- Email: [email protected]
Perception of the risk of an accident by young and older drivers
Accid. Anal. & Prev. Vol. Printed in Great Britain.
18, No. 4. pp. 289-298.
OcQl-4575186 $3.00 + .cm 0 1986 Pergamon Journals Ltd.
1986
PERCEPTION OF THE RISK OF AN ACCIDENT YOUNG AND OLDER DRIVERS
BY
PETERFINN Abt Associates Inc., 55 Wheeler Street, Cambridge, MA 02138, U.S.A. and BARRY W. E. BRAGG Program Evaluation Branch, Department of Communications, Ontario KlA OC8, Canada
300 Slater Street, Ottawa,
Abstract-Young drivers are significantly overrepresented among all drivers involved in traffic accidents and fatalities. Excessive risk taking by young drivers appears to be largely responsible for this disproportionate involvement. This excessive risk taking could be due to (1) being more willing to take risks than older drivers are, (2) failing to perceive hazardous situations as being as dangerous as older drivers do or (3) both causes. This paper reports the results of a study which attempted to determine whether misperception of risk could be an explanation for the high rates of traffic accidents among youth by testing whether young drivers perceive driving to be less hazardous than do older drivers. Three different methods of estimating the risk of accident involvement were used to compare risk estimates of young and older drivers. The methods included general questions about accident involvement, rating the riskiness of ten specific driving situations illustrated in still photographs, and rating the riskiness of fifteen videotaped driving situations. Young drivers perceived their own chances of an accident to be significantly lower than those of both their peers and older male drivers, while older male drivers saw their chances of accident involvement as comparable to those of their male peers and less than those of young male drivers. These findings lend support to the thesis that young male drivers are overrepresented in traffic accidents at least in part because they fail to perceive specific driving situations as being as risky as older drivers perceive them. INTRODUCTION
Young drivers are significantly overrepresented among all drivers involved in traffic accidents and fatalities. In 1979, drivers under the age of 25 constituted 22% of the registered driver population and nearly 37% of fatal accident-involved drivers (U.S. Department of Transportation, 1981). Furthermore, young drivers are much more likely than older drivers to be responsible for the crashes in which they are involved (Williams and Karpf, 1984). In reviewing the research on young driver accident involvement, Hodgdon, Bragg and Finn (1981) found that this overrepresentation in collisions could not be explained by more miles driven, less driving experience or poorer mechanical condition of vehicles driven by young drivers as compared with older drivers. What did account for greater collision involvement by young drivers was the manner in which they drove. Young drivers speed more frequently, drive through yellow lights more frequently, wear safety belts less frequently and accept shorter temporal gaps when entering traffic-in other words, young drivers take more risks than older drivers. The excessive risk taking by young drivers could be due to (1) being more willing than older drivers to take risks, (2) failing to see hazardous situations as being as dangerous as often as older drivers do or (3) both. The ultimate design of a countermeasure program might be very different depending on the extent to which risk perception, as well as risk choice, is a determinant of young driver behind-the-wheel behavior. This paper reports the results of a study which attempted to determine whether misperception of risk could be an explanation for the high rates of traffic accidents among youth by testing whether young drivers perceive driving to be less hazardous than do older drivers. If the results of the study showed that young drivers perceive driving to be less risky than do older drivers, strategies could be developed to alter risk perceptions along with attempts to influence young drivers to make less risky driving decisions. The main research hypothesis which the experiment was designed to test was that male drivers aged 18-24 would perceive less risk associated with a given driving situation
290
P. FINN and B. W. E. BRAGG
than would older male drivers. In the present study, we concentrated exclusively on the influence of age and, by proxy, driving experience as potentially direct influences on the risk perception of driving situations. While some researchers (e.g. Slavic, Fischoff and Lichtenstein, 1979) have found that people have difficulty estimating the frequency with which low probability, catastrophic events may occur, others (e.g. Bragg and Cousins, 1976; Cousins and Bragg, 1977; Cousins, 1980) have indicated that subjects can make meaningful and reliable estimates of some low probability events such as the likelihood of arrest for impaired driving. STUDY
METHODOLOGY
Risk perception measurement approaches
We developed three off-road approaches for obtaining the subjects’ estimates of the risk of accident involvement. The first approach involved a 33-question closed-ended questionnaire. Several questions sought driving-related information about the subjects. One series of questions focused on the subjects’ estimates of the number of people injured and killed in traffic accidents in Massachusetts and the nation. Another series asked subjects to estimate the percentage of three categories of male drivers who would get into a traffic accident next year-drivers aged 18-24, drivers aged 38-50, and the respondents themselves. Certain questions were included because they tapped the subjects’ perception of driving risk in a global sense (e.g. “How many people were killed in traffic accidents in Massachusetts last year?“), while other questions focused on the subjects’ perceptions of driving risks related to two specific risk-taking behaviours (failure to wear safety belts, drinking before driving). Three sets of questions were designed to elicit the subjects’ comparative estimates of the likelihood of traffic accidents for male drivers their age and their father’s age for use later in comparing the subjects’ comparative estimates of accident risk in response to photographed and videotaped driving scenes. In the second risk perception measurement approach, each subject was shown ten still photographs, with descriptive captions, of different driving situations and asked to indicate how much greater or less the risk of an accident would be for male drivers his own age in each situation compared to the risk for his peers driving during the daytime on normal suburban road conditions illustrated in an eleventh (baseline) photograph. Subjects repeated this assignment for male drivers in the other driver age group and then repeated the rating a third time for themselves. The following driving situations were represented in the photographs and described in the accompanying captions: (1) driving on bald tires; (2) nighttime driving; (3) driving on an undivided two-lane rural road; (4) urban driving; (5) driving on an eight-lane divided highway at 45 mph when other traffic is travelling at 55 mph; (6) driving at 65 mph on an eight-lane divided highway when other traffic is travelling at 55 mph; (7) driving on wet roads when it is raining; (8) following one-and-one-half car lengths behind another car while travelling at 30 mph; (9) driving on snow and ice covered roads and (10) driving after consuming six 12-ounce cans of beer within one hour. The third risk perception test was to view a videotape of fifteen driving situations and estimate the relative likelihood of an accident in each compared to the chances of an accident in a baseline videotaped sequence. Each sequence was designed to give the viewer an overall view of the roadway and traffic conditions and then a driver’s eye view of the driving situation. The driving situations included, for example, a vehicle driving on a divided six-lane highway with no traffic present (baseline); a vehicle entering a freeway causing another vehicle to change lanes to avoid a collision; a vehicle at a suburban traffic-free intersection; a vehicle running a stop sign causing a second vehicle to swerve to avoid a collision; and a pedestrian suddenly emerging in front of a vehicle. The subject was instructed to rate the risk of an accident for each scenario using the same relative magnitude estimation method he used for the photographs. Subjects
The major focus of the study was to compare the risk perceptions of young male drivers with those of older male drivers. Female drivers were excluded from the study
Perception of accident risk by drivers
291
because the high accident rate for young drivers is primarily a high accident rate for young male drivers (Hodgdon, Bragg and Finn, 1981). For example, 18.5% of 19-yearold male drivers have an accident in one year, whereas only 5.9% of 19-year-old female drivers have an accident in one year (National Accident Sampling System, 1981). Ideally, we should have compared a representative sample of young male drivers and older male drivers. To obtain a sample of sufficient size to be nationally or even regionally representative, however, would have been prohibitively expensive and administratively impractical. Instead, young and older male drivers were recruited through a newspaper advertisement in the major Boston newspaper. If this recruitment procedure resulted in any bias it should have produced a more homogeneous group of participants thus making it more difficult to detect age differences which might exist in the population. However, we minimized young driver/older driver sampling bias by controlling demographic effects. A comparison between the young and older drivers in the sample indicates that they are similar in the number of miles they drive, their frequency of accidents and violations, their use of safety belts, and the percentage of their driving done in the city. The young drivers differ from the older drivers in the distribution of violations, with higher variability among the young drivers (suggesting that a smaller proportion of young drivers are responsible for more of the violations), and in the proportion of driving done at night, with young drivers doing a significantly greater proportion of their driving at night. Subjects were first administered the questionnaire, they then rated the still photographs and they then concluded with the videotape ratings. Ninety-three subjects were tested, 45 subjects between the ages of 18-24, and 48 subjects between the ages of 3850. RESULTS
The findings from the study are reported separately below for the interview questions, still photograph ratings and videotape ratings. Results of interview questions on perception of risk
Previous research on perception of risk (Berger and Pensinger, 1980) has shown that young drivers tend to see driving as being more hazardous than do older drivers, insofar as young drivers provide higher estimates of accident and injury likelihood. We attempted to replicate these findings by asking the subjects to estimate the number of traffic fatalities and injuries per year in the United States and in Massachusetts (their own state). The results are presented in Table 1. As may be seen, no differences exist between young and older subjects in the average estimates of the number of persons killed in traffic accidents in the United States and the number of persons injured in traffic accidents in Table 1. Estimates by young and older male drivers of the number of persons injured and killed in traffic accidents in Massachusetts and the United States mom
Driver
loution
Vai&le -Injured
Killed
EZFZ Young Older
45 48
1g9,ooo 203,000
.95
U.S.
