You are on page 1of 14

Accident Analysis and Prevention 191 (2023) 107190

Contents lists available at ScienceDirect

Accident Analysis and Prevention


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

Perceptions of novice driver education needs; Development of a scale based


on the Goals for driver education using young driver and parent samples
David Rodwell a, b, *, Natalie Watson-Brown a, Lyndel Bates c
a
Queensland University of Technology (QUT), Centre for Accident Research & Road Safety-Queensland (CARRS-Q), School of Psychology and Counselling, Faculty of
Health, 130 Victoria Park Road, Kelvin Grove, QLD 4059, Australia
b
Queensland University of Technology (QUT), Australian Centre for Health Services Innovation (AusHSI), School of Public Health and Social Work Centre for
Healthcare Transformation, Faculty of Health, 60 Musk Ave, Kelvin Grove, QLD 4059, Australia
c
School of Criminology and Criminal Justice and Griffith Criminology Institute, Griffith University, Messines Ridge Road, Mt Gravatt, QLD 4122, Australia

A R T I C L E I N F O A B S T R A C T

Keywords: Professional driver education and training has traditionally been used to improve the driving skills of young
Goals for driver education people with the expectation that they will drive more safely and therefore be less likely to crash. Despite this,
GADGET Matrix much research has failed to find such an effect. The Goals for Driver Education (GDE) is a theoretical framework
Young driver
developed to improve the effectiveness of driver education. The GDE seeks to draw attention to a broader array
Graduated driver licensing
of factors that may influence the driving of young people and hence increase the scope of what is included in
Driver training
Scale development driver education and training. Within the GDE, factors affecting driving and driver education are grouped into
combinations of four hierarchical levels that represent influences on driving at increasing levels of abstraction
(GDE level 1: Vehicle manoeuvring; GDE level 2: Mastery of traffic situations; GDE level 3: Goals and contexts of
driving; GDE level 4: Goals for life and skills for living) and three types of individual competencies (Knowledge
and skills; Risk-increasing factors; Self-evaluation skills). However, to date there have been few, if any, attempts
to operationalise the GDE in a quantitative way. In response, the Perceptions of Driver Education Scale (PDES)
was developed to gain an understanding of what two important stakeholder groups in the driver education and
training ecosystem perceived to be important to be taught to novice drivers. Understanding these perceptions
may lead to greater acceptance of the driver education and training that is designed and make it more effective.
Following rigorous scale development procedures, thirty-five items were developed a priori using expert opinion
and deployed, via an online survey, to a sample of parents of novice drivers (N = 518) and a separate sample of
novice drivers (N = 247), both from Queensland, Australia. Exploratory Factor Analysis was performed on the
parent responses and Confirmatory Factor Analysis was performed using the novice driver responses. A 29-item 9
factor solution was judged to be the best fit with the overall scale and each of the factors displaying strong levels
of internal consistency. The 9 factors cover most hierarchical level and competency combinations, although
removal of items meant that there is no representation concerning perceptions of Self-evaluation skills related to
Level 3: Goals and contexts of driving within the PDES. Overall, the PDES is a valid and reliable instrument for
investigating the aspects of driver education and training that are perceived as important by novice drivers and
parents. The scale may also be applicable for use with other stakeholder groups and could also be used to
evaluate pre-existing driver education and training programs.

1. Introduction promotes health and well-being, and enables continued connectivity


within communities (Collins et al., 2021; Ivers et al., 2016). However,
Having a licence to drive and regular access to a vehicle is beneficial young people have high involvement in road crashes. Globally, in 2019,
for young people, particularly those from marginalised communities, as 58% of deaths and 52% of Disability Adjusted Life Years (DALYs) of
it facilitates education and employment, increases social mobility, young people aged 10–24 years were transport-related (Peden et al.,

* Corresponding author at: Queensland University of Technology (QUT), Centre for Accident Research & Road Safety-Queensland (CARRS-Q), School of Psy­
chology and Counselling, Faculty of Health, 130 Victoria Park Road, Kelvin Grove, QLD 4059, Australia.
E-mail address: d3.rodwell@qut.edu.au (D. Rodwell).

https://doi.org/10.1016/j.aap.2023.107190
Received 30 March 2023; Received in revised form 11 June 2023; Accepted 22 June 2023
Available online 11 July 2023
0001-4575/© 2023 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
D. Rodwell et al. Accident Analysis and Prevention 191 (2023) 107190

Fig. 1. The hierarchical organisation of the GDE (adapted from Rodwell et al., 2021).

