Scratch/groom Talk/sing Control panel Read Reach Adjust glasses Gesture Mobile telephone

18.50 5.10 13.30 100.00 50.00 33.30 40.00 100.00

5.60 7.70 0.00 100.00 100.00 33.30 0.00 100.00

a

Logistic General Estimating Equation Models were estimated to identify driver, road and driving manoeuvre-related predictors of the percentage of intersection manoeuvre time engaged in secondary activities. Results from the model showing the estimated odds of engagement in secondary activities are summarised in Table 5. The statistical significance of each variable and 95% confidence limits is also shown in the table. Odds estimates greater than 1 indicate proportionately greater time engaged in secondary activities for the specified parameter compared with the relevant referent group. Low statistical significance values indicate the result is unlikely to have been obtained through chance variation in the data. The odds of engagement were three times higher at fully controlled intersections compared with uncontrolled intersections, p < 0.0001; and 65% lower at partly controlled intersections compared with uncontrolled intersections, p < 0.002. Odds of engagement in secondary activities were also 75% lower when the vehicle was moving compared to when the vehicle was stationary, p < 0.001; and 47% lower when traffic density was low compared with moderate/high. Turning direction (left/right), road type (divided/undivided) and gender were not significantly associated with drivers’ engagement in secondary activities, p’s > 0.05.

Percentage of time per activity both hands off wheel/eyes not on roadway.

While previous studies have examined sources of driver distraction using a naturalistic, observational approach (e.g., Stutts et al., 2003; Sayer et al., 2005; Klauer et al., 2006), the study reported here was the first to engage naturalistic driving methods to focus on older drivers’ engagement in secondary activities in response to challenging road and traffic conditions. Across the 200 intersection manoeuvres studied in depth, the most frequently observed secondary activities were scratching/grooming (42%), talking/singing (30%) and manipulating the control panel (12%). These findings are consistent with previous naturalistic driving studies in which conversation and grooming were amongst the most common activities (Sayer et al., 2005; Stutts et al., 2003). Interestingly, while eating or drinking have been reported to be amongst the highest frequency activities in previous studies (Stutts et al., 2003), no food or drink was consumed during intersection negotiation in the current study. It is possible that this discrepancy could reflect the specific driving manoeuvre under investigation; in the current study, analyses were restricted to intersection negotiation only, while previous investigations have observed across a full range of driving contexts. Alternatively, differences in findings may relate to differences in the age of study

276

J.L. Charlton et al. / Accident Analysis and Prevention 58 (2013) 271–278

Table 5 Summary results for GEE Model predicting engagement in secondary activities expressed as a percentage of total intersection manoeuvre time. Variables

Referrent

Relative odds

Statistical significance

LCL

UCL

Intersection type: fully controlled Intersection type: partly controlled Intersection type: roundabout Turn direction: left Vehicle status: moving Road type: undivided Traffic density: low Gender: males