Young Older
45 40
7,019,000 2,096,OOO
.9g
Mass.
Young Older
45 48
5,483 1,769
2.44
U.S.
Young Older
45 40
227,000 516,000
-.66
prepared
January
3
1980
Federal
Annual Hiahway Traffic Safety Trend Re for the National Highway Traffic Safety 1982).
Fatal Accident Reporting Administration (Washington,
Highway
Aetu1
t
Mass.
1 Fatal Injury and Accident Ratea. (Washington, D.C.: June 1982). 2 First
lmiutr
!!_
(19a) 67,760l
3,260,0002
8811
51,0773
Administration
rt, Abt Associates Inc., Administration (Cambridge
System, the National Highway Traffic Safety D.C.: 1981). Fatalities as of August 1980.
MA:
P. FINN and B. W. E. BRAGG
292
the United States and in Massachusetts. Young drivers did estimate that significantly more persons were killed in traffic accidents in Massachusetts (X = 5483 per year) than did older drivers (x = 1769). In fact, 881 persons died in traffic accidents in Massachusetts in 1981, the year the study was conducted (U.S. Department of Transportation, 1982). The older drivers were, therefore, more accurate in their lower estimate. In general, it appears that all subjects had very inaccurate perceptions of the actual magnitude of the traffic accident problem. Subjects overestimated by almost three times the frequency of injuries in Massachusetts and then either underestimated by one half (older drivers) or overestimated by over two times (younger drivers) the frequency of injuries in United States. Subjects overestimated the traffic fatalities in Massachusetts by two to six times, and they overestimated the actual rate of fatalities for the United States on average by almost four to eight times. It appears that Massachusetts was seen as a relatively more hazardous location in which to drive than is the United States as a whole. In fact, in 1981, the fatality rate per 100 million miles of travel was lower in Massachusetts (2.7) than for the United States as a whole (3.39) (Federal Highway Administration, 1982). In addition, there was significantly greater variability in the estimates of young drivers of the number of persons killed in Massachusetts (F = 10.54, df = 44,47, p < .OOl). Conversely, older drivers had significantly greater variability in their estimates of the number of persons injured in Massachusetts (F = 7.56, df = 44,47,p < .OOl), the number of persons injured in the United States (F = 2.64, df = 44,47, p < .OOl) and the number of persons killed in the United States (F = 38.96, @ = 44,47, p < .OOl). We had predicted that young drivers would perceive less risk in driving than would older drivers. These data only partially support that hypothesis, with young subjects estimating fewer fatalities in Massachusetts than older drivers. We did not expect, however, that older drivers would be more variable than young drivers in their estimatesas we found in three of the four estimates. Respondents were also asked to estimate the proportion of all drivers who were involved in an accident in the previous year and to estimate the chances of 18-24 year old males, 38-50 year old males and themselves being involved in an accident in the next 12 months. The results are shown in Table 2. As can be seen in Table 2, 18-24 year old males were judged by both young and older drivers to be significantly more likely to be involved in an accident than all other categories of drivers. Male drivers aged 38-50 were judged significantly more likely to be involved in an accident by young drivers but not by older drivers. Of most relevance, however, is that young drivers estimated their own chances of accident involvement (21.4%) as significantly lower than the chances of the average 18-24 year old male (37%), while older drivers, aged 38-50, judged their own chances of accident involvement (13%) to be comparable to those of other male drivers their age (14.8%). Thus, while young male drivers recognized that their age group and sex is at greater risk of accident involvement than are older male drivers they perceived their own risk to be significantly
Table 2. Young and older male drivers’ estimates of average percentage of drivers involved in a traffic accident’ in the previous year2 18-24 ,ar old
male*
All Llrioaa
Bate%
38-50
year old --males
Self
Young drivers
29.1,
37.Od
21.1b
21.4b
Older
26.4,
36.64
14.aa
13.08
15.1
6.8
drivers
Actual (Police
6.8
1 An accident was personal injury.
’
Any
different
NA
Reported)3
fwa
defined
means not (c - 1.96,
3 Police-reported
as
a collision
containing the df
-
accidents
91,
involving
same
subscript
at are
least
$100
significantly
pt.05).
with
$400
damane
or
personal
injury.
damage
or
Perception of accident risk by drivers
293
lower than that of their mule peers. By contrast, older male drivers saw their own chances
of accident involvement as comparable to those of their male peers. In summary, generalized questions about the number of persons killed or injured in a jurisdiction produce highly erroneous estimates. Estimates of the percentage of drivers involved in an accident per annum produce less erroneous overestimates. Young drivers see themselves as less likely to be accident involved than their peers while older drivers see their own chances of accident involvement as similar to that of their peers. Results of risk estimates of general driving situations (in still photographs) Subjects provided estimates of the risks of general driving situations by providing magnitude estimates of the risk of an accident for their peers, the other age group (young or older drivers), and themselves in ten driving situations depicted in photographs. Subjects were asked to estimate how much greater or less the risk of an accident was in each situation compared with a standard condition depicted in an eleventh photograph that showed a vehicle described as travelling at the speed of other traffic on a divided sixlane highway, in light traffic, in dry weather, during daylight. All judgments were made relative to this baseline. The data were subjected to a 2 transformation where all the magnitude estimates for a given subject (on all three target groups-own age group, other age group and self-for all ten driving situations) were used as the set for the transformation. By using a Z transformation each subject’s scores are converted to a common normalized scale. This eliminates the possibility of exploring age differences in the use of scales-an interesting but academic issue-but allows for the comparison of different age groups’ perceptions of different target groups using the same scale. A three factor analysis of variance with one random factor (age) and two repeated fixed effects (target group and driving situation) was conducted. Driving situation is nested within target group. The results of the analysis of the Z transformed scores are shown in Table 3. These results show a significant effect for target (2) (i.e. the group being evaluated); a significant target by age-of-subject (TA) interaction; a significant driving situation (P) main effect; a significant target group by driving situation (7’P) interaction; and a significant age-of-subject by target-group by driving-situation (WA) interaction. Table 4 shows the means for the age of subject by target group interaction. (The lower the mean, the lower the magnitude of the risk of an accident.) The data indicate that, over all driving situations, young drivers saw their chances of an accident us signif-
Table 3. Analysis of variance of Z transformed magnitude estimates of risk of accident involvement for three groups rating ten driving situations, by young and older drivers
Meall
0.00003
1
0.00003
0.54
0.4658
A (driver group)
0.00011
1
0.00011
2.40
0.2150
Error
0.00431
91
0.00005
T (target
group)
TA Error
P (driving
35.95768
2
17.97884 3.97018
7.94037
2
153.68895
182
0.84444
1237.83375
9
137.53708
21.29 4.70
136.06
0.0000 0.0102
0.0000
sitIl.¶tilXl) PA Error
7.733853 827.88493
9
819
0.81539
0.81
0.6102
1.01085
TP
8.05914
18
0.44773
1.79
0.0217
TPA
7.73700
18
0.42983
1.72
0.0305
409.85664
1638
0.25022
Error
294
P.
