Introduction
Surveys show that while older Canadians make fewer trips and travel for different reasons than those still in the labour force, the personal automobile remains the primary mode of transportation (e.g., Newbold, Scott, Spinney, Kanaroglou, & Páez, Reference Newbold, Scott, Spinney, Kanaroglou and Páez2005). For many seniors, driving symbolizes independence and mobility (e.g., Dickerson et al., Reference Dickerson, Molnar, Eby, Adler, Bédard and Berg-Weger2007). Driving cessation can lead to depression, reduced out-of-home activity, social isolation, dependence on others (e.g., Johnson, Reference Johnson1999; Marottoli et al., Reference Marottoli, Mendes de Leon, Glass, Williams, Cooney and Berkman1997, Reference Marottoli, Mendes de Leon, Glass, Williams, Cooney and Berkman2000; Turcotte, Reference Turcotte2006), and possibly early mortality (Edwards, Perkins, Ross, & Reynolds, Reference Edwards, Perkins, Ross and Reynolds2009). Not surprisingly, older adults are often reluctant to stop driving (e.g., Johnson, Reference Johnson1999, Reference Johnson2002; Rudman, Friedland, Chipman, & Sciortino, Reference Rudman, Friedland, Chipman and Sciortino2006).
The concern is that age-related declines, together with development of medical conditions, can compromise driving safety. Beginning around age 70 and escalating thereafter, older drivers are disproportionately involved in fatal collisions when crash rates are adjusted for mileage (Dobbs, Reference Dobbs2008; Hopkins, Kilik, Day, Rows, & Tseng, Reference Hopkins, Kilik, Day, Rows and Tseng2004). Context must be considered because low-mileage drivers tend to drive more in congested urban areas, which pose a higher risk of collisions (Keall & Frith, Reference Keall and Frith2004; Langford, Methorst, & Hakimies-Blomqvist, Reference Langford, Methorst and Hakamies-Blomqvist2006). Time use surveys also indicate that Canadian seniors are driving more during peak periods (Scott et al., Reference Scott, Newbold, Spinney, Mercado, Páez and Kanaroglou2009). It is well-known that older drivers tend to be involved in different types of crashes than younger drivers, namely, those (a) involving multiple vehicles, (b) occurring during the day, (c) at lower speeds, (d) at intersections, and (e) when turning (e.g., Cooper, Reference Cooper1990; Preusser, Williams, Ferguson, Ulmer, & Weinstein, Reference Preusser, Williams, Ferguson, Ulmer and Weinstein1998; Ryan, Legge, & Rosman, Reference Ryan, Legge and Rosman1998; Zhang, Lindsay, Clarke, Robbins, & Mao, Reference Zhang, Lindsay, Clarke, Robbins and Mao2000).
Efforts to regulate older drivers through age-based licensing requirements are expensive and, except for in-person renewal for drivers aged 85 and older, have not shown any safety benefits (Grabowski, Campbell, & Morrisey, Reference Grabowski, Campbell and Morrisey2004; Langford & Koppel, Reference Langford and Koppel2006). Restricted licenses (e.g., daytime only) are being used in several provinces, although not in Ontario. A survey of Ottawa seniors found that restrictions were preferable to losing one’s license; however, restrictions that might limit seniors’ autonomy (e.g., being restricted to drive a maximum 10 km from home) were not highly endorsed (Marshall, Man-Son-Hing, Molnar, Wilson, & Blair, Reference Marshall, Man-Son-Hing, Molnar, Wilson and Blair2007). Restrictions are difficult to enforce and may not be necessary if seniors recognize their limitations and adapt their driving accordingly (e.g., Ball et al., Reference Ball, Owsley, Stalvey, Roenker, Sloane and Graves1998). Difficulty seeing is the most frequently reported reason for driving restrictions (Satariano, MacLeod, Cohn, & Ragland, Reference Satariano, MacLeod, Cohn and Ragland2004), which is not surprising as vision is impaired under low illumination and glare even in seniors with otherwise good eye health (e.g., Ball et al., Reference Ball, Owsley, Stalvey, Roenker, Sloane and Graves1998; Owens, Wood, & Owens, Reference Owens, Wood and Owens2007).
Drivers who lack insight due to cognitive impairment are unlikely to self-regulate or comply with licensing restrictions and must be carefully monitored (e.g., Eby & Molnar, Reference Eby and Molnar2009; Hopkins et al., Reference Hopkins, Kilik, Day, Rows and Tseng2004). However, many seniors are capable of regulating their driving and can control their risk either tactically (e.g., reducing speed or their following distance) or strategically (e.g., postponing a trip). Decisions at the strategic level in terms of how much to drive and under what conditions, together with lifestyle choices (such as where to live), may present the greatest opportunity for safe self-regulation (Eby & Molnar, Reference Eby and Molnar2009).
