Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Accident Analysis and Prevention 145 (2020) 105709

Contents lists available at ScienceDirect

Accident Analysis and Prevention


journal homepage: www.elsevier.com/locate/aap

Measuring the perception of aggression in driving behavior T


a, b
Thomas Kerwin *, Brad J. Bushman
a
Driving Simulation Laboratory, The Ohio State University, United States
b
School of Communication, The Ohio State University, United States

A R T I C LE I N FO A B S T R A C T

Keywords: The study of aggressive driving is an important step in the reduction of (often fatal) crashes due to this behavior.
Aggression However, even though various measures of aggressive driver behavior have been proposed, a more thorough
Perception examination of what the driving public perceives as aggressive driving behavior can be performed. A nationally
Driver behavior representative sample of 198 American adults saw and rated the aggressiveness of various driving behaviors in
Driving
videos. The videos were shown from a first-, second-, or third-person perspective. Some videos depicted close
Survey
following, varying in speed and distance from the car ahead. Participants also saw illegal passing videos and
collision or near collision videos. A number of variables that might influence judgments of aggressive driving
were included as controls (i.e., trait anger, aggressive and prosocial driving attitudes, driving experience).
Following other drivers closely was rated as aggressive, especially when viewed from a third-person perspective.
Illegal passes were viewed as more aggressive than speeding. Faster speeds didn’t increase aggressive ratings
much, regardless of perspective. Aggressiveness ratings were especially high for acts that could be considered
“road rage” (i.e., hitting or nearly hitting vehicles, cyclists, and pedestrians). People high in trait anger have a
bias to view many driving behaviors as intentionally aggressive.

1. Introduction definition: What behaviors show the intent to harm? Certain behaviors,
like following closer than two seconds behind a vehicle, could be
Worldwide, about 1.35 million people die each year as a result of considered normal driving behavior, although unsafe in the event of an
road traffic crashes (World Health Organization, 2019). In the United unexpected event on the road.
States (US), about 40,000 people die each year in motor vehicle crashes Dula and Ballard, in their development of the Dula Dangerous
(US Department of Transportation, 2016). According to the American Driving Index (Dula and Ballard, 2003), had participants rate driving
Automobile Association, 56 % of fatal crashes in the US from 2003 to behaviors on how strongly they felt the behaviors are associated with
2007 were associated with driving behaviors that are often labeled aggressive driving. We take a similar approach, but provide concrete
aggressive (AAA Foundation for Traffic Safety, 2009). However, de- examples of driving scenarios, with videos from the perspective of the
termining aggressive behavior from police reports is challenging, In driver in various driving situations. Showing participants the same
addition, the behaviors associated in the crashes are often, but, not driving behaviors should be more objective than having them read
always, seen as an aggressive behavior. The question of exactly when about and then imagine those behaviors.
driving behaviors are considered aggressive by the driving public is not It is useful to distinguish between aggressive driving and “road
one that has been thoroughly examined. rage”. The National Highway Safety and Traffic Administration
One definition of aggressive driving is “any behavior emitted by a (NHTSA) defines aggressive driving as “driving actions that markedly
driver while driving, that is intended to cause physical and/or psy- exceed the norms of safe driving behavior and that directly affect other
chological harm to any sentient being.” (Dula and Geller, 2003) This road users by placing them in unnecessary danger” whereas road rage is
definition captures the idea of intent in aggressive driving. It is also defined as “an intentional assault by a driver or passenger with a motor
consistent with how researchers typically define aggressive behavior: vehicle or a weapon that occurs on the roadway or is precipitated by an
“Aggression is any form of behavior directed toward the goal of incident on the roadway.” (Richard et al., 2018)
harming or injuring another living being who is motivated to avoid Although conceptual definitions of aggressive driving and road rage
such treatment”. (Baron, 1977) However, a question follows from this are well accepted, operational definitions are not. In both naturalistic


Corresponding author.
E-mail address: kerwin.6@osu.edu (T. Kerwin).