FINN
and B. W. E.
BRAGG
Table 4. Mean Z transformed ratings’ of risk of an accident for young male drivers, older male drivers and self as estimated by young and older male drivers* Target Drivers
Young
Older
Drivers
-Self
Judge
Young
Drivers
.038b
Older
Drivers
.22,
1 2
High
score
Any means
= high in
risk
a row
significantly
not
different
0.69,
-.09,
-0.41,
_.17b
rating containing (t
=
the 1.96,
same
df
= 91,
subscript
are
p (.05).
icantly less than those of both similar age other male drivers and older male drivers. Older drivers saw the risk of an accident as significantly greater for young male drivers than for older male drivers and saw their own chances of an accident as significantly lower than those of both similar aged male drivers and young male drivers. When the ratings are examined for specific driving conditions (the age-of-subject by target-group by drivingsituation interaction), the differentiations of young drivers become apparent (see Table 5). Young drivers saw themselves as less at risk of an accident than their own peer group in two of the ten situations: driving on wet roads and driving after consuming six beers within one hour. In no situation did young drivers see themselves at greater risk of an accident than their peer group. Young drivers saw themselves as less at risk of an accident than older drivers in five driving situations: tailgating (driving one-and-a-half car lengths behind another vehicle at 30 mph), driving at night, speeding, driving on snow-covered roads and driving after consuming six beers within one hour. In none of the ten driving
Table 5. Mean Z transformed ratings1 of risk of accident involvement for young male drivers, older male drivers and self by young and older male drivers rating ten driving situations Group
situation I.
2.
3.
4.
5.
6.
Urban
Rural
driving
driving
Bald
tires
Tailgating
Slow
Driver
Night
Driver
7.
Wet
8.
Speeding
9.
10.
road
covered
Young Male
Target Older Hale
B
Drivers
Driven
-self
Young Older
-.43b -.34b
-.Slab -.57b
-.54ab -.71a
Young
-.71a
-.69a
-.76a
Older
-.32b
-.60a
-.73a
Young
-.46ab
.47ab
Older
-.86a
-.58b
-.35a -.47ab
Young
-.35ab
-.28a
-.43b
Older
-.l&
-.32ab
-.53b
Young
-.50ab
-.5&b
-.55ab
Older
-.49a
-.65ab
-.71b
Young Older
-. 19b .05a
.13a -.15b
-.30bc -.42c
Young Older
.13a .20a
.Olab -.OOb
-.13bc -.24c
-. 25bc -.47a
-.37ab -.50a
-.46a -.14c
Young Older
50bc
.69c
.39b .05a
road
Young Older
.62c
Drinking
Young
1.74cd
1.7oc
1.45ab
Older
1.94d
1.31a
1.59bc
Snow
I Any means within a given subscript are significantly
driving different
situation (t =
which do 1.98, df -
.46b
not have the 1638, P (.05).
same
Perception of accident risk by drivers
295
situations did young drivers see their own risk of an accident as greater than that of older drivers. Young drivers saw their peer group of young male drivers as significantly less likely to be involved in an accident than experienced drivers in two driving situations: night driving and speeding. Once again, the young drivers saw their own chances of accident involvement as less than those of their peer group and particularly less than the chances of an accident for older drivers. Older drivers saw their own risk of accident involvement as signficantly lower than that of their same age male peer group on three of the ten driving situations: night driving, driving on wet roads and driving on snow-covered roads. They saw their own chances of an accident as significantly higher than those of their same aged male peer group in one situation: drinking and driving. Older drivers saw young drivers as significantly more likely to be involved in an accident than older drivers in eight of the ten driving situations. Only driving 10 mph slower than other traffic and tailgating were seen as situations of comparable risk for young and older drivers. In summary, the photo rating data support and refine the finding that young drivers see their accident likelihood as lower than that of their peers. Five of ten driving situations are seen by young drivers as having a lower risk of an accident for themselves than for their peers. However, the photo rating data also identified three situations where older drivers saw their accident likelihood as lower than that of their peers, and one where the risk was seen as higher. Magnitude estimates of videotaped driving sequences
Each subject was shown a specially filmed videotape of a series of 15 driving situations. The sequences were randomly ordered with relation to the degree of risk of an accident. Each subject was asked to make magnitude estimates of the relative risk of an accident in each situation compared with a baseline sequence shown at the beginning of the videotape. The baseline comparison sequence showed a vehicle driving on a six-lane divided highway with no other traffic present. A correlation of the average risk estimate indicates high agreement (r = .85, df = 13, p < .Ol) in the average risk rating between the young drivers and the older age group. However, significant differences in the variances suggested that the data needed to be transformed in order to determine more precisely if there was any difference in the average ratings. A Z transformation on each subject’s video ratings was used, and the analysis of the transformed ratings revealed a significant age by driving situation interaction (see Table 6). Analysis of the means reveals that only two of the driving situations were rated significantly differently by young and older drivers. The tailgating sequence was rated significantly more dangerous by older drivers than by young drivers (t = 3.51, df = 91, p < .OOl), and the pedestrian sequence was rated significantly more hazardous by young drivers (t = 2.90, df = 91, p < .OOl) than by older drivers. These findings are not surprising. The tailgating sequence involves risk in which the skill of the tailgating driver is crucial to avoid an accident. Skill is more likely to be Table 6. Analysis of variance of Z transformed magnitude estimates of risk of accident involvement in 15 driving situations depicted on videotape as rated by young an older male drivers SW of gquarel,
Source MeaIl A (driver EKI-OK S (driving
group)
Degrees Freedom
of
hkan squarea
Tail RObability
p 55.29 .Ol
0.0000 0.9226
39.74756
81.23
0.0000
0.96372 0.48935
1.97
0.0170
0.04108 0.00000 0.00678
1 1 91
0.04108 0.00000 0.00007
556.46587
14
13.49210 623.42589
14 1274
SitLdiOll)
SA
ErrOr
296
P. FINNand B. W. E. BRAGG
assumed by the young driver. The pedestrian sequence involves the sudden emergence of a pedestrian in front of the driver. Here the unexpected danger and lack of control is likely to result in young drivers rating the situation as more dangerous whereas older drivers have, through years of exposure, learned that such unexpected actions occur quite frequently without usually resulting in an accident. In summary, the videotape data add support to the driving situation specificity found in the photo rating data. In the videotape data one sequence (pedestrian) is seen as more risky by young drivers, while another sequence (tailgating) is seen as more risky by older drivers. The underlying dimension that appears to explain this difference is a skill-control dimension with situations requiring skill and control seen as less risky by young drivers and a situation where skill and control would not reduce the risk seen as more risky by young drivers.