There is substantial evidence that older drivers adapt their driving with age. Compared to younger drivers, older adults tend to drive less often, closer to home, in the daytime, on weekdays, and in familiar areas (Collia, Sharp, & Giesbrecht, Reference Collia, Sharp and Giesbrecht2003; Keall & Frith, Reference Keall and Frith2004). Older drivers, particularly women, also tend to avoid driving at night, in bad weather, in rush hour, on highways, and making complex manoeuvres such as left-hand turns (Baldock, Mathias, McLean, & Berndt, Reference Baldock, Mathias, McLean and Berndt2006; Charlton et al., Reference Charlton, Oxley, Fildes, Oxley, Newstead and Koppel2006; Hakamies-Blomqvist & Wahlstrom, Reference Hakamies-Blomqvist and Wahlstrom1998; Kostyniuk & Molnar, Reference Kostyniuk and Molnar2008). Driver perceptions (particularly of confidence or comfort) may be a key determinant of self-regulatory practices (Baldock et al., Reference Baldock, Mathias, McLean and Berndt2006; Charlton et al., Reference Charlton, Oxley, Fildes, Oxley, Newstead and Koppel2006; MacDonald, Myers, & Blanchard, Reference MacDonald, Myers and Blanchard2008; Marottoli & Richardson, Reference Marottoli and Richardson1998; Molnar & Eby, Reference Molnar and Eby2008; Myers, Paradis, & Blanchard, Reference Myers, Paradis and Blanchard2008; Oxley, Charlton, Scully, & Koppel, Reference Oxley, Charlton, Scully and Koppel2010; Rudman et al., Reference Rudman, Friedland, Chipman and Sciortino2006). Estimates of the proportion of older drivers who restrict their driving, however, vary widely from study to study and until recently have been based almost exclusively on self-report data (i.e., questionnaires, diaries, and interviews).
Self-reported measures of driving exposure and patterns are subject to recall and social desirability bias (Lajunen & Summala, Reference Lajunen and Summala2003). There is now evidence that self-reported estimates of distance driven (km) are inaccurate (Blanchard, Myers, & Porter, Reference Blanchard, Myers and Porter2010; Huebner, Porter, & Marshall, Reference Huebner, Porter and Marshall2006); moreover, there are preliminary findings that show discrepancies between self-reported and objectively measured driving patterns (Blanchard & Myers, Reference Blanchard and Myers2010; Blanchard et al., Reference Blanchard, Myers and Porter2010).
Blanchard and colleagues were the first to examine naturalistic driving patterns of seniors using electronic devices, in relation to self-reported regulatory practices, driver characteristics, and perceptions. These findings need to be replicated. Additionally, driving was monitored only for a one-week period between June and October, when there was substantial daylight and generally good weather.
In northern climates, drivers must deal with winter conditions. Snow makes vehicle handling more difficult and reduces driver visibility (Andrey, Reference Andrey2010; Zhang et al., Reference Zhang, Lindsay, Clarke, Robbins and Mao2000), especially when combined with darkness which reduces preview time and ability to assess road conditions (Kilpelainen & Summala, Reference Kilpelainen and Summala2007). A Finnish study showed that on winter days with bad-weather forecasts, work-related trips were far more common than leisure trips. These researchers speculated that retirees have more freedom to postpone discretionary trips on bad days; however, their sample consisted of drivers (of various ages) who drove ≥ 20 km on the days of the survey and did not include those who chose not to drive on those days (Kilpelainen & Summala, Reference Kilpelainen and Summala2007).
Only one study to date has examined geographical and climatic influences on seniors’ driving behaviour. Compared to a Florida sample, older drivers in western New York reported driving less in the winter; over half said specifically that they avoided snow, sleet, or icy conditions (Sabback & Mann, Reference Sabback and Mann2005). However, these findings were based on interview data. The purposes of this study were twofold: (a) examine the influence of winter conditions on naturalistic driving exposure and patterns in older adults using objective measures; and (b) explore factors such as driver characteristics and perceptions that may be associated with self-restrictions.
Methods
Participants
To be eligible, participants had to be age 65 or older, hold a valid driving license, drive at least once a week, live in the Kitchener-Waterloo (K-W) region, and own and operate a non-hybrid vehicle, 1996 or newer. Older vehicles and those with alternating power sources are not compatible with the CarChip, one of the recording devices used in this study. Volunteers were recruited from seniors’ centres and lecture series, as well as from the Ministry of Transportation of Ontario’s (MTO) Senior Drivers Renewal Program (SDRP), required for drivers over age 80.