2022). Road crashes continue to be a problem in motorised countries malleable and can be influenced by education and training or other
even despite the advent of comprehensive Graduated Driver Licensing factors.
(GDL) programs in many of these places. For example, in the United In the GDE, influences on driving behaviour are categorised and
States, in 2020 there was a 15% increase in road fatalities in the 16–24 organised hierarchically, from those that are more concrete to those that
age group compared to 2019 (National Highway Traffic Safety Admin­ are more abstract, although each level can exert influence on all other
istration, 2021). In Australia, an analysis of a decade of data indicates levels (Hatakka et al., 2002). As depicted in Fig. 1 (adapted from Rod­
that road deaths peak at around 18 years of age (Bradshaw et al., 2015). well et al., 2021), the lowest level in the hierarchy is focused on psy­
Thus, there remains a great need to find ways to reduce the number of chomotor driving and vehicle manoeuvring skills directly responsible
road crashes, fatalities, and injuries of young people. One method that is for operation of the vehicle. The second most basic level concentrates on
popular is driver education and training for young people (Bates et al., specific real-time actions and decisions made by drivers to interact
2014; Lonero & Mayhew, 2010) aimed at developing driving skills safely with each other and the road environment. The next highest level
(Mayhew, 2007). focuses on more general skills and factors that related to a particular
A variety of driver education and training methods have been journey, which may not be specific to a particular in-the-moment
developed, however, most have focused on teaching basic psychomotor driving situation. The highest and most abstract level includes larger
driving skills or preparing drivers to negotiate specific common driving macro-contextual factors that may or may not be directly related to
situations that they may encounter (Beanland et al., 2013). It was hoped driving but may have some effect on driving behaviour. Conceived a
that by providing driver education and training, young people would different way, the first three levels of the GDE relate to operational,
become more skilled and safer drivers earlier, and therefore be involved tactical, and strategic decision-making while operating a vehicle (Berg,
in fewer crashes. 2006) while the highest level is concerned with factors that may affect
Unfortunately, evaluations have not found a consistent and strong indirectly or directly these operational, tactical, and strategic decisions.
link between traditional skills-focused professional driver education and Critically, Hatakka et al. (2002) assert that failure to attend to factors on
training and reductions in novice driver crash risk (Christie, 2001; Elvik the higher levels of the hierarchy cannot be compensated fully by im­
et al., 2009; Glendon et al., 2014; Haworth et al., 2000; Ker et al., 2005; provements in vehicle manoeuvring.
Lonero & Mayhew, 2010; Lund et al., 1986; Mayhew et al., 2002; Three types of competencies are included in the GDE (see Fig. 1)
Mayhew et al., 1998; Roberts & Kwan, 2001). To combat this lack of (Peraaho et al., 2003). Thus, aligning with the suggestion that driver
efficacy, the Goals for Driver Education (GDE) Framework (Hatakka education effectiveness might be increased by tailoring what is taught to
et al., 2002) was developed to guide the design, evaluation, and the needs of individual novices (Mayhew et al., 2002), the GDE specifies
implementation of professional driver education. This paper reports on that individualised knowledge and skills, risk-increasing factors, and
the development of the Perceptions of Driver Education Scale (PDES), self-evaluation skills should be attended to at each level of the hierarchy
which was designed to assist in operationalising the GDE. The PDES is (Rodwell et al., 2018). Knowledge and skills encapsulate the declarative
intended to measure perceptions of what is important to be taught when and non-declarative information a person requires to drive. Risk-
learning to drive. With this knowledge, organisations, legislators, and increasing factors refer to the specific lifestyle influences and psycho­
other interested parties can design driver education and training that is logical makeup of an individual that lead to higher risk for that person.
more likely to be accepted by important stakeholders. Such acceptance Self-evaluation skills refer to a driver’s insight into themselves and how
may lead to greater effectiveness of driver education. this might impact the way they drive. It is argued that these compe­
tencies have impacts at each level of the GDE hierarchy (Hatakka et al.,
1.1. Goals for driver education 2002).
Consequently, the GDE can be conceptualised as a matrix of twelve
The GDE (Hatakka et al., 2002) is a conceptual framework that distinct combinations of GDE level and competencies that is unique to a
identifies factors to be considered for effective professional driver edu­ person. Nevertheless, some generalisations can be made regarding each
cation and training (Bates et al., 2019; Rodwell et al., 2018). The GDE of these twelve facets of driver education that should be considered
draws from social cognitive theories of mental models and schematic when designing and conducting effective driver education and training.
processing (Hatakka et al., 2002; Keskinen, 2007). In essence, the GDE Table A1 in appendix A, adapted from Rodwell et al. (2018), provides an
asserts that driving behaviour results from the activation of an appro­ example of a skill or type of knowledge that may appear in each facet.
priate (or inappropriate) mental model, and that driving behaviour is The GDE has been strongly endorsed by road safety researchers and