DISCUSSION
The purpose of this study was to determine whether young male drivers perceive driving to be less hazardous than do older male drivers. If they do, this finding would lend support to the hypothesis that the disproportionate involvement of young drivers in accidents may be due not merely or even primarily to a willingness to take more risks than older drivers but to a failure to see hazardous situations as being risky and, hence, to see the need to exercise caution. One constraint to the generalizability of these findings is the lack of representativeness of the samples of young and older drivers. As we noted in the study methodology, advertizing for subjects probably led to more homogeneous samples than the populations. This is reflected in the lack of a difference in the accident rate between young and older drivers. This homogeneity makes the observed differences recorded in this study even more important in that the differences are likely to be larger in the populations. The importance of the differences reported in this study are further strengthened by the findings of Matthews and Moran (1986) reported elsewhere in this issue. Matthews and Moran (1986) found similar differences using a slightly different set of stimuli and a sample that was homogeneous on a different dimension (education). The study suggests that young drivers do perceive the risk of an accident significantly differently than do older drivers-and in the direction of perceiving less risk than do older drivers. Both young and older drivers were asked to estimate the probability of an accident to all drivers, young drivers, older drivers and themselves. Both groups saw young male drivers as having a greater risk of an accident than all drivers, who in turn had a greater risk of an accident than older drivers. However, young drivers saw their own chances of accident involvement as significantly lower than those of other young male drivers, while older drivers saw their own chances of accident involvement as comparable to those of their peers. In addition, and in contrast to previous research (e.g. Berger and Persinger, 1980), young drivers in the sample for the most part did not see driving as more dangerous than did the older drivers. No differences were found between young and other subjects in the average estimates of the number of persons killed in traffic accidents in the United States and the number of persons injured in traffic accidents in the United States and Massachusetts. Only in terms of the number of fatalities did the young drivers provide higher estimates than older drivers. Although we found that young drivers see driving as more dangerous when these general questions about the risk of an accident were asked, and they recognize that their age group is at greater risk of accident involvement than are drivers 38-50 years old, young drivers see their own chances of an accident as lower than those of their own age group or older drivers when specific questions about their own risk are asked. This discrepancy between general risk rating and specific risk rating may help to explain why young drivers are able to see driving in general as more dangerous than do older drivers and yet still take more risks than older drivers.
Perception of accident risk by drivers
297
of the subjects’ ratings of the videotaped sequences showed that the older drivers rated one situation (tailgating) as significantly more dangerous than did the young drivers, but the surprise pedestrian situation was rated significantly more hazardous by the young drivers. In terms of on-road behaviour, however, the tailgating discrepancy between the groups is likely to reflect frequent hazardous driving on the part of young drivers, while the pedestrian discrepancy is not likely to lead to safer driving by young drivers. Avoiding collisions with pedestrians who have yet to be spotted requires a consistently safe driving approach that anticipates danger-an approach young drivers appear to lack-while failing to see the risk of tailgating can lead to constantly placing oneself (and others) in jeopardy by failing to allow a sufficient gap between cars. Thus, we would expect to find the different ratings between the two groups for tailgating and surprise pedestrians to be reflected in an overall higher rate of accidents among young drivers compared with older drivers. Thus, all three of our risk estimation measurement approaches supported our hypothesis that young male drivers would perceive less risk in driving situations than older male drivers would. The findings reported here are based, as noted, on off-road tests of subject risk perception. However, another part of the study, reported elsewhere (Bragg and Finn, 1985), tested the same subjects’ on-road perception of driving risks in fifteen locations along a predetermined four-mile course in normal traffic. Findings from the on-road part of the study confirm the findings reported here that the young drivers saw less risk in specific driving situations than did the older drivers. The on-road findings showed that young drivers, but not older drivers, perceived less risk as they became familiar with a traffic location-that is, after they had driven the test route a second time. Taken together, the findings from both the off-road and on-road parts of the study suggest that risk misperception-or at least seeing less risk in driving situations than older male drivers-may account in significant measure for the disproportionate involvement of young male drivers in traffic accidents. These findings suggest several avenues for further research and programmatic change. In terms of additional research, if young drivers see themselves as less likely to be involved in an accident than their peers, then the logic or rationalization behind this perception should be explored. Do young drivers see themselves as more skilled than their peers? more cautious? more likely to avoid risky locations? luckier? If, for example, it were found that young drivers see themselves as more skillful than their peers and that this claimed superior level of skill was seen as sufficient to avoid accidents, then an educational program which demonstrated the limits of driving skill in accident avoidance (and the need for defensive driving) could alter this misperception of young drivers. Driver education courses should also be evaluated to determine if they are increasing or decreasing the perception of risk of an accident. Finally, drivers’ misperception of the risk of an accident at a particular location could be modified by posting signs indicating the number of accidents at the location, or “black spot” signs similar to those used in England. Alternatively, the presence of crosses marking fatality locations might be evaluated for their impact on perceived risk of an accident. Analysis
Acknowledgments-This research was supported by Contract No. DTNH22-80-C-07360 with the National Highway Traffic Safety Administration, U.S. Department of Transportation. We wish to thank Jonathan Hodgdon, Martin Stein, Michael F. Smith and Raymond C. Peck for valuable reviews of earlier reports on which this paper is based.
REFERENCES Berger R. J. and Persinger G. S., 1980 Survey of Public Perception of Highway Safefy. Automated Services Inc., McLean, VA, 1980. Bragg W. E. and Cousins L. S., Changing the perceived likelihood of arrest for impaired driving. Presented at the Eleventh Annual Conference of the Canadian Foundation on Alcohol and Drug Dependencies, June 22, 1976. Bragg B. W. E. and Finn P., Young driver risk-taking research: Technical report of experimentalstudy. Prepared for the National Highway Traffic Safety Administration. National Technical Information Service, Springfield, VA, 1982.
298
P. FINNand B. W. E. BRAGG
Bragg B. W. E. and Finn P., Influence of safety belt usage on perception of risk of an accident. Accid. Anal. & Prev. 17, 15-23, 1985. Cousins L. S. and Bragg B. W. E., The effects of public advertising on subjective probability of arrest for impaired driving: A telephonesurvey. Paper presented at the Twelfth Annual Conference on the Canadian Foundation on Alcohol and Drug Dependencies, Winnipeg, June 1977. Cousins L. S., The effects of public education on subjective probability of arrest for impaired driving: A field study. Accid. Anal. & Prev. 40, 216-227, 1980. Federal Highway Administration, Fatal Injury and Accident Rates, 1980. Government Printing Office. Washington, D.C., 1982. Hodgdon J., Bragg B. W. E. and Finn P., Young Driver Risk-Taking Research: The State of the Art. Report to the National Highway Traffic Safety Administration. National Technical Information Service, Springfield, VA, 1981. Konecni V., Ebbeson E. B. and Konecni D. K., Decision process and risk taking in traffic: Driver responses to the onset of a yellow light. J. Appl. Psych. 6, 359-367, 1976. Mathews M. L. and Moran A. R., Age differences in male drivers’ perception of accident risk: The role of perceived driving ability. Accid. Anal. & Prev. 18, 299-313, 1986. McGuire E and Kersh R. C., A Study of the History, Philosophy, Research, Methodology and Effectiveness in the Field of Driver Education. Publications in Education, Volume 19. University of California Press, Los Angeles, 1969. National Accident Sampling System (NASS), Abt Associates Inc., 1981. Analysis of 1979 Computerized Data. Slavic P., Fischoff B. and Lichtenstein S., Rating the risks. Environment 21(3). 14-20 and 36-39, 1979. Stevens S. S., Psychophysics. Wiley, New York, 1975. U.S. Department of Transportation, Fatal Accident Reporting Sysfem, 1979, p. 3. U.S. Government Printing Office, Washington, D.C., 1981. U.S. Department of Transportation, Fatal Injury and Accidenr Rates, 1980. U.S. Government Printing Office, Washington, D.C., 1982. Wilde G. J. S., Age, Sex, and Climatic Influence on Seat Belt Use. Transport Canada, Ottawa, 1977. Williams A. F. and Karpf R. S., Teenaged drivers and fatal crash responsibility. Law Policy Quarterly 6. 101113, 1984.