A total of 47 drivers, 24 men (51%) and 23 women (49%), ranging in age from 65 to 91 (M = 77.2, SD = 6.6), primarily urban/suburban dwellers (93.6%), took part in the study between late November 2008 and March 2009. The average age of men (M = 77.4, SD = 7.1) and women (M = 76.9, SD = 6.3) in the sample was comparable. Of the total, 62 per cent were living with a spouse, and seven couples enrolled in the study together. Three quarters (76.6%) had completed either college or university; only two people were still working. Most rated their health as excellent (29%) or good (69%), and their eyesight as better (37%) or the same (61%) as most people their age. Diagnosed health conditions that were frequently reported included (a) high blood pressure, high cholesterol or heart-related ailments (62%), (b) vision disorders (47%), (c) back problems (30%), (d) hearing problems (28%), and (e) arthritis, rheumatism, or osteoporosis (26%). With respect to vision disorders, three participants reported glaucoma, one reported macular degeneration, and 18 reported cataracts (all of whom had corrective surgery). Only 15.6 per cent used a cane or walker, and 93 per cent reportedly could walk a quarter of a mile.
The study sample had relatively few functional impairments (M = 1.8, SD = 1.2, range 0 to 4), on the basis of performance on several tests. These tests included the Roadwise Review® (Staplin & Dinh-Zarr, Reference Staplin and Dinh-Zarr2006) battery of computerized tests of driving-related abilities: the Rapid Paced Walk test, head/neck flexibility, high and low contrast visual acuity, the Motor-Free Visual Perception Test, subtest 2 of the Useful Field of View, the Trail Making Test, and delayed word recall.
All participants had substantial driving experience, ranging from 34 to 75 years. Reported driving problems over the past year were hitting curbs or medians, getting lost (16% each), and near misses (14%); only three people (6%) reported a collision. Continuing to drive was rated by most as either extremely (51%) or very (42%) important. The most common reasons for driving included the following: (a) shopping or errands (98%), (b) visiting (98%), (c) appointments (96%), (d) recreation or social events (96%), (e) volunteering (65%), and (f) religious services (56%). The majority (94%) preferred to drive themselves; six per cent (all women) preferred someone else to drive; 41 per cent said that others relied on them to drive. Women versus men (χ2 = 6.17, p = .01) and those over (versus under) age 80 (χ2 = 14.02, p < .001) were more likely to be the only driver in their household. Only 18 per cent had thought about reducing their driving; only one had thought about quitting (an 81-year-old woman who cited poor vision as the reason).
Measures
Self-Reported Driving Habits and Restrictions
We used a questionnaire to obtain information on usual driving habits (e.g., times of day, types of roadways) and preferences for getting around. The 14-item Situational Driving Frequency (SDF) and 20-item Situational Driving Avoidance (SDA) scales were used to assess self-reported driving restrictions. The SDF asks people to rate how often (from never to very often: 4–7 days a week) they drive in challenging situations (e.g., at night, on highways, in rush hour, making left-hand turns), while the SDA asks people to check particular situations they try to avoid, if possible (e.g., driving at night, in bad weather, on highways with three or more lanes and speed limits of 100 km or more). Scores on the SDF can range from 0 to 56, and on the SDA from 0 to 20, with higher scores indicating greater frequency and avoidance of challenging driving situations, respectively (MacDonald et al., Reference MacDonald, Myers and Blanchard2008). Both scales have good internal consistency and test-retest reliability (Blanchard & Myers, Reference Blanchard and Myers2010; MacDonald et al., Reference MacDonald, Myers and Blanchard2008).
Perceived Driving Comfort and Abilities
Perceived driving comfort was assessed using the 13-item Day (DCS-D) and 16-item Night (DCS-N) Driving Comfort Scales. Both scales were developed with older drivers, are hierarchical and have good person, item, and test-retest reliability (Myers et al., Reference Myers, Paradis and Blanchard2008). When rating their level of comfort (0%, 25%, 50%, 75%, 100%), respondents are asked to consider confidence in their own abilities and driving skills, as well as the situation itself, and to assume normal traffic flow. Scores on the DCS-D and DCS-N scales can range from 0 per cent to 100 per cent, with higher scores indicating higher levels of comfort (Myers et al., Reference Myers, Paradis and Blanchard2008).
Perceived driving abilities were assessed using the 15-item Perceived Driving Abilities (PDA) scale (4-point rating from poor to very good) and the 15-item PDA change scale comparing one’s abilities to 10 years ago (a 4-point rating from “a lot worse” to “better”). Higher scores (range 0 to 45) indicate more positive perceptions and fewer declines, respectively (MacDonald et al., Reference MacDonald, Myers and Blanchard2008). The scales have shown good item and person reliability (MacDonald et al., Reference MacDonald, Myers and Blanchard2008), and moderate test-retest reliability, although internal consistency was better for the current than for the change PDA scale (Blanchard & Myers, Reference Blanchard and Myers2010).
Objective Driving Data
Similar to the way that Blanchard and colleagues conducted their study (described in Blanchard et al., Reference Blanchard, Myers and Porter2010, and Blanchard & Myers, Reference Blanchard and Myers2010), in our study we assessed naturalistic driving using two devices: a CarChip® (Pro model; Davis Instruments, Hayward, California) and the Otto Driving Companion® (Otto; Persen Technologies Inc., Winnipeg, Manitoba). The CarChip plugs into the on-board diagnostic port, while the Otto (a small GPS device) is mounted on the dashboard to pick up satellite signals. Each device can record about 300 hours of driving data and logging begins automatically when the engine is turned on. Although both devices collect similar date- and time-stamped information, the CarChip is more accurate than GPS devices in recording distance (Huebner et al., Reference Huebner, Porter and Marshall2006). As the loss of satellite signals can result in missing GPS data, CarChip data were used for most of the driving variables (distance, duration, number of trips, and stops). The Otto data were used to examine roadways and radius or distance driven from home.