https://doi.org/10.1016/j.aap.2020.105709
Received 13 January 2020; Received in revised form 16 July 2020; Accepted 29 July 2020
0001-4575/ © 2020 Elsevier Ltd. All rights reserved.
T. Kerwin and B.J. Bushman Accident Analysis and Prevention 145 (2020) 105709

and driving simulation studies about aggressive driving, a critical de- 2. Methods
cision is how to quantify the degree of aggressiveness a driver is en-
gaging in. In the aggressive driving literature, the quantification of 2.1. Participants
aggressive driving which is commonly done in three ways: (1) asking
participants how much they engage in a list of behaviors pre-de- Recruitment was stratified so that we obtained a nationally re-
termined to be aggressive (e.g., tailgating, speeding), (2) measuring presentative sample. Our sample size was 198 American adults, ranging
particular driving metrics (e.g., following distance between vehicles, in age from 18 to 89 ( xAge = 47.35, sd = 16.49). Roughly half of the
speed) and generating an aggressive driving score, or (3) using a scoring participants (53.03 %) identified as female. Racial composition was
guide for research assistants to rate behaviors as aggressive. However, 64.65 % White, 13.13 % Black, 14.14 % Hispanic, 5.05 % Asian, and
all of these methods presuppose that certain driving behaviors are ag- 3.03 % other responses. All participants were required to have a valid
gressive. driver’s license and be at least 18 years old. Participants were recruited
Matthews and Noris (Matthews and Norris, 2002) surveyed drivers and paid by Qualtrics Panel.
in a study designed to see if there were factors that affected whether a
driver performing a notable action in the car would be perceived as
aggressive or not. They found that trait aggressiveness increased the 2.2. Measurement of control variables
likelihood of a borderline case being attributed to malign intent. The
driving scenarios given to the participants to evaluate were complex, Trait anger has been shown to influence driving behavior (Bogdan
but were provided via text descriptions only. et al., 2016; Nesbit et al., 2007) and could influence attitudes about
A purely text-based questionnaire like the Perception of aggressive aggressive driving behaviors as well. For example, one study found that
driving scale (PAD) (Alonso et al., 2019) asks respondants to imagine people high in trait aggressiveness are especially likely to attribute
driving situations and provide an aggressiveness value. This approach is malign intent to other drivers (Matthews and Norris, 2002). Recall that
well-suited to determine, for example, if drivers consider honking to be definitions of aggressive driving and road rage both include intent to
more aggressive than speeding. However, there are still abiguities when harm.
trying to determine, given a specific driving dataset, how aggressive is a We included other control variables that might relate to judgments
person’s driving. For example, an item in the PAD asks how aggressive of aggressive driving behavior: gender, age, and aggressive and proso-
the behavior “Approaching the vehicle behind” is. If we were to mea- cial driving attitudes. We also included a number of measures related to
sure the following distance of a driver in a simulator or from natur- driving frequency and quantity (i.e., how long the participants have had
alistic driving data, how would we determine if that behavior is ag- a driving license, how many times per week they drove, how long per
gressive, and how aggressive? The distance between the vehicles during week they drove, how many miles per week they drove). We were not
the close following is a factor in that determination, and including items sure of the effect driving experience would have the ratings: driving
that refer to concrete examples with various following distances can experience may lead to more acceptance of unsafe driving behavior due
help create such a mapping. to continual exposure on the road, or personal experience with dan-
We believe that examining what the driving public perceives as gerous driving may make one more likely to attribute malice from
aggressive driving is important, not only for informing future work that them.
asks the public about aggressive driving in general, but also for de- The participants were first asked demographic questions, including
termining policy. Drivers routinely exceed speed limits on highways, age, gender and race. Next, they were asked four questions related to
but this common behavior is not considered aggressive. The aim of this driving frequency and quantity: how long they have had a license, how
work is to better understand to what degree drivers consider various many times per week they drive, how long per week they drive and how
behaviors to be aggressive. Furthermore, identifying the specific mea- many miles per week they drive. The survey also included a trait anger
surable values associated with perceived aggressive driving behaviors scale (Spielberger and Sydeman, 1994), which contained 10 items (e.g.,
can be a useful tool in categorizing those behaviors in experimental or “I have a fiery temper” and “When I get mad, I say nasty things”) that
observational studies of driver behavior in simulation or on-road are rated on a 4-point scale (1=almost never to 4=almost always).
driving. They also completed a measure of driving attitudes (Harris et al.,
In this study, we present participants with video examples of various 2014) with 17 items to measure prosocial driving attitudes (e.g., “Drive
driving behaviors and ask them to quantify the aggressiveness of the with extra care around pedestrians” and “Slow down in a construction
driving. We describe the results of this survey and discuss future ap- zone”), and 12 items to measure aggressive driving attitudes (e.g.,
plications for a tool to quantify aggressive driving behavior from “Make rude gestures at other drivers when they do something I do not
driving data in a way that is informed by the views of the driving like” and “Speed up when another vehicle tries to overtake me”). All
public. driving survey items were rated on a 6-point scale (1=never to
6=always).