2
D. Rodwell et al. Accident Analysis and Prevention 191 (2023) 107190

Table 1 Table 2
An overview of the process for scale development. Items developed for the PDES.
Item development Identification of domain and item generation Item GDE Level and Item
number competency
Content validity
1 1 KS The learner driver knows how to position the car
Scale Pre-testing of questions
in the lane
development Sampling and survey administration
2 1 KS The learner driver knows about the surface grip
Item reduction
of the car
Extraction of factors
3 1 KS The learner driver knows about centrifugal force
Scale Evaluation Tests of dimensionality (e.g., confirmation factor analysis)
when turning
Tests of reliability (e.g., Cronbach’s alpha; test-rest reliability)
4 1 RF The learner driver knows it is risky to drive
Tests of validity (e.g., concurrent validity; discriminant
when it was raining.
validity)
5 1 RF The learner driver knows it is risky to not use my
Source: Adapted from Boateng et al., 2018. seatbelt properly
6 1 RF The learner driver knows about the best seating
position.
experts (e.g. Bailey et al., 2022; Fisher & Dorn, 2016) and has been the 7 1 SE The learner driver can identify their abilities at
foundation for several European Union driver training and assessment manoeuvring the car in dangerous situations
projects (Panou et al., 2007), and used to reformulate State-based driver 8 1 SE The learner driver can identify their abilities at
manoeuvring the car in everyday situations
education curricula and testing regimes in Scandinavia (Henriksson
9 1 SE The learner driver can understand their
et al., 2007). Several studies have utilised the GDE as a theoretical strengths and weaknesses at driving a car
framework (Bates et al., 2019; Cuenen et al., 2014; Molina et al., 2014; 10 2 KS The learner driver knows about traffic rules
Rodwell et al., 2021; Rodwell et al., 2018; Rodwell et al., 2020; Tron­ 11 2 KS The learner driver knows about how to
smoen, 2010). For example, Watson-Brown and colleagues (Watson- anticipate what the traffic was going to do
12 2 KS The learner driver knows how much room is
Brown et al., 2018, Watson-Brown et al., 2019; Watson-Brown et al., needed to leave between them and the car in
2020a; Watson-Brown et al., 2021b; Watson-Brown et al., 2020b) used front to be a safer driver
the GDE as basis to examine professional driving instruction in a natu­ 13 2 RF The learner driver knows about the risks
ralistic setting and develop a model of Higher-Order Driving Instruction associated with poor decision-making
14 2 RF The learner driver knows about the risks
aimed at facilitating professional instruction of skills located on the
associated with breaking the road rules
higher levels of the GDE. 15 2 RF The learner driver knows about the risks
associated with trying to do too much while
1.2. The current study driving
16 2 SE The learner driver can see their own driving
tendencies
The GDE is a recognised theoretical framework that allows thinking 17 2 SE The learner driver has an appreciation of their
about driver education to be organised and can serve as a basis to design, limitations and abilities to respond to
implement, and evaluate novice driver education. However, no stand­ unexpected traffic situations
ardised quantitative instrument to operationalise the GDE has been 18 2 SE The learner driver has a realistic idea of their
own driving skill in traffic
developed (Rodwell et al., 2018). With this in mind, Bates et al. (2019) 19 3 KS The learner driver knows the best way to choose
deployed a survey that incorporated an author-devised 12-item scale, their travel route while driving
based on the GDE, to a sample of 114 novice drivers that had attended a 20 3 KS The learner driver knows how to estimate their
professional driver education course. These authors suggested that the travel time when driving
21 3 KS The learner driver knows about how having
items they created mapped to the twelve combinations of level and
passengers of their age in their care while
competency in the GDE. Participants were instructed to