18, No. 4. pp. 289-298.
OcQl-4575186 $3.00 + .cm 0 1986 Pergamon Journals Ltd.
1986
PERCEPTION OF THE RISK OF AN ACCIDENT YOUNG AND OLDER DRIVERS
BY
PETERFINN Abt Associates Inc., 55 Wheeler Street, Cambridge, MA 02138, U.S.A. and BARRY W. E. BRAGG Program Evaluation Branch, Department of Communications, Ontario KlA OC8, Canada
300 Slater Street, Ottawa,
Abstract-Young drivers are significantly overrepresented among all drivers involved in traffic accidents and fatalities. Excessive risk taking by young drivers appears to be largely responsible for this disproportionate involvement. This excessive risk taking could be due to (1) being more willing to take risks than older drivers are, (2) failing to perceive hazardous situations as being as dangerous as older drivers do or (3) both causes. This paper reports the results of a study which attempted to determine whether misperception of risk could be an explanation for the high rates of traffic accidents among youth by testing whether young drivers perceive driving to be less hazardous than do older drivers. Three different methods of estimating the risk of accident involvement were used to compare risk estimates of young and older drivers. The methods included general questions about accident involvement, rating the riskiness of ten specific driving situations illustrated in still photographs, and rating the riskiness of fifteen videotaped driving situations. Young drivers perceived their own chances of an accident to be significantly lower than those of both their peers and older male drivers, while older male drivers saw their chances of accident involvement as comparable to those of their male peers and less than those of young male drivers. These findings lend support to the thesis that young male drivers are overrepresented in traffic accidents at least in part because they fail to perceive specific driving situations as being as risky as older drivers perceive them. INTRODUCTION
Young drivers are significantly overrepresented among all drivers involved in traffic accidents and fatalities. In 1979, drivers under the age of 25 constituted 22% of the registered driver population and nearly 37% of fatal accident-involved drivers (U.S. Department of Transportation, 1981). Furthermore, young drivers are much more likely than older drivers to be responsible for the crashes in which they are involved (Williams and Karpf, 1984). In reviewing the research on young driver accident involvement, Hodgdon, Bragg and Finn (1981) found that this overrepresentation in collisions could not be explained by more miles driven, less driving experience or poorer mechanical condition of vehicles driven by young drivers as compared with older drivers. What did account for greater collision involvement by young drivers was the manner in which they drove. Young drivers speed more frequently, drive through yellow lights more frequently, wear safety belts less frequently and accept shorter temporal gaps when entering traffic-in other words, young drivers take more risks than older drivers. The excessive risk taking by young drivers could be due to (1) being more willing than older drivers to take risks, (2) failing to see hazardous situations as being as dangerous as often as older drivers do or (3) both. The ultimate design of a countermeasure program might be very different depending on the extent to which risk perception, as well as risk choice, is a determinant of young driver behind-the-wheel behavior. This paper reports the results of a study which attempted to determine whether misperception of risk could be an explanation for the high rates of traffic accidents among youth by testing whether young drivers perceive driving to be less hazardous than do older drivers. If the results of the study showed that young drivers perceive driving to be less risky than do older drivers, strategies could be developed to alter risk perceptions along with attempts to influence young drivers to make less risky driving decisions. The main research hypothesis which the experiment was designed to test was that male drivers aged 18-24 would perceive less risk associated with a given driving situation
290
P. FINN and B. W. E. BRAGG
than would older male drivers. In the present study, we concentrated exclusively on the influence of age and, by proxy, driving experience as potentially direct influences on the risk perception of driving situations. While some researchers (e.g. Slavic, Fischoff and Lichtenstein, 1979) have found that people have difficulty estimating the frequency with which low probability, catastrophic events may occur, others (e.g. Bragg and Cousins, 1976; Cousins and Bragg, 1977; Cousins, 1980) have indicated that subjects can make meaningful and reliable estimates of some low probability events such as the likelihood of arrest for impaired driving. STUDY
METHODOLOGY
Risk perception measurement approaches
We developed three off-road approaches for obtaining the subjects’ estimates of the risk of accident involvement. The first approach involved a 33-question closed-ended questionnaire. Several questions sought driving-related information about the subjects. One series of questions focused on the subjects’ estimates of the number of people injured and killed in traffic accidents in Massachusetts and the nation. Another series asked subjects to estimate the percentage of three categories of male drivers who would get into a traffic accident next year-drivers aged 18-24, drivers aged 38-50, and the respondents themselves. Certain questions were included because they tapped the subjects’ perception of driving risk in a global sense (e.g. “How many people were killed in traffic accidents in Massachusetts last year?“), while other questions focused on the subjects’ perceptions of driving risks related to two specific risk-taking behaviours (failure to wear safety belts, drinking before driving). Three sets of questions were designed to elicit the subjects’ comparative estimates of the likelihood of traffic accidents for male drivers their age and their father’s age for use later in comparing the subjects’ comparative estimates of accident risk in response to photographed and videotaped driving scenes. In the second risk perception measurement approach, each subject was shown ten still photographs, with descriptive captions, of different driving situations and asked to indicate how much greater or less the risk of an accident would be for male drivers his own age in each situation compared to the risk for his peers driving during the daytime on normal suburban road conditions illustrated in an eleventh (baseline) photograph. Subjects repeated this assignment for male drivers in the other driver age group and then repeated the rating a third time for themselves. The following driving situations were represented in the photographs and described in the accompanying captions: (1) driving on bald tires; (2) nighttime driving; (3) driving on an undivided two-lane rural road; (4) urban driving; (5) driving on an eight-lane divided highway at 45 mph when other traffic is travelling at 55 mph; (6) driving at 65 mph on an eight-lane divided highway when other traffic is travelling at 55 mph; (7) driving on wet roads when it is raining; (8) following one-and-one-half car lengths behind another car while travelling at 30 mph; (9) driving on snow and ice covered roads and (10) driving after consuming six 12-ounce cans of beer within one hour. The third risk perception test was to view a videotape of fifteen driving situations and estimate the relative likelihood of an accident in each compared to the chances of an accident in a baseline videotaped sequence. Each sequence was designed to give the viewer an overall view of the roadway and traffic conditions and then a driver’s eye view of the driving situation. The driving situations included, for example, a vehicle driving on a divided six-lane highway with no traffic present (baseline); a vehicle entering a freeway causing another vehicle to change lanes to avoid a collision; a vehicle at a suburban traffic-free intersection; a vehicle running a stop sign causing a second vehicle to swerve to avoid a collision; and a pedestrian suddenly emerging in front of a vehicle. The subject was instructed to rate the risk of an accident for each scenario using the same relative magnitude estimation method he used for the photographs. Subjects
The major focus of the study was to compare the risk perceptions of young male drivers with those of older male drivers. Female drivers were excluded from the study
Perception of accident risk by drivers
291
because the high accident rate for young drivers is primarily a high accident rate for young male drivers (Hodgdon, Bragg and Finn, 1981). For example, 18.5% of 19-yearold male drivers have an accident in one year, whereas only 5.9% of 19-year-old female drivers have an accident in one year (National Accident Sampling System, 1981). Ideally, we should have compared a representative sample of young male drivers and older male drivers. To obtain a sample of sufficient size to be nationally or even regionally representative, however, would have been prohibitively expensive and administratively impractical. Instead, young and older male drivers were recruited through a newspaper advertisement in the major Boston newspaper. If this recruitment procedure resulted in any bias it should have produced a more homogeneous group of participants thus making it more difficult to detect age differences which might exist in the population. However, we minimized young driver/older driver sampling bias by controlling demographic effects. A comparison between the young and older drivers in the sample indicates that they are similar in the number of miles they drive, their frequency of accidents and violations, their use of safety belts, and the percentage of their driving done in the city. The young drivers differ from the older drivers in the distribution of violations, with higher variability among the young drivers (suggesting that a smaller proportion of young drivers are responsible for more of the violations), and in the proportion of driving done at night, with young drivers doing a significantly greater proportion of their driving at night. Subjects were first administered the questionnaire, they then rated the still photographs and they then concluded with the videotape ratings. Ninety-three subjects were tested, 45 subjects between the ages of 18-24, and 48 subjects between the ages of 3850. RESULTS
The findings from the study are reported separately below for the interview questions, still photograph ratings and videotape ratings. Results of interview questions on perception of risk
Previous research on perception of risk (Berger and Pensinger, 1980) has shown that young drivers tend to see driving as being more hazardous than do older drivers, insofar as young drivers provide higher estimates of accident and injury likelihood. We attempted to replicate these findings by asking the subjects to estimate the number of traffic fatalities and injuries per year in the United States and in Massachusetts (their own state). The results are presented in Table 1. As may be seen, no differences exist between young and older subjects in the average estimates of the number of persons killed in traffic accidents in the United States and the number of persons injured in traffic accidents in Table 1. Estimates by young and older male drivers of the number of persons injured and killed in traffic accidents in Massachusetts and the United States mom
Driver
loution
Vai&le -Injured
Killed
EZFZ Young Older
45 48
1g9,ooo 203,000
.95
U.S.