Procedure
Two visits were scheduled with each participant (all were agreeable to meeting at their homes). At the first visit, the researcher obtained consent, asked the person to complete a background questionnaire, and explained the trip logs (used to record who drove, the trip purpose and location, and general weather and road conditions). The CarChip and Otto were then installed in the person’s vehicle and a set of trip logs placed in the car. Subjects were asked to drive as usual over the subsequent two weeks. Drivers with multiple vehicles (28% of the sample) were asked to use only one for the study period. Couples enrolled in the study were permitted to use separate or shared vehicles; logs were used to identify the driver of each trip.
At the second (post-monitoring) visit, the researcher met with participants to collect the devices and trip logs. Subjects were asked to complete, in order, a general driving habits questionnaire (usual patterns) and the DCS, PDA, SDF, and SDA scales. Following an interview, participants were asked to complete the Roadwise Review. The researcher loaded the CD-ROM on her computer and controlled the mouse for consistency. As five of the eight tasks require a partner for cueing and scoring, the researcher provided this assistance. Couples were assessed separately.
Analysis
The perception (DCS and PDA) and self-regulation (SDF and SDA) scales were scored according to the developers’ instructions (MacDonald et al., Reference MacDonald, Myers and Blanchard2008; Myers et al., Reference Myers, Paradis and Blanchard2008). All but one person completed these scales. CarChip data were lost for one subject as the device was inadvertently removed during vehicle servicing. Otto data were lost for three people, while four other participants were missing a third or more of the data from their trips, mainly due to connectivity problems. No one was missing both CarChip and Otto data, and trip logs were returned by all 47 subjects.
CarChip (for 46 subjects) and Otto data (for 40 subjects) were cleaned prior to analysis (i.e., removing trips with 0.0 km and by drivers not in the study). To determine complete trips, segments were linked by cross-referencing CarChip data with logs and/or Otto data. Night driving was defined as the period of darkness (between sunrise and sunset) and included both complete and partial trips (i.e., those beginning or completed in darkness). Archives were consulted for daily times of sunrise and sunset (http://www.sunrisesunset.com), temperature, and weather conditions for the region (http://www.weatheroffice.gc.ca). Weather advisories and conditions (e.g., severe snowstorm, freezing rain) were also obtained from the local daily newspaper and the airport weather station. As the conditions that people observe may exert a greater influence on their driving (Kilpelainen & Summala, Reference Kilpelainen and Summala2007), trip log descriptions were weighed more heavily than regional reports when there were discrepancies (which occurred 22% of the time). Trip logs from other participants driving on the same days were also consulted.
The GPS data and digital maps were used to examine roadways driven, including freeways (divided, multi-lane, with speed limits ≥ 90 km/h) and highways (two-lane, with speed limits ≥ 70 km/h), as well as to calculate the average and maximum radii each person drove from his/her home for each trip. To examine the frequency of actual driving in challenging situations, vehicle data were used to examine 11 of the 14 situations depicted in the SDF scale. Reverse parking and driving in unfamiliar areas could not be examined, while participants had been told that speeding would not be examined. As the study had a two-week monitoring period, the first three response options on the SDF scale (never, less than once a month but not more than weekly, more than once a month but not weekly) were collapsed. Method error (ME) and coefficient of variation (CV; expressed as a percentage) were calculated to examine agreement between actual and self-rated frequency scores; CVs of ≤ 10% are generally considered acceptable (Holmbäck, Porter, Downham, & Lexell, Reference Holmbäck, Porter, Downham and Lexell1999).
Analyses were conducted with SPSS (version 17). Comparative statistics (parametric and non-parametric, as appropriate) were used to examine associations between driver perceptions, and self-reported and actual driving exposure and patterns. Sex and age group (< age 80 vs. age 80 and older) differences were examined throughout. Consistency of driving over the two weeks was examined using the intraclass correlation coefficient (ICC3,2) (Shrout & Fleiss, Reference Shrout and Fleiss1979).
Results
Self-Reported Driving Patterns
Half the sample (54%) said that they drove less now compared to 10 years ago; 44 per cent drove about the same amount, and 2 per cent said that they drove more. Overall, the sample reported driving marginally less (fewer days) in the winter than the summer (M = 5.13, SD = 1.5 vs. M = 5.70, SD = 1.38). The older group (age 80 and older) reportedly drove fewer days in the winter than the younger group (M = 4.53 vs. 5.50, z = –1.96, p < .05). The younger group was also more likely to change to snow tires in the winter (p < .01), although only one third of the sample reported this practice. Most said they drove on residential (98%) and city streets (96%), followed by highways (78%), freeways (76%), and rural roads (59%). Men were more likely to report freeway driving (p < .01). Most said they usually drove in the morning (96%), afternoon (96%), and evening (91%); however, only 67% (78% of men; 57% of women) typically drove at night.