Uncontrolled Uncontrolled Uncontrolled Right Stationary Divided High Females

2.982 0.342 0.887 1.550 0.259 0.865 0.531 1.176

0.000* 0.002* 0.845 0.239 0.001* 0.665 0.044 0.859

2.008 0.172 0.268 0.748 0.114 0.449 0.288 0.196

4.429 0.680 2.938 3.210 0.588 1.667 0.982 7.047

participants and potential age-related differences in the propensity to engage in eating/drinking behaviour. Indeed, older drivers have been found to be almost four times less likely than younger drivers to report eating/drinking on regular occasions (Young and Lenne, 2010). The finding is suggestive of a strategic decision or form of self-regulation practice by older drivers. Two aspects of driver performance which have high relevance for safety are hands off the wheel (Stutts et al., 2003) and eyes off the roadway for longer than 2 s (Klauer et al., 2006). Driving with hands off the wheel has been shown to result in variability in vehicle lane position and drifting into adjacent lanes (Stutts et al., 2003). In the current study, secondary activities that were most commonly associated with both hands being taken off the wheel were reading (100%), mobile telephone use (100%) and reaching for an object (50%). This was not surprising since these activities potentially involve a manual component. All of these secondary activities were restricted exclusively to times when the vehicle was stationary, suggesting that the older drivers in this study chose to perform these behaviours when driving demands were perceived to be less complex and risk of a crash lower. Given the high prevalence of failure-to-yield errors in crash reports for older drivers, it was of interest to examine the association between secondary activities and driver looking behaviour, in particular, the activities for which drivers’ eyes were diverted away from the roadway for longer than 2 s. Previous research has shown that crash risk almost doubles when a driver’s eyes are diverted from the roadway for 2 s or more (Klauer et al., 2006). In the current study, drivers’ eyes were diverted from the roadway for 86–100% of the time when using a mobile telephone, reading, reaching for an object and manipulating the control panel. With the exception of control panel manipulation, the occurrence of these activities was restricted to when the vehicle was stationary. These findings were similar to those reported by Stutts et al. (2003) who found that drivers’ diversion of gaze was associated with reaching, reading and dialing/answering a mobile telephone. One interesting point of difference between the studies was that Stutts and colleagues reported that manipulating the control panel was not significantly related to eye diversion. It is possible that the older drivers in the current study were more reliant on visual guidance due to lack of familiarity with the positioning of the audio control unit in the study vehicle. Hence it will be important to explore this effect further in a more natural context in the drivers’ own vehicle. Using Logistic General Estimating Equation modelling, it was possible to explore the relative importance of a number of variables in determining older drivers’ propensity to engage in secondary activities while driving. Of particular interest was whether engagement in potentially distracting secondary activities at intersections varied depending on the complexity of the driving situation. The analysis showed that the percentage of time engaged in secondary activities was influenced by intersection type and vehicle status (moving vs. stationary). A marginal effect was also found for traffic volume, with the odds of engagement in secondary activities lower in low-volume traffic compared with high-volume traffic. This effect may be because the vehicle is more likely to be moving

when the traffic volume is lower. It is also possible that effects for traffic volume may be influenced by intersection type. For example, intersections with traffic lights are likely to be on roads which carry high traffic volume while uncontrolled intersections are more commonly found on residential roads with lower traffic volume. However, given the relatively small data set, it was not possible to examine these potential interaction effects. Neither driver gender nor road type significantly affected engagement in secondary activities. Similarly, no significant association was found between turning direction (left/right) and drivers’ propensity to engage in secondary activities. This finding can be attributed to a lack of power to detect differences between left and right turn behaviours. Given the evidence for higher rates of crashes amongst older drivers when making right turns (across traffic) compared with left turns (e.g., Griffin, 2004; Chandraratna and Stamatiadis, 2003), it might be expected that differences might be observed based on real or perceived difficulty of right turns. This finding warrants further research with a larger data set. The findings of the current study from observations of on-road driving suggest that older drivers engaged less in secondary activities at uncontrolled intersections where the complexity of gap judgements was greatest and when the car was moving, compared with less demanding intersection manoeuvres. This is consistent with evidence reported by Sayer et al. (2005) who showed that drivers are less likely to engage in secondary activities under certain driving conditions that may be considered more challenging. For example, Sayer and colleagues reported lower use of mobile telephones when braking and when driving on curved roads. Findings from this study are also consistent with self-reported avoidance of challenging driving situations (Baldock et al., 2006; Charlton et al., 2003, 2006; Smiley, 2004) and more specifically, with respect to potential driver distractions, the current findings concur with drivers’ reports that they engage in secondary activities only when stopped in traffic (Young and Lenne, 2010). While the findings of this study suggest that the older drivers may reduce some of the risks associated with secondary activities by refraining from these activities at intersections with lower decision-making demands and while the vehicle is stationary, this does not mean that it is a safe practice. A concern is that drivers may underestimate the potential risk associated with secondary activities, particularly when undertaking intersection manoeuvres. As discussed, intersections pose a significant problem for older drivers as evidenced by their over-representation in intersection crashes (Braitman et al., 2007). Moreover, there is a substantial body of literature demonstrating large and robust age-related deficiencies in dual-task environments (Salthouse, 1991) and management or co-ordination of multiple tasks (Korteling, 1994). This has serious implications for driver distraction because the additional cognitive load in having to process different sources of information in a serial mode may slow the drivers’ decision-making process to dangerous levels and in turn, lead to safety errors and increased risk of crash (Hakamies-Blomqvist et al., 1999). As highlighted by Sayer et al. (2005), the interpretation of drivers’ decisions to engage in potentially distracting secondary