Fig. 1. Example still images from the videos presented to the participants. On the left, the car is about to narrowly miss a collision with a bicyclist. On the right, the
rear car is driving one second behind the truck.

2
T. Kerwin and B.J. Bushman Accident Analysis and Prevention 145 (2020) 105709

2.3. Videos bicyclist, and a car. Speeding videos show the driven car or a third-
person car overtaking at 50 MPH (86 KPH) a line of cars going 35 MPH
Next, the participants viewed short videos (7–14 seconds) taken (56 KPH). Additionally, participants rated videos of the driven vehicle
from scenarios performed in a driving simulator. A total of 32 videos passing vehicles by crossing a double yellow line into the oncoming
were watched and rated. The order of the videos was randomized per lane or going off the road to the left shoulder.
participant. Participants rated the videos on aggressiveness im- The participants were asked to rate the aggressiveness of the driving
mediately after viewing. An example of the images in the videos is behavior on a scale of 0 (not aggressive) to 10 (highly aggressive) and a
shown in Fig. 1. All videos were created specifically for the current binary scale (not aggressive or aggressive).
study, in simulator software (SimCreator) that is used for real-time Survey questions and videos were refined prior to delivery to the
driving simulator studies. All scenarios were of a four-lane undivided participants by having internal lab research assistants answer a draft
roadway. There was no additional traffic depicted in the videos in order version of the survey and give comments. Video length and precise
to reduce confusion. The behaviors chosen were based on common formatting of the questions was determined in this way.
elements from both the PAD and the Dula Dangerous Driving Index.
The videos show close following (tailgating), speeding, illegal pas- 3. Results
sing, collisions and near collisions. The close following videos are from
three visual perspectives, first-person (i.e., the car driven is performing Our target was 200 participants. During data collection, four re-
the aggressive action), second-person (i.e., the car driven is being ag- spondents provided erroneous data: one claimed to make 200 drives a
gressed against), and third-person (i.e., the car driven is in the other week for a total of 10 miles driven, one had a blank race field, which
lane from the perpetrator and victim cars). We used different visual should have been disallowed, one claimed to drive 70,000 miles a
perspectives because participants might view the behavior as less ag- week, and a fourth said they drive 2 times every week for a total of 2
gressive when they are performing it than when someone else is per- miles for 2 h (and listed their age as 22). These were replaced with new
forming it. (Campbell and Sedikides, 1999; @ respondents. After the survey was ended, two more errors were found:
mischkowskiFliesWallAre, 2012) The tailgating videos also vary in how one respondent drove an average amount of hours in a week greater
fast the vehicles are traveling: 35 miles per hour (MPH) (56 km per hour than the total number of hours in a week, and another respondent drove
(KPH)) or 65 MPH (105 KPH). These speeds were chosen to cover an average of 2000 miles a week in an average of 1.5 h, giving an
common speeds on moderate-speed urban roads and highways. They average speed of over 1300 MPH. These two respondents were ex-
also vary in how close the tailgating car is from the car in front of it. cluded, bringing the total number of valid respondents to 198.
This closeness is measured in Time to Collision (TTC), which describes
the time the tailgating car would take to hit the car in front of it if the
3.1. Personality variables
car stopped immediately. This reflects the idea of the “two second rule,”
a common rule of thumb for safe driving.
We use Cronbach’s Alpha to measure the internal consistency of the
The idea of the “two second rule” however, is variable and incon-
personality trait assessments. High internal consistency gives evidence
sistent in states across the United States. After reviewing different
that the questions in the assessments measure the same intended idea.
state’s driving manuals, stating the driving rules, regulations and
In the assessments, the Alpha was 0.93 for trait anger, 0.98 for the
guidelines for each state, the time varied from a suggested two second
prosocial driving assessment, and 0.95 for the aggressive driving as-
following distance, to four or six depending on driving conditions. For
sessment. These Alpha values are all greater than 0.9, suggesting very
example, the Ohio Department of Public Safety states, “Some safety
high internal consistency.
experts advise drivers to allow one car length ahead for each 10 mph of
Spearman’s rank correlation coefficient between trait anger and
speed. Others suggest following the ‘three-second rule’” (Ohio
prosocial attitudes to driving was −0.3, between trait anger and ag-
Department of Motor Vehicles, 2019). Alternatively, Pennsylvania’s
gressive driving was 0.46 and between prosocial and aggressive driving
Driver Safety Manual states, “A four second following distance, will
was −0.35. With adjusting for false discovery rate, all p-values for
allow you to steer or brake to avoid a hazard safely on the highway, if
these correlations were found to be lower than 0.05.
the pavement is dry” (Pennsylvania Bureau of Driver Licensing, 2019).
The collision videos were all in first-person, with the driver either
swerving (without slowing down) to avoid a slow moving vehicle or 3.2. Influences on perceived aggressiveness
person in their lane, or actually colliding with that vehicle or person.
There were six videos, with each action occurring with a pedestrian, a We performed exploratory data analysis to better understand the
survey results. Graphs showing the mean values and 95 % confident