Reported frequency of driving in challenging situations (SDF scores) was moderate overall (M = 33.5, SD = 6.5, out of 56), but significantly higher for men than women (M = 37.6, SD = 5.5 vs. M = 29.7, SD = 5.1, t = 5.03, p < .001). Reported avoidance of challenging driving situations was quite low overall (M = 6.3, SD = 4.1, out of 20); however, women had significantly higher total SDA scores than men (M = 8.5, SD = 3.9 vs. M = 4.2, SD = 3.0, t = –4.18, p < .001). When individual items were examined, women were significantly more likely to say they tried to avoid driving at night in general (χ2 = 10.27, p < .001) as well as at night in bad weather (χ2 = 7.22, p < .01). The older age group had lower SDF and higher SDA scores, although the differences were not significant. Scores on the SDF and SDA were inversely related (r = –.52, p < .001) and significantly associated with DCS and PDA scores in the expected directions. That is, driving comfort (day and night) and perceived abilities correlated positively with SDF scores (.44, .53, and .34) and negatively with SDA scores (–.65, –.66, and –.63), respectively.
Actual Driving over the Monitoring Period
Driving was fairly consistent from week to week: ICC values were over .70 for all the indicators, except for night km (.65), and radius (maximum .28 and average .07). Ten people did not drive at night their first week, while 15 did not drive in their second week. However, only five of the 46 (11%) did not drive at least once at night over the full two-week monitoring period. Following comparison of week one versus week two data, cumulative scores were calculated for each indicator (e.g., total distance over the full two weeks), then averaged to one week for comparison with prior studies. The results for the sample as a whole, as well as by sex and age group, are presented in Table 1.
Driving behavior by sex and age groupFootnote *
* Values denote Mean (SD; standard deviation) and range (averaged to one week). N = 46, except for radii (n = 40).
a Significant sex difference (p < .05)
b Significant sex difference (p < .01)
c Significant sex difference (p < .001)
Compared with women, men drove significantly more often (# days, t = –2.31, p < .05), making more trips (t = 3.40, p < .001) and stops (t = 3.47, p < .001). They also drove a greater distance (km) in general (t = –2.31, p < .001) and at night (t = –2.20, p < .05), for a longer duration (z = 4.96, p < .001), and further from home (average z = –2.70 and maximum radius z = –2.53, p < .01). No significant age group differences emerged.
Table 2 presents the scores on the driver perception measures, while Table 3 shows the associations with actual driving behaviour. Overall, the participants were quite comfortable driving during the day and at night in good weather and traffic conditions (DCS-N Item # 1), and DCS-N scores were significantly lower than DCS-D scores (t = 6.95, p < .001). Men had significantly higher comfort scores for both day (z = –2.35, p < .05) and night driving (t = 3.21, p < .01). Although men perceived more change in their driving abilities (t = –2.63, p < .05), they had better perceptions of their current driving abilities than women (approached significance), as did the younger versus older age group (t = –2.63, p < .05).
Perception scores by sex and age groupFootnote *
* Values denote Mean (SD; standard deviation) and range. DCS-D, DSC-N: Day and Night Driving Comfort Scales, respectively DCS-N #1: first item (driving at night in good weather and traffic conditions)
PDA: Perceived Driving Abilities: current (now) and change (compared to 10 years ago)
a Significant sex difference (p < .05)
b Significant sex difference (p < .01)
c Significant age group difference (p < .05)
Associations between perceptions and actual drivingFootnote *
* DCS-D, DSC-N: Day and Night Driving Comfort Scales, respectively
DCS-N #1: first item (driving at night in good weather and traffic conditions)
PDA: Perceived Driving Abilities Scale. All values are Pearson r or Spearman ρ.
N = 45, except for radii (n = 40).
a p < .05
b p < .01
As shown in Table 3, daytime driving comfort scores (DCS-D) were moderately related to total trips, duration, and night distance. Driving comfort at night (DCS-N scores) correlated more strongly with most of the indicators, as well as overall frequency of driving in challenging situations (frequency index). Driving at night in good weather and traffic conditions (item #1 on the DCS-N scale) was also related to multiple indicators. Perceived driving abilities (PDA scores), meanwhile, were significantly related to distance (km), radius, and the frequency index.
Scores on the frequency index were significantly higher than self-ratings on the corresponding 11-item SDF scale (M = 9.0, SD = 2.7 vs. M = 6.0, SD = 3.9, z = –4.18, p < .001). Only two people scored identically on both measures. The ME was 2.76, while the CV was 36 per cent. Forty people drove at least once at night over the two weeks, only 27.5 per cent of whom reported on the SDA scale that they usually try to avoid night driving. Mean distance (km) driven at night was significantly lower (t = –2.66, p < .01) for reported avoiders (M = 13.1, SD = 16.9) versus non-avoiders (M = 39.4, SD = 44.0).