J.L. Charlton et al. / Accident Analysis and Prevention 58 (2013) 271–278

activities is somewhat speculative. Notwithstanding that naturalistic driving studies of the kind presented here have enhanced our understanding the way in which older drivers negotiate complex manoeuvres and their propensity to engage in secondary behaviours, there are inherent difficulties in interpreting these behaviours as drivers’ conscious choice to adjust their driving behaviour to minimise risk. Some limitations of this study are noted. Firstly, drivers were observed driving a study vehicle rather than their own private car. While every attempt was made to recruit participants who drove a large vehicle with auto-transmission control in order to closely match the study vehicle, it is possible that participants’ lack of familiarity with the vehicle influenced their driving. Indeed in previous naturalistic driving research in which both leased and private vehicles were used, participants who drove a loan vehicle had a slightly higher risk of crash than participants who used their own vehicle (Dingus et al., 2006). Secondly, it is possible that drivers modified their driving because they knew they were being observed. However, it is unlikely that this fully explains the findings since it could reasonably be expected that drivers might modify their behaviour under both complex and less complex driving situations. Thirdly, the sample size was small and the findings are based on twenty intersection manoeuvres for ten drivers. Appropriate statistical methods were applied to study distraction events with adjustment for potential correlations due to multiple observations for each participant. Nevertheless, further research is warranted to demonstrate that the findings are robust for a larger, more representative sample of older drivers. Thirdly, it is acknowledged that there was no comparison with a younger group, nor was there longitudinal data to study changes across the age span. Further research across a broader age range of drivers is needed to investigate whether the findings are unique to older drivers or whether the same level of self-regulation might be apparent in other age groups. 5. Conclusions The paper describes a novel application of naturalistic driving methods to study older driver distraction at intersections. Drivers appeared to engage selectively in secondary activities according to roadway/driving situations. The findings support the notion that older drivers self-regulate by limiting their engagement in secondary activities when the driving task is more challenging compared with less demanding situations. Further research is planned to expand the data set and explore further the effects of age, functional abilities, presence of passengers and other variables on older driver distraction. For the first time, this study provides realworld driving evidence of older drivers’ self-regulation in relation to potentially distracting secondary activities that may impact on their safety. As the population ages and increasing numbers of older drivers are on the road, it will become increasingly important that road safety strategies are targeted to reduce specific behaviours that predispose this group to crash risk and promote safer driving practices. The findings of the current study offer some preliminary information to guide the development of older driver awareness/training programmes on safe driving strategies. Acknowledgments This project was funded through the Australian Cooperative Research Centre for Advanced Automotive Technology (AutoCRC) in collaboration with GM Holden Australia. The project team acknowledges the significant supplementary funding provided by the Monash University Accident Research Centre (MUARC). Valuable technical advice was provided throughout the project by Mike