Young Older
45 40
7,019,000 2,096,OOO
.9g
Mass.
Young Older
45 48
5,483 1,769
2.44
U.S.
Young Older
45 40
227,000 516,000
-.66
prepared
January
3
1980
Federal
Annual Hiahway Traffic Safety Trend Re for the National Highway Traffic Safety 1982).
Fatal Accident Reporting Administration (Washington,
Highway
Aetu1
t
Mass.
1 Fatal Injury and Accident Ratea. (Washington, D.C.: June 1982). 2 First
lmiutr
!!_
(19a) 67,760l
3,260,0002
8811
51,0773
Administration
rt, Abt Associates Inc., Administration (Cambridge
System, the National Highway Traffic Safety D.C.: 1981). Fatalities as of August 1980.
MA:
P. FINN and B. W. E. BRAGG
292
the United States and in Massachusetts. Young drivers did estimate that significantly more persons were killed in traffic accidents in Massachusetts (X = 5483 per year) than did older drivers (x = 1769). In fact, 881 persons died in traffic accidents in Massachusetts in 1981, the year the study was conducted (U.S. Department of Transportation, 1982). The older drivers were, therefore, more accurate in their lower estimate. In general, it appears that all subjects had very inaccurate perceptions of the actual magnitude of the traffic accident problem. Subjects overestimated by almost three times the frequency of injuries in Massachusetts and then either underestimated by one half (older drivers) or overestimated by over two times (younger drivers) the frequency of injuries in United States. Subjects overestimated the traffic fatalities in Massachusetts by two to six times, and they overestimated the actual rate of fatalities for the United States on average by almost four to eight times. It appears that Massachusetts was seen as a relatively more hazardous location in which to drive than is the United States as a whole. In fact, in 1981, the fatality rate per 100 million miles of travel was lower in Massachusetts (2.7) than for the United States as a whole (3.39) (Federal Highway Administration, 1982). In addition, there was significantly greater variability in the estimates of young drivers of the number of persons killed in Massachusetts (F = 10.54, df = 44,47, p < .OOl). Conversely, older drivers had significantly greater variability in their estimates of the number of persons injured in Massachusetts (F = 7.56, df = 44,47,p < .OOl), the number of persons injured in the United States (F = 2.64, df = 44,47, p < .OOl) and the number of persons killed in the United States (F = 38.96, @ = 44,47, p < .OOl). We had predicted that young drivers would perceive less risk in driving than would older drivers. These data only partially support that hypothesis, with young subjects estimating fewer fatalities in Massachusetts than older drivers. We did not expect, however, that older drivers would be more variable than young drivers in their estimatesas we found in three of the four estimates. Respondents were also asked to estimate the proportion of all drivers who were involved in an accident in the previous year and to estimate the chances of 18-24 year old males, 38-50 year old males and themselves being involved in an accident in the next 12 months. The results are shown in Table 2. As can be seen in Table 2, 18-24 year old males were judged by both young and older drivers to be significantly more likely to be involved in an accident than all other categories of drivers. Male drivers aged 38-50 were judged significantly more likely to be involved in an accident by young drivers but not by older drivers. Of most relevance, however, is that young drivers estimated their own chances of accident involvement (21.4%) as significantly lower than the chances of the average 18-24 year old male (37%), while older drivers, aged 38-50, judged their own chances of accident involvement (13%) to be comparable to those of other male drivers their age (14.8%). Thus, while young male drivers recognized that their age group and sex is at greater risk of accident involvement than are older male drivers they perceived their own risk to be significantly
Table 2. Young and older male drivers’ estimates of average percentage of drivers involved in a traffic accident’ in the previous year2 18-24 ,ar old
male*
All Llrioaa
Bate%
38-50
year old --males
Self
Young drivers
29.1,
37.Od
21.1b
21.4b
Older
26.4,
36.64
14.aa
13.08
15.1
6.8
drivers
Actual (Police
6.8
1 An accident was personal injury.
’
Any
different
NA
Reported)3
fwa
defined
means not (c - 1.96,
3 Police-reported
as
a collision
containing the df
-
accidents
91,
involving
same
subscript
at are
least
$100
significantly
pt.05).
with
$400
damane
or
personal
injury.
damage
or
Perception of accident risk by drivers
293
lower than that of their mule peers. By contrast, older male drivers saw their own chances
of accident involvement as comparable to those of their male peers. In summary, generalized questions about the number of persons killed or injured in a jurisdiction produce highly erroneous estimates. Estimates of the percentage of drivers involved in an accident per annum produce less erroneous overestimates. Young drivers see themselves as less likely to be accident involved than their peers while older drivers see their own chances of accident involvement as similar to that of their peers. Results of risk estimates of general driving situations (in still photographs) Subjects provided estimates of the risks of general driving situations by providing magnitude estimates of the risk of an accident for their peers, the other age group (young or older drivers), and themselves in ten driving situations depicted in photographs. Subjects were asked to estimate how much greater or less the risk of an accident was in each situation compared with a standard condition depicted in an eleventh photograph that showed a vehicle described as travelling at the speed of other traffic on a divided sixlane highway, in light traffic, in dry weather, during daylight. All judgments were made relative to this baseline. The data were subjected to a 2 transformation where all the magnitude estimates for a given subject (on all three target groups-own age group, other age group and self-for all ten driving situations) were used as the set for the transformation. By using a Z transformation each subject’s scores are converted to a common normalized scale. This eliminates the possibility of exploring age differences in the use of scales-an interesting but academic issue-but allows for the comparison of different age groups’ perceptions of different target groups using the same scale. A three factor analysis of variance with one random factor (age) and two repeated fixed effects (target group and driving situation) was conducted. Driving situation is nested within target group. The results of the analysis of the Z transformed scores are shown in Table 3. These results show a significant effect for target (2) (i.e. the group being evaluated); a significant target by age-of-subject (TA) interaction; a significant driving situation (P) main effect; a significant target group by driving situation (7’P) interaction; and a significant age-of-subject by target-group by driving-situation (WA) interaction. Table 4 shows the means for the age of subject by target group interaction. (The lower the mean, the lower the magnitude of the risk of an accident.) The data indicate that, over all driving situations, young drivers saw their chances of an accident us signif-
Table 3. Analysis of variance of Z transformed magnitude estimates of risk of accident involvement for three groups rating ten driving situations, by young and older drivers
Meall
0.00003
1
0.00003
0.54
0.4658
A (driver group)
0.00011
1
0.00011
2.40
0.2150
Error
0.00431
91
0.00005
T (target
group)
TA Error
P (driving
35.95768
2
17.97884 3.97018
7.94037
2
153.68895
182
0.84444
1237.83375
9
137.53708
21.29 4.70
136.06
0.0000 0.0102
0.0000
sitIl.¶tilXl) PA Error
7.733853 827.88493
9
819
0.81539
0.81
0.6102
1.01085
TP
8.05914
18
0.44773
1.79
0.0217
TPA
7.73700
18
0.42983
1.72
0.0305
409.85664
1638
0.25022
Error
294
P.