Fig. 2. Effects of time to collision, speed, and viewpoint on perceived driving aggressiveness. Vertical lines show 95 % confidence intervals.

3
T. Kerwin and B.J. Bushman Accident Analysis and Prevention 145 (2020) 105709

intervals for perceived aggressiveness of the videos depicting tailgating ( p < 0.01). The differences between all other perspectives were sig-
can be seen in Fig. 2. Shorter time to collision values generally receive nificant as well ( p < 0.01, except for 1 st vs. 2nd perspective at
higher aggressiveness ratings. However, a higher speed is associated p < 0.05), as were the differences between the 35 and 65 mile per hour
with lower aggressiveness ratings at lower TTC values. A different way videos ( p < 0.01). However, perspective did not have a significant effect
to measure tailgating is the distance between the two vehicles. Distance on perceived aggressiveness for the speeding videos. For the passing
and TTC are directly related (distance equals TTC multiplied by speed), videos, passing on the left into the oncoming lane was considered to be
but their effect on the participants perceptions may be different. more aggressive than passing on the right onto the shoulder ( p < 0.01).
Many of the participants thought none of the tailgating videos For the collision videos, the fact that the action was a near-miss rather
showed aggression. Fig. 4 shows the distribution of responses for the than a collision was a significant ( p < 0.01) effect, but there was no
most aggressive seeming video: a half-second following distance at 65 significant effect on perceived aggressiveness due to the type of object
MPH. Zero is the most common rating for all perspectives, although the interacted with (i.e. person, cyclist or vehicle).
increase in mean perceived aggressiveness can be seen in the third- Using this model as a base, a generalized linear mixed model with a
person perspective histogram. binomial link function was fit for the binary responses. We refer to these
A plot of the collision responses can be found in Fig. 5. All objects as “logistic models”. Like the previous mixed models, they were gen-
had similar means, with the actual collisions perceived aggressiveness erated using the lme4 R package. In these models, none of the driving
ratings mean being slightly above 9 and the near-collisions averaging demographic data had a significant effect on perceived aggressiveness.
around 8. A plot of the speeding and passing responses can be found in For the car following videos, trait anger had a significant effect, and for
Fig. 6. Mean speeding perceived aggressiveness was slightly under 2.5, the collisions videos, prosocial driving attitudes had a significant effect.
while the illegal passes were scored much higher, around 8, with pas- The variable estimates for these models with the demographic data are
sing on the right (the shoulder) scoring lower than passing on the left shown in two figures. Fig. 9 shows the common elements between the
(into the oncoming lane of traffic). For the speeding videos, perspective models and Fig. 10 shows the variables that varied inside each video
was not found to be significant. group.
We modeled the responses using a linear mixed model fit by max-
imum likelihood. Possible predictors we considered are the per-in-
4. Discussion
dividual personality and attitude assessments, driving-related demo-
graphics, along with the per-video characteristics such as time to
Smaller distances between cars are associated with higher perceived
collision, speed and perspective. We also include by-response random
aggressiveness ratings by respondents. However, at a half-second time
intercepts. The R package lme4 was used to generate and compare the
to collision, a majority of respondents thought that first-person driving
mixed models. Based on graphing the data (Fig. 3), we performed a log
video had the lowest possible value of aggressiveness (as shown in
transformation of the distance variable for the following distance. A
Fig. 4). This provides evidence that close tailgating at a steady distance
likelihood ratio test comparison was used to determine that the log
is never considered aggressive by a large portion of the driving public.
transformation was a significantly better fit for the data. ( p < 0.05).
The strongest result from the survey data is that third-person
The videos fell into four groups: Following distance, passing,
viewing of tailgating behavior is much more likely to be perceived as
speeding, and collisions. We constructed models for each of those
aggressive than is first or second-person viewing. There are several