277

Hammer, Haydn Leembruggen, Dion Thomas and Jeff Brown at Holden Innovation, GM Holden. In addition, we thank Drs. Tom Dingus, Jonathon Hankey and Jon Antin from Virginia Tech Transportation Institute for their generous assistance and advice on vehicle data logging methods in the early stages of the project. We also acknowledge significant input from MUARC researchers David Taranto, Ron Laemmle and Nebojsa Tomasevic who were responsible for installing the vehicle instrumentation and data parsing, Anna Devlin for assistance with data collection and Jennie Oxley for advice on study design. References Bailey, I., Lovie, J., 1976. New design principles for visual acuity letter charts. American Journal of Optometry and Physiological Optics 53, 740–745. Baldock, M.R.J., Mathias, J.L., McLean, J., Berndt, A., 2006. Self-regulation of driving and older drivers’ functional abilities. Clinical Gerontologist 30, 53–70. Benekohal, R., Michaels, R., Shim, E., Resende, P., 1994. Effects of aging on older drivers’ travel characteristics. Transportation Research Record 1438, 91–98. Braitman, K.A., Kirley, B.B., Ferguson, S., Chaudhary, N.K., 2007. Factors leading to older drivers’ intersection crashes. Traffic Injury Prevention 8 (3), 267–274. Chandraratna, S., Mitchell, L., Stamatiadis, N., 2002. Evaluation of the Transportation Safety Needs of Older Drivers. Department of Civil Engineering, University of Kentucky, Lexington, KY. Chandraratna, S., Stamatiadis, N., 2003. Problem driving maneuvers of elderly drivers. Transportation Research Record 1843, 89–95. Charlton, J., Oxley, J., Fildes, B., Oxley, P., Newstead, S., O’Hare, M., Koppel, S., 2003. An Investigation of Self-Regulatory Behaviours of Older Drivers (Report No. 208). Monash University Accident Research Centre, Melbourne, pp. 1–83. Charlton, J.L., Oxley, J., Fildes, B., Oxley, P., Newstead, S., Koppel, S., O’Hare, M., 2006. Characteristics of older drivers who adopt self-regulatory driving behaviours. Transportation Research Part F 9, 363–373. Dingus, T.A., Klauer, S.G., Neale, V.L., Petersen, A., Lee, S.E., Sudweeks, J., Perez, M.A., Hankey, J., Ramsey, D., Gupta, S., Bucher, C., Doerzaph, Z.R., Jermeland, J., Knipling, R.R., 2006. The 100-Car Naturalistic Driving Study, Phase II; Results of the 100-Car Field Experiment. National Highway Traffic Safety Administration, Washington, DC. Ball, K.K., Roenker, D.L., Wadley, V.G., Edwards, J.D., Roth, D.L., McGwin, G., Raleigh, R., Joyce, J.J., Cissell, G.M., Dube, T., 2006. Can high-risk older drivers be identified through performance-based measures in a department of motor vehicles setting? Journal of the American Geriatrics Society 54, 77–87. Fildes, B., Corben, B., Morris, A., Oxley, J., Pronk, N., Brown, L. et al., 2000. Road Safety Environment and Design for Older Drivers. Report No. AP-R169, Austroads, Sydney, Australia. Folstein, M.F., Folstein, S.E., McHugh, P.R., 1975. “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research 12 (3), 189–198. Griffin III, L.I., 2004. Older Driver Involvement in Injury Crashes in Texas, 1975–1999. AAA Foundation for Traffic Safety, Washington, DC. Hakamies-Blomqvist, L.E., 1994. Compensation in older drivers as reflected in their fatal accidents. Accident Analysis and Prevention 26, 107–112. Hakamies-Blomqvist, L.E., Mynttinen, S., Backman, M., Mikkonen, V., 1999. Agerelated differences in driving: are older drivers more serial? International Journal of Behavioural Development 23, 575–589. Hancock, P.A., Wulf, G., Thom, D., Fassnacht, P., 1990. Driver workload during differing driving maneuvers. Accident Analysis and Prevention 22, 281–290. Hardin, J., Hilbe, J., 2003. Generalized Estimating Equations. Chapman & Hall/CRC, London, ISBN 978-1-58488-307-4. Klauer, S.G., Dingus, T.A., Neale, V.L., Sudweeks, J., Ramsey, D.J., 2006. The Impact of Driver Inattention on Near-Crash/Crash Risk: An Analysis Using the 100-Car Naturalistic Driving Study Data. DOT HS 810 594. Koppel, S., Charlton, J., Fildes, B., 2009. Distraction and the older driver. In: Regan, M., Lee, J., Young, K. (Eds.), Driver Distraction: Theory, Effects and Mitigation. CRC Press, London, pp. 353–382. Korteling, J., 1994. Effects of ageing, skill modification and demand alternation on multiple-task performance. Human Factors 36 (1), 27–43. Liang, K.-Y., Zeger, S., 1986. Longitudinal data analysis using generalized linear models. Biometrika 73 (1), 13–22. Lansdown, T.C., 2012. Individual differences and propensity to engage with invehicle distractions – a self-report survey. Transportation Research Part F 15, 1–8. Lerner, N., Singer, J., Huey, R., 2008. Driver Strategies for Engaging in Distracting Tasks Using In-vehicle Technologies (Report No. DOT HS 810 919). The National Highway Traffic Safety Administration, Oak Ridge, pp. 1–120. McEvoy, S.P., Stevenson, M.R., Woodward, M., 2007. The prevalence of, and factors associated with, serious crashes involving a distracting activity. Accident Analysis and Prevention 39 (3), 475–482. Mayhew, D.R., Simpson, H.M., Ferguson, S.A., 2006. Collisions involving senior drivers: high-risk conditions and locations. Traffic Injury Prevention 7, 117–124. Oxley, J., Fildes, B., Corben, B., Langford, J., 2006. Intersection design for older drivers. Transportation Research Part F 9, 335–346.