FINN
and B. W. E.
BRAGG
Table 4. Mean Z transformed ratings’ of risk of an accident for young male drivers, older male drivers and self as estimated by young and older male drivers* Target Drivers
Young
Older
Drivers
-Self
Judge
Young
Drivers
.038b
Older
Drivers
.22,
1 2
High
score
Any means
= high in
risk
a row
significantly
not
different
0.69,
-.09,
-0.41,
_.17b
rating containing (t
=
the 1.96,
same
df
= 91,
subscript
are
p (.05).
icantly less than those of both similar age other male drivers and older male drivers. Older drivers saw the risk of an accident as significantly greater for young male drivers than for older male drivers and saw their own chances of an accident as significantly lower than those of both similar aged male drivers and young male drivers. When the ratings are examined for specific driving conditions (the age-of-subject by target-group by drivingsituation interaction), the differentiations of young drivers become apparent (see Table 5). Young drivers saw themselves as less at risk of an accident than their own peer group in two of the ten situations: driving on wet roads and driving after consuming six beers within one hour. In no situation did young drivers see themselves at greater risk of an accident than their peer group. Young drivers saw themselves as less at risk of an accident than older drivers in five driving situations: tailgating (driving one-and-a-half car lengths behind another vehicle at 30 mph), driving at night, speeding, driving on snow-covered roads and driving after consuming six beers within one hour. In none of the ten driving
Table 5. Mean Z transformed ratings1 of risk of accident involvement for young male drivers, older male drivers and self by young and older male drivers rating ten driving situations Group
situation I.
2.
3.
4.
5.
6.
Urban
Rural
driving
driving
Bald
tires
Tailgating
Slow
Driver
Night
Driver
7.
Wet
8.
Speeding
9.
10.
road
covered
Young Male
Target Older Hale
B
Drivers
Driven
-self
Young Older
-.43b -.34b
-.Slab -.57b
-.54ab -.71a
Young
-.71a
-.69a
-.76a
Older
-.32b
-.60a
-.73a
Young
-.46ab
.47ab
Older
-.86a
-.58b
-.35a -.47ab
Young
-.35ab
-.28a
-.43b
Older
-.l&
-.32ab
-.53b
Young
-.50ab
-.5&b
-.55ab
Older
-.49a
-.65ab
-.71b
Young Older
-. 19b .05a
.13a -.15b
-.30bc -.42c
Young Older
.13a .20a
.Olab -.OOb
-.13bc -.24c
-. 25bc -.47a
-.37ab -.50a
-.46a -.14c
Young Older
50bc
.69c
.39b .05a
road
Young Older
.62c
Drinking
Young
1.74cd
1.7oc
1.45ab
Older
1.94d
1.31a
1.59bc
Snow
I Any means within a given subscript are significantly
driving different
situation (t =
which do 1.98, df -
.46b
not have the 1638, P (.05).
same
Perception of accident risk by drivers
295
situations did young drivers see their own risk of an accident as greater than that of older drivers. Young drivers saw their peer group of young male drivers as significantly less likely to be involved in an accident than experienced drivers in two driving situations: night driving and speeding. Once again, the young drivers saw their own chances of accident involvement as less than those of their peer group and particularly less than the chances of an accident for older drivers. Older drivers saw their own risk of accident involvement as signficantly lower than that of their same age male peer group on three of the ten driving situations: night driving, driving on wet roads and driving on snow-covered roads. They saw their own chances of an accident as significantly higher than those of their same aged male peer group in one situation: drinking and driving. Older drivers saw young drivers as significantly more likely to be involved in an accident than older drivers in eight of the ten driving situations. Only driving 10 mph slower than other traffic and tailgating were seen as situations of comparable risk for young and older drivers. In summary, the photo rating data support and refine the finding that young drivers see their accident likelihood as lower than that of their peers. Five of ten driving situations are seen by young drivers as having a lower risk of an accident for themselves than for their peers. However, the photo rating data also identified three situations where older drivers saw their accident likelihood as lower than that of their peers, and one where the risk was seen as higher. Magnitude estimates of videotaped driving sequences
Each subject was shown a specially filmed videotape of a series of 15 driving situations. The sequences were randomly ordered with relation to the degree of risk of an accident. Each subject was asked to make magnitude estimates of the relative risk of an accident in each situation compared with a baseline sequence shown at the beginning of the videotape. The baseline comparison sequence showed a vehicle driving on a six-lane divided highway with no other traffic present. A correlation of the average risk estimate indicates high agreement (r = .85, df = 13, p < .Ol) in the average risk rating between the young drivers and the older age group. However, significant differences in the variances suggested that the data needed to be transformed in order to determine more precisely if there was any difference in the average ratings. A Z transformation on each subject’s video ratings was used, and the analysis of the transformed ratings revealed a significant age by driving situation interaction (see Table 6). Analysis of the means reveals that only two of the driving situations were rated significantly differently by young and older drivers. The tailgating sequence was rated significantly more dangerous by older drivers than by young drivers (t = 3.51, df = 91, p < .OOl), and the pedestrian sequence was rated significantly more hazardous by young drivers (t = 2.90, df = 91, p < .OOl) than by older drivers. These findings are not surprising. The tailgating sequence involves risk in which the skill of the tailgating driver is crucial to avoid an accident. Skill is more likely to be Table 6. Analysis of variance of Z transformed magnitude estimates of risk of accident involvement in 15 driving situations depicted on videotape as rated by young an older male drivers SW of gquarel,
Source MeaIl A (driver EKI-OK S (driving
group)
Degrees Freedom
of
hkan squarea
Tail RObability
p 55.29 .Ol
0.0000 0.9226
39.74756
81.23
0.0000
0.96372 0.48935
1.97
0.0170
0.04108 0.00000 0.00678
1 1 91
0.04108 0.00000 0.00007
556.46587
14
13.49210 623.42589
14 1274
SitLdiOll)
SA
ErrOr
296
P. FINNand B. W. E. BRAGG
assumed by the young driver. The pedestrian sequence involves the sudden emergence of a pedestrian in front of the driver. Here the unexpected danger and lack of control is likely to result in young drivers rating the situation as more dangerous whereas older drivers have, through years of exposure, learned that such unexpected actions occur quite frequently without usually resulting in an accident. In summary, the videotape data add support to the driving situation specificity found in the photo rating data. In the videotape data one sequence (pedestrian) is seen as more risky by young drivers, while another sequence (tailgating) is seen as more risky by older drivers. The underlying dimension that appears to explain this difference is a skill-control dimension with situations requiring skill and control seen as less risky by young drivers and a situation where skill and control would not reduce the risk seen as more risky by young drivers.