groups that included trait anger, the aggressive driving behavior score
potential explanations for this. A hostile perception bias, one type of
and the prosocial driving score. They also included the per-video
hostile attribution error, could influence the difference between first-
characteristics. We then constructed models that included all the
person and third-person values (Dill et al., 1997). The hostile perception
driving-related demographic data (e.g. the number of times they drive
bias is the tendency to perceive ambiguous social interactions as being
per week). The driving-related demographic data did not significantly
aggressive. For example, if you see one driver tailgating another, a
improve the model in any group except for the passing videos
hostile perception would be that they tailgater is an aggressive driver.
( p < 0.001). The variable estimates for these models with the demo-
In the first person condition, participants placed themselves in the
graphic data are shown in two figures. Fig. 7. shows the common ele-
standpoint of the driver, and would be less likely to believe themselves
ments between the models and Fig. 8 shows the variables that varied
as an aggressor than to believe a third party was an aggressor. However,
inside each video group. All models included random intercepts per
this does not explain the similar results between the first-person and
respondent.
second-person videos. A possible cause of the similarity is that the size
For the car following videos, the difference in aggressiveness ratings
of the image of aggressor car in the rear-view mirror is smaller than that
between the videos in first-person (i.e. driver tailgating) and second-
in the other conditions and therefore easier to dismiss. It could also be
person (i.e. driver being tailgated) perspectives was significant
that judgement of the distance between the driver’s car and the car

Fig. 3. Effects of vehicle speed and log of distance to vehicle on percieved driving agressiveness when viewed from a third-person perspective. Vertical lines show 95
% confidence intervals.

4
T. Kerwin and B.J. Bushman Accident Analysis and Prevention 145 (2020) 105709

Fig. 4. Histogram comparing perceived aggressiveness responses for the three perspectives at 0.5 s following time at 65 MPH.

Fig. 5. Mean perceived aggressiveness values of collisions and near-collisions. Bars show 95 % confidence intervals.

Fig. 6. Mean perceived aggressiveness values of speeding and off-road passing. Bars show 95 % confidence intervals.

behind it is difficult from the rear-view mirror image. and perhaps there is an interaction effect that contributes to this fact.
In general, the values of perceived aggression from the tailgating No significant effects on type of object was seen in the linear model.
videos was lower than anticipated. Even at a very unsafe following The fact that an illegal pass on the shoulder is found to be less ag-
distance (0.5 s TTC at 65 MPH), 70 out of 198 (35 %) respondents gressive than one into the oncoming lane matches our intuition. Passing
considered the behavior to be a 0 on a 0–10 scale – as non-aggressive as on the oncoming lane when it is disallowed by a double yellow line is
possible – when viewing the behavior in first person perspective. There both illegal and puts the people in the oncoming lane in grave danger.
could be several reasons for this. A possible explanation is that close The threat of collision with another vehicle is lower in passing on the
driving is considered by many to be a normal driving behavior. It also shoulder. The transgression in going on the other side of the road may
could be that the steady-state nature of our video caused many people have a much greater weight in attributing aggression intent than en-
to be assume a non-malign source of the close driving. A hostile attri- tering into an area that could be considered a unused part of the road.
bution error could play a role here as well, but even in the third-person Simple speeding was not considered very aggressive and did not
perspective, 35 out of 198 (18 %) participants gave the behavior a 0 exhibit the change in perceived aggressiveness based on first-person
score. versus third-person as did the tailgating videos. This could be because
Collisions were considered highly aggressive. In the logistic model, distances are perceived differently in the two perspectives but relative
hitting or nearly hitting a pedestrian, was slightly less likely to be speeds are not.
considered aggressive than hitting a vehicle. This is a confusing result, In general, the amount of time, number of trips, or distance a person