Winter Conditions and Driving Behaviour
Table 4 shows the weather and road conditions for the period in which devices were installed in participant vehicles. Over one half of the monitoring period (53 of 94 days) had inclement weather in the form of either snow (on 41 days) or rain/freezing rain (on 12 days). The month of January had the most snow, number of cold days (temperatures ≥ –15º Celsius), and weather advisories issued for the region. Poor road conditions (snow-covered, icy/slushy, or wet), meanwhile, occurred on 63 days, or 67 per cent of the 94-day monitoring period.
Weather and road conditions over the monitoring periodFootnote *
* Days with no participant monitoring were excluded
Over the monitoring period, there were a total of 644 opportunities (14 days × 46 subjects with vehicle data) for participants to either drive or not drive. Table 5 shows the proportion of the sample that drove or did not drive on days with inclement (vs. favorable) weather, and poor (vs. clear/dry) road conditions. The sample as a whole was more likely to drive (69%) than not drive (31%) on days with bad weather (either snow or rain). Two thirds of the sample (67%) also drove on days in which weather advisories had been issued for the region. Similarly, the sample was more likely to drive (69%) than not drive (30%) when road conditions were poor. Compared to men, however, women were significantly less likely to have driven (than not driven) on days with bad weather (χ2 = 9.16, p < .01) and poor road conditions (χ2 = 10.83, p < .001). No significant age group differences emerged.
Instances of driving and not driving according to weather and road conditionsFootnote *
* Values are frequency (%) of driving by opportunities in good versus poor conditions
a Significant sex difference (p < .01)
b Significant sex difference (p < .001)
Amount of daylight and driving exposure were also examined by month of study participation. As Table 6 indicates, the average amount of daylight was considerably less for the first wave of participants (who did most of their driving in December), particularly compared to the last wave in March. Amount of night driving (distance) was significantly different across months (Kruskal-Wallis H = 8.29, p < .05). Pairwise comparisons showed significant differences between December versus March, as well as February versus March (p < .05). Although not significant, the total amount of driving (regardless of time of day) was the highest for the December participants and lowest for the January participants.
Average daylight and kilometers driven by month of study participationFootnote *
* Values for total distance and night driving shown are one week average
a Significant difference by month (p < .05)
Trip Purposes and Cancellations
Trip purposes, as described in participant logs, were categorized and examined by driving instances. Half the driving trips (327/644) were for shopping or errands, followed by social and entertainment (211/644 or 33%), helping others (97 instances, 15%), and active leisure (57 instances, 9%). Other trip purposes recorded less often included these: going to church, paid work or classes, volunteer work, visiting someone in hospitals or nursing homes, and attending funerals. Family and friend get-togethers (e.g., birthday parties) were included in the social/entertainment category. The sample of participants was more likely to make trips for social or entertainment purposes on days with good versus inclement weather (χ2 = 9.18, p < .01), and out-of-town trips on days with clear versus poor road conditions (χ2 = 5.70, p < .05). Participants were also less likely to drive for social/entertainment purposes than for other purposes (χ2 = 8.32, p < .01) on days with weather advisories.
The most common reasons given for trip postponement or cancellation were as follows: poor weather (85%), not feeling well (37%), scheduling conflicts (13%), emergencies (4%), and car problems (2%). However, two individuals (both sole drivers) said nothing would persuade them to cancel or postpone planned trips. In the final interview, participants were asked whether the devices affected their driving (all said no) and whether their driving was typical over the two-week period (80% said yes). Five people said they drove less than usual, while four said they drove more. About one third of the sample (37%) noted special events such as out-of-town trips. One fifth, meanwhile, said that they did not take a planned trip due to an event’s being cancelled, poor driving conditions, and other reasons. Notwithstanding, everyone said that they did most or all of their regularly scheduled activities over the monitoring period.
Discussion
Determining the extent to which older drivers self-restrict or comply with ministry-imposed restrictions, and ultimately the effectiveness of such practices on crash reduction, requires objective measures (Marshall, Wilson, et al., Reference Marshall, Man-Son-Hing, Molnar, Wilson and Blair2007). Self-estimates of exposure (km driven), even over a short recall period (one week), have been shown to be inaccurate (Blanchard et al., Reference Blanchard, Myers and Porter2010; Huebner et al., Reference Huebner, Porter and Marshall2006), which causes us to question the validity of other self-reported regulatory practices. Electronic devices can also provide objective data on when, where (using GPS data, paired with digital maps) and under what conditions (using weather archives) people drive. Importantly, these devices do not appear to affect driver behaviour, which is consistent with reports from prior samples (Blanchard et al., Reference Blanchard, Myers and Porter2010; Huebner et al., Reference Huebner, Porter and Marshall2006; Marshall, Wilson, et al., Reference Marshall, Man-Son-Hing, Molnar, Wilson and Blair2007).