DISCUSSION
The purpose of this study was to determine whether young male drivers perceive driving to be less hazardous than do older male drivers. If they do, this finding would lend support to the hypothesis that the disproportionate involvement of young drivers in accidents may be due not merely or even primarily to a willingness to take more risks than older drivers but to a failure to see hazardous situations as being risky and, hence, to see the need to exercise caution. One constraint to the generalizability of these findings is the lack of representativeness of the samples of young and older drivers. As we noted in the study methodology, advertizing for subjects probably led to more homogeneous samples than the populations. This is reflected in the lack of a difference in the accident rate between young and older drivers. This homogeneity makes the observed differences recorded in this study even more important in that the differences are likely to be larger in the populations. The importance of the differences reported in this study are further strengthened by the findings of Matthews and Moran (1986) reported elsewhere in this issue. Matthews and Moran (1986) found similar differences using a slightly different set of stimuli and a sample that was homogeneous on a different dimension (education). The study suggests that young drivers do perceive the risk of an accident significantly differently than do older drivers-and in the direction of perceiving less risk than do older drivers. Both young and older drivers were asked to estimate the probability of an accident to all drivers, young drivers, older drivers and themselves. Both groups saw young male drivers as having a greater risk of an accident than all drivers, who in turn had a greater risk of an accident than older drivers. However, young drivers saw their own chances of accident involvement as significantly lower than those of other young male drivers, while older drivers saw their own chances of accident involvement as comparable to those of their peers. In addition, and in contrast to previous research (e.g. Berger and Persinger, 1980), young drivers in the sample for the most part did not see driving as more dangerous than did the older drivers. No differences were found between young and other subjects in the average estimates of the number of persons killed in traffic accidents in the United States and the number of persons injured in traffic accidents in the United States and Massachusetts. Only in terms of the number of fatalities did the young drivers provide higher estimates than older drivers. Although we found that young drivers see driving as more dangerous when these general questions about the risk of an accident were asked, and they recognize that their age group is at greater risk of accident involvement than are drivers 38-50 years old, young drivers see their own chances of an accident as lower than those of their own age group or older drivers when specific questions about their own risk are asked. This discrepancy between general risk rating and specific risk rating may help to explain why young drivers are able to see driving in general as more dangerous than do older drivers and yet still take more risks than older drivers.
Perception of accident risk by drivers
297
of the subjects’ ratings of the videotaped sequences showed that the older drivers rated one situation (tailgating) as significantly more dangerous than did the young drivers, but the surprise pedestrian situation was rated significantly more hazardous by the young drivers. In terms of on-road behaviour, however, the tailgating discrepancy between the groups is likely to reflect frequent hazardous driving on the part of young drivers, while the pedestrian discrepancy is not likely to lead to safer driving by young drivers. Avoiding collisions with pedestrians who have yet to be spotted requires a consistently safe driving approach that anticipates danger-an approach young drivers appear to lack-while failing to see the risk of tailgating can lead to constantly placing oneself (and others) in jeopardy by failing to allow a sufficient gap between cars. Thus, we would expect to find the different ratings between the two groups for tailgating and surprise pedestrians to be reflected in an overall higher rate of accidents among young drivers compared with older drivers. Thus, all three of our risk estimation measurement approaches supported our hypothesis that young male drivers would perceive less risk in driving situations than older male drivers would. The findings reported here are based, as noted, on off-road tests of subject risk perception. However, another part of the study, reported elsewhere (Bragg and Finn, 1985), tested the same subjects’ on-road perception of driving risks in fifteen locations along a predetermined four-mile course in normal traffic. Findings from the on-road part of the study confirm the findings reported here that the young drivers saw less risk in specific driving situations than did the older drivers. The on-road findings showed that young drivers, but not older drivers, perceived less risk as they became familiar with a traffic location-that is, after they had driven the test route a second time. Taken together, the findings from both the off-road and on-road parts of the study suggest that risk misperception-or at least seeing less risk in driving situations than older male drivers-may account in significant measure for the disproportionate involvement of young male drivers in traffic accidents. These findings suggest several avenues for further research and programmatic change. In terms of additional research, if young drivers see themselves as less likely to be involved in an accident than their peers, then the logic or rationalization behind this perception should be explored. Do young drivers see themselves as more skilled than their peers? more cautious? more likely to avoid risky locations? luckier? If, for example, it were found that young drivers see themselves as more skillful than their peers and that this claimed superior level of skill was seen as sufficient to avoid accidents, then an educational program which demonstrated the limits of driving skill in accident avoidance (and the need for defensive driving) could alter this misperception of young drivers. Driver education courses should also be evaluated to determine if they are increasing or decreasing the perception of risk of an accident. Finally, drivers’ misperception of the risk of an accident at a particular location could be modified by posting signs indicating the number of accidents at the location, or “black spot” signs similar to those used in England. Alternatively, the presence of crosses marking fatality locations might be evaluated for their impact on perceived risk of an accident. Analysis
Acknowledgments-This research was supported by Contract No. DTNH22-80-C-07360 with the National Highway Traffic Safety Administration, U.S. Department of Transportation. We wish to thank Jonathan Hodgdon, Martin Stein, Michael F. Smith and Raymond C. Peck for valuable reviews of earlier reports on which this paper is based.
REFERENCES Berger R. J. and Persinger G. S., 1980 Survey of Public Perception of Highway Safefy. Automated Services Inc., McLean, VA, 1980. Bragg W. E. and Cousins L. S., Changing the perceived likelihood of arrest for impaired driving. Presented at the Eleventh Annual Conference of the Canadian Foundation on Alcohol and Drug Dependencies, June 22, 1976. Bragg B. W. E. and Finn P., Young driver risk-taking research: Technical report of experimentalstudy. Prepared for the National Highway Traffic Safety Administration. National Technical Information Service, Springfield, VA, 1982.
298
P. FINNand B. W. E. BRAGG
Bragg B. W. E. and Finn P., Influence of safety belt usage on perception of risk of an accident. Accid. Anal. & Prev. 17, 15-23, 1985. Cousins L. S. and Bragg B. W. E., The effects of public advertising on subjective probability of arrest for impaired driving: A telephonesurvey. Paper presented at the Twelfth Annual Conference on the Canadian Foundation on Alcohol and Drug Dependencies, Winnipeg, June 1977. Cousins L. S., The effects of public education on subjective probability of arrest for impaired driving: A field study. Accid. Anal. & Prev. 40, 216-227, 1980. Federal Highway Administration, Fatal Injury and Accident Rates, 1980. Government Printing Office. Washington, D.C., 1982. Hodgdon J., Bragg B. W. E. and Finn P., Young Driver Risk-Taking Research: The State of the Art. Report to the National Highway Traffic Safety Administration. National Technical Information Service, Springfield, VA, 1981. Konecni V., Ebbeson E. B. and Konecni D. K., Decision process and risk taking in traffic: Driver responses to the onset of a yellow light. J. Appl. Psych. 6, 359-367, 1976. Mathews M. L. and Moran A. R., Age differences in male drivers’ perception of accident risk: The role of perceived driving ability. Accid. Anal. & Prev. 18, 299-313, 1986. McGuire E and Kersh R. C., A Study of the History, Philosophy, Research, Methodology and Effectiveness in the Field of Driver Education. Publications in Education, Volume 19. University of California Press, Los Angeles, 1969. National Accident Sampling System (NASS), Abt Associates Inc., 1981. Analysis of 1979 Computerized Data. Slavic P., Fischoff B. and Lichtenstein S., Rating the risks. Environment 21(3). 14-20 and 36-39, 1979. Stevens S. S., Psychophysics. Wiley, New York, 1975. U.S. Department of Transportation, Fatal Accident Reporting Sysfem, 1979, p. 3. U.S. Government Printing Office, Washington, D.C., 1981. U.S. Department of Transportation, Fatal Injury and Accidenr Rates, 1980. U.S. Government Printing Office, Washington, D.C., 1982. Wilde G. J. S., Age, Sex, and Climatic Influence on Seat Belt Use. Transport Canada, Ottawa, 1977. Williams A. F. and Karpf R. S., Teenaged drivers and fatal crash responsibility. Law Policy Quarterly 6. 101113, 1984.