5
T. Kerwin and B.J. Bushman Accident Analysis and Prevention 145 (2020) 105709

Fig. 7. Effects of predictors common to all linear models. This includes driving-related demographic data and the personality assessments. Effect values on the
original 0-10 scale are shown for all video groups. Horizontal lines show 95 % confidence intervals.

Fig. 8. Effects of per-video variables in all linear models. Effect values on the original 0-10 scale are shown for all video groups. Horizontal lines show 95 %
confidence intervals.

Fig. 9. Effects of predictors common to all logistic models. This includes driving-related demographic data and the personality assessments. Odds ratios are shown for
all video groups. Horizontal lines show 95 % confidence intervals.

drives is not strongly related to their impressions of how aggressive the for the speeding scenarios but higher for the collision scenarios. High
driving scenarios were. For the illegal passing scenario, if a person had aggressive driving attitudes were associated with higher perceived ag-
more years driving and drove longer distances, they scored the video gressiveness in the speeding and car following scenarios.
actions as more aggressive. For the car following scenario, if a person
drove for more time per week, they scored the video actions as more 5. Conclusion
aggressive. These effects disappeared entirely in the logistic models.
There was no strong overall effect from driving experience or general Higher trait anger and aggressive driving attitude both lead to
vehicle trip behavior on driver’s rating of aggressiveness. people rating tailgating behavior as more highly aggressive. Using the
For the personality assessments, trait anger was associated with logarithm of the distance between cars is a good estimate of perceived
higher perceived aggressiveness in the car following and illegal passing aggressiveness. The same distance between cars at a high speeds is
scenarios. The higher perceived aggressiveness effect remained for the perceived as slightly more aggressive, whereas people tend to rate
car following scenarios even in the logistic models. High prosocial tailgating between two other drivers on the road as much more ag-
driving attitudes were associated with lower perceived aggressiveness gressive than the person tailgating themselves or even someone

6
T. Kerwin and B.J. Bushman Accident Analysis and Prevention 145 (2020) 105709

Fig. 10. Effects of per-video variables for the logistic models. Odds ratios are shown for all video groups. Horizontal lines show 95 % confidence intervals.