Huebner et al. (Reference Huebner, Porter and Marshall2006) and Marshall, Wilson et al. (Reference Marshall, Man-Son-Hing, Molnar, Wilson and Blair2007) used similar devices to monitor one-week exposure of older drivers in Winnipeg and Ottawa, respectively; however, neither study examined driving patterns or weather conditions. The study by Blanchard et al. (reported in Blanchard & Myers, Reference Blanchard and Myers2010; Blanchard et al., Reference Blanchard, Myers and Porter2010) was the first to examine naturalistic driving patterns of older adults in relation to self-reported regulatory practices and perceptions. In their study, 61 drivers (mean age: 80) from Southwestern Ontario were monitored for one week between June and October 2007.
Participants in this study were recruited from the same area and assessed with the same protocol, albeit for a longer monitoring period. Except for radius and night km, driving patterns were found to be quite consistent from week to week. The extra week captured more instances of night driving and adverse weather conditions; however, trade-offs included more technical problems with devices, additional time for analysis, and increased subject burden (more trip logs to complete).
As participants were forewarned about taking their car in for servicing, CarChip data were lost for only one person. Problems with GPS devices, such as cold starts and signal loss, are more common (e.g., Blanchard & Myers, Reference Blanchard and Myers2010; Grengs, Wang, & Kostyniuk, Reference Grengs, Wang and Kostyniuk2008; Huebner et al., Reference Huebner, Porter and Marshall2006). Triangulation of multiple data sources (devices, trip logs, digital maps, and archives) is necessary to overcome the inherent limitations of each method and provide a more complete depiction of driving patterns, purposes, and context (daylight, weather, and road conditions).
To compare the findings to prior studies, driving data were averaged to one week. Total driving distance (157 ± 109 km) was similar to the sample in the Blanchard et al. (Reference Blanchard, Myers and Porter2010) study (164 ± 158) despite being monitored at different times of the year. Both sets of participants from Southwestern Ontario drove less than did Huebner et al.’s (Reference Huebner, Porter and Marshall2006) Winnipeg sample (340 ± 159) of 20 seniors (mean age 73), monitored in the summer using CarChips, and Marshall, Wilson, et al.’s (Reference Marshall, Man-Son-Hing, Molnar, Wilson and Blair2007) Ottawa sample (185 ± 82) of 20 seniors (mean age: 78) monitored in February, also using CarChips. Differences in exposure may be due in part from the higher proportion of men in the Winnipeg (70%) and Ottawa (75%) samples, compared to the proportion of men in the study by Blanchard et al. (41%) and in ours (51%).
We found that men drove significantly more than did women with respect to most of the driving indicators. In Blanchard’s study (gender and age comparisons are reported in Blanchard & Myers, Reference Blanchard and Myers2010), all values were higher for men but significant only for night km. Age group differences did not emerge in either study. Average results were similar for number of days driven, trips, duration, and radius; however, only 28 per cent of their sample drove at night, compared to 89 per cent of the sample in our study. Although two weeks captured more instances of night driving, this explained only 11 per cent of the increase we observed.
In our study, distance (km) driven at night varied significantly by month. Participants drove substantially more total and night km in December than in January, which may be explained in part by the busy holiday season. The weather was also much colder in January, with more days of snow (and total accumulation), weather advisories, and poor road conditions. Total distance increased in February and further in March as the weather improved. Conversely, night km decreased noticeably from February to March, when daylight savings time also took effect.
Although both samples cited bad weather as the main reason for postponing or cancelling trips, in the Blanchard et al. study (Reference Blanchard, Myers and Porter2010), inclement weather occurred on only 34 days or 23 per cent of their 148-day monitoring period. In contrast, over half of the 94-day winter monitoring period had inclement weather (either snow or rain/freezing rain), and 67 per cent of the period had poor road conditions. A key finding in this study was that women were significantly less likely than were men to drive on days with bad weather or poor road conditions.
Confidence may be a key factor in self-regulation (e.g., Rudman et al., Reference Rudman, Friedland, Chipman and Sciortino2006) and may also explain the gender effect often found in studies on older drivers’ self-regulatory practices (e.g., Blanchard & Myers, Reference Blanchard and Myers2010; Charlton et al., Reference Charlton, Oxley, Fildes, Oxley, Newstead and Koppel2006; Kostyniuk & Molnar, Reference Kostyniuk and Molnar2008). Prior to the development of the Driving Comfort Scales, however, this construct had not been reliably measured. Consistent with prior studies using these scales (Blanchard & Myers, Reference Blanchard and Myers2010; MacDonald et al., Reference MacDonald, Myers and Blanchard2008; Myers et al., Reference Myers, Paradis and Blanchard2008), perception scores (DCS-D, DCS-N, and PDA) were significantly correlated with each other and with scores on the SDF and SDA (self-reported regulatory practices) in the expected directions. Night driving comfort scores were also significantly lower than daytime driving comfort scores, as found in prior studies.
Night-time driving comfort scores overall and in good conditions (DCS-N item 1) were significantly higher in men and related to most indicators of driving exposure and patterns, supporting the findings of Blanchard and Myers (Reference Blanchard and Myers2010). Daytime comfort scores, also higher in men, were significantly related to night km in both studies, as well as total trips and duration in this study. Similarly, perceived driving abilities (PDA scores) were significantly associated with distance, maximum radius, and frequency of driving in challenging situations.
Over one half of both samples reportedly reduced their driving over the past 10 years. Changes in driving patterns, however, may be due to purposeful avoidance, changes in lifestyle (e.g., more flexible schedules, less need to drive), or simply common sense. Consistent with the findings by Blanchard et al. (Reference Blanchard, Myers and Porter2010), people in our sample drove in challenging situations more often than they self-reported. Circumstances (such as being the only driver in the household) and commitments may dictate when and where people drive, even though they may not feel totally comfortable and would prefer not to drive in certain situations (Baldock et al., Reference Baldock, Mathias, McLean and Berndt2006; Blanchard et al., Reference Blanchard, Myers and Porter2010).
In countries such as Canada and Finland, people may simply become accustomed to winter driving, confident that if authorities keep roads open and maintained, they are safe to drive (Kilpelainen & Summala, Reference Kilpelainen and Summala2007). The fact that our sample reportedly drove only marginally less (fewer days) in the winter than the summer, and relatively few (30%) changed to snow tires, suggests that they were not overly concerned with winter driving. As a whole, the sample was more likely to drive (than not drive), regardless if conditions were good or poor, which may not be representative of the behaviour of older drivers in general.
Study Limitations
Older adults who volunteer for driving studies tend to be well educated, generally healthy, and lead active lifestyles (e.g., Baldock et al., Reference Baldock, Mathias, McLean and Berndt2006; Charlton et al., Reference Charlton, Oxley, Fildes, Oxley, Newstead and Koppel2006; Rudman et al., Reference Rudman, Friedland, Chipman and Sciortino2006). Scores on the Roadwise Review showed little evidence of impaired driving-related abilities. Moreover, 40 per cent had completed Ontario’s SDRP, required every two years for drivers aged 80 and older to retain their license. Thus, generalization is limited to a convenience sample of primarily urban-dwelling, healthy, and active older drivers from one region of Southwestern Ontario. Although two thirds of Canadians live in urban areas (Andrey, Reference Andrey2010), rural drivers require special attention as they may be even more reliant on their vehicles (e.g., Johnson, Reference Johnson2002). Seniors living in northern areas with more extreme weather may restrict their driving more; some may even put their cars away for the winter.
Limitations of the weather and road data must also be considered. Descriptions derived from participant trip logs did not always correspond with Environment Canada reports, which was expected because regional forecasts may differ from local conditions. Although advisories were taken from the local newspaper, participants were not asked whether they accessed traffic-related weather information (from television, radio, or newspapers) prior to their trips or en route (via radio). Nor did we examine the additive effects of poor weather, road conditions, and time of day. Kilpelainen and Summala (Reference Kilpelainen and Summala2007) found that drivers were more likely to rate conditions better than the forecast in daylight and worse at night. In Southwestern Ontario, conditions are highly variable and frequently change throughout the day. It is possible that people may have waited to go out until conditions improved, or shortened their trips, or altered their routes on bad days.
Although the study employed a longer monitoring period than prior studies did with older drivers, two weeks provides only a snapshot of driving patterns. Monthly comparisons were cross-sectional and based on relatively small groups of 10–13 participants who were sequentially enrolled in the study. Larger samples are also required to examine the relative influence of driver perceptions, characteristics (e.g., gender, rural versus urban, availability of other household drivers), and other factors (such as geographical differences) on naturalistic driving practices.
Conclusions
The findings support the assertion by Blanchard et al. (Reference Blanchard, Myers and Porter2010) that older drivers may not regulate as much as they say they do on questionnaires, warranting the continued use of objective measures. Although total driving distance was similar, the present sample drove substantially more at night showing that seasonal factors (particularly the amount of daylight) are important in studying self-regulatory practices. Older drivers were more likely to make trips for social/entertainment purposes on days with no precipitation or weather advisories and out-of-town trips on days with dry road conditions, supporting the notion that retirees may postpone or cancel discretionary trips on bad winter days (Kilpelainen & Summala, Reference Kilpelainen and Summala2007; Zhang et al., Reference Zhang, Lindsay, Clarke, Robbins and Mao2000). Future studies need to consider the severity of conditions (e.g., amount of precipitation, wind velocity) and acquisition of weather and traffic information, in conjunction with trip purposes, to better understand strategic decisions made by older drivers. Prospective studies, such as the five-year Candrive cohort study are required to determine temporality concerning the relationships between driver perceptions, abilities/skills, naturalistic driving practices, and crash outcomes.