278

J.L. Charlton et al. / Accident Analysis and Prevention 58 (2013) 271–278

Pelli, D.G., Robson, J.G., Wilkins, A.J., 1988. The design of a new letter chart for measuring contrast sensitivity. Clinical Vision Science 2, 187–199. Preusser, D.F., Williams, A.F., Ferguson, S.A., Ulmer, R.G., Weinstein, H.B., 1998. Fatal crash risk for older drivers at intersections. Accident Analysis and Prevention 30, 151–159. Salthouse, T., 1991. A Theory of Cognitive Aging. Elsevier Press, Amsterdam. Sayer, J.R., Devonshire, J.M., Flannagan, C.A., 2005. The Effects of Secondary Tasks on Naturalistic Driving Performance (Report No. UMTRI-2005-29). The University of Michigan Transportation Research Institute, Ann Arbour, pp. 1–51. Smiley, A., 2004. Adaptive strategies of older drivers. In: Transportation in An Aging Society, Transportation Research Board Conference Proceedings 27, Washington, DC, USA. Snapper© Webbsoft Technologies, 2008. http://www.webbsoft.biz/prod snapper.php (accessed 08.05.12). Stamatiadis, N., Taylor, W., McKelvey, F., 1991. Elderly drivers and intersection accidents. Transportation Quarterly 45 (3), 559–571. Staplin, L., Lococo, K.H., McKnight, A.J., McKnight, A.S., Odenheimer, G.L., 1998. Intersection Negotiation Problems of Older Drivers; Volume II: Background Synthesis

on Age and Intersection Driving Difficulties. National Highway Traffic Safety Administration, Washington, DC. Staplin, L., Lococo, K., Byington, S., Harkey, D., 2001. Guidelines and Recommendations to Accommodate Older Drivers and Pedestrians. Federal Highway Administration, McClean. Stutts, J.C., Reinfurt, D.W., Rodgman, E.A., 2001. The role of driver distraction in crashes: an analysis of 1995–1999 Crashworthiness Data System Data. In: Annual Proceedings/Association for the Advancement of Automotive Medicine, pp. 45287–45301. Stutts, J., Feaganes, J., Rodgman, E., Hamlett, C., Reinfurt, D., Gish, K., Mercadante, M., Staplin, L., 2003. The causes and consequences of distraction in everyday driving. In: Annual Proceedings/Association for the Advancement of Automotive Medicine, vol. 47, pp. 235–251. Viano, D.C., Ridella, S., 1996. Significance of Intersection Crashes for Older Drivers SAE Technical Paper Series 960457. Society of Automotive Engineers, Warrendale, PA, pp. 115–121. Young, K.L., Lenne, M.G., 2010. Driver engagement in distracting activities and the strategies used to minimise risk. Safety Science 48, 326–332.