tailgating them. People’s amount of driving experience, in general, did meta-analysis of the association between anger and aggressive driving. Transp. Res.
not have much effect on the perceived aggressiveness of driving beha- Part F Traffic Psychol. Behav. 42 (October), 350–364. https://doi.org/10.1016/j.trf.
2016.05.009.
viors. Campbell, W. Keith, Sedikides, Constantine, 1999. Self-threat magnifies the self-serving
Although this work does not construct a metric of how aggressive a Bias: a meta-analytic integration. Rev. Gen. Psychol. 3 (1), 23–43. https://doi.org/10.
driver’s behavior is, it provides valuable information towards that goal. 1037/1089-2680.3.1.23.
Dill, Karen E., Anderson, Craig A., Anderson, Kathryn B., Deuser, William E., 1997. Effects
The perceived aggressiveness ratings gathered through this study pro- of aggressive personality on social expectations and social perceptions. J. Res. Pers.
vide concrete values that can be used to give a measurement of ag- 31 (2), 272–292. https://doi.org/10.1006/jrpe.1997.2183.
gressive driving behavior as observed in naturalistic or simulation Dula, Chris S., Ballard, Mary E., 2003. Development and evaluation of a measure of
dangerous, aggressive, negative emotional, and risky driving1. J. Appl. Soc. Psychol.
driving studies. By basing those estimates on descriptive data such as 33 (2), 263–282. https://doi.org/10.1111/j.1559-1816.2003.tb01896.x.
those gathered through this study, rather than prescriptive rules from Dula, Chris S., Geller, E. Scott, 2003. Risky, aggressive, or emotional driving: addressing
governments or safety agencies, the strength of validity evidence for the the need for consistent communication in research. J. Saf. Res. 34 (5), 559–566.
https://doi.org/10.1016/j.jsr.2003.03.004.
use of those measurements is increased.
Harris, Paul B., Houston, John M., Vazquez, Jose A., Smither, Janan A., Harms, Amanda,
Dahlke, Jeffrey A., Sachau, Daniel A., 2014. The prosocial and aggressive driving
CRediT authorship contribution statement inventory (PADI): a self-report measure of safe and unsafe driving behaviors. Accid.
Anal. Prev. 72 (November), 1–8. https://doi.org/10.1016/j.aap.2014.05.023.
Matthews, B.Alex, Norris, Fran H., 2002. When is believing "Seeing"? Hostile attribution
Thomas Kerwin: Conceptualization, Methodology, Writing - ori- Bias as a function of self-reported Aggression1. J. Appl. Soc. Psychol. 32 (1), 1–31.
ginal draft, Software, Validation, Formal analysis, Investigation, https://doi.org/10.1111/j.1559-1816.2002.tb01418.x.
Visualization, Data curation. Brad J. Bushman: Conceptualization, Mischkowski, Dominik, Kross, Ethan, Bushman, Brad J., 2012. Flies on the wall are less
aggressive: self-distancing ‘in the heat of the moment’ reduces aggressive thoughts,
Methodology, Resources, Writing - review & editing. angry feelings and aggressive behavior. J. Exp. Soc. Psychol. 48 (5), 1187–1191.
https://doi.org/10.1016/j.jesp.2012.03.012.
Appendix A. Supplementary data Nesbit, SundéM., Conger, Judith C., Conger, Anthony J., 2007. A quantitative review of
the relationship between anger and aggressive driving. Aggress. Violent Behav. 12
(2), 156–176. https://doi.org/10.1016/j.avb.2006.09.003.
Supplementary material related to this article can be found, in the Ohio Department of Motor Vehicles, 2019. Digest of Ohio Motor Vehicle Laws.
online version, at doi:https://doi.org/10.1016/j.aap.2020.105709. Pennsylvania Bureau of Driver Licensing, 2019. Pennsylvania’s Driving Manual.
Richard, Christian M., Magee, Kelly, Bacon-Abdelmoteleb, Paige, Brown, James L., 2018.
“Countermeasures That Work: A Highway Safety Countermeasure Guide for State
References Highway Safety Offices, Ninth Edition.” DOT HS 812 478. National Highway Traffic
Safety Administration.
Spielberger, Charles D., Sydeman, Sumner J., 1994. State-trait anxiety inventory and
AAA Foundation for Traffic Safety, 2009. Aggressive Driving: 2009 Research Update.
state-trait anger expression inventory. The Use of Psychological Testing for
Alonso, Francisco, Esteban, Cristina, Montoro, Luis, Serge, Andrea, 2019.
Treatment Planning and Outcome Assessment. Lawrence Erlbaum Associates, Inc,
Conceptualization of aggressive driving behaviors through a perception of aggressive
Hillsdale, NJ, US, pp. 292–321.
driving scale (PAD). Transp. Res. Part F Traffic Psychol. Behav. 60 (January),
US Department of Transportation, 2016. “2016 Fatal Motor Vehicle Crashes: Overview.”
415–426. https://doi.org/10.1016/j.trf.2018.10.032.
DOT HS 812 456.
Baron, Robert A., 1977. Human Aggression. Springer US, Boston, MA. https://doi.org/10.
World Health Organization, 2019. Global Status Report on Road Safety 2018. S.l. WORLD
1007/978-1-4615-7195-7.
HEALTH ORGANIZATION.
Bogdan, Smaranda Raluca, Măirean, Cornelia, Havârneanu, Corneliu-Eugen, 2016. A

You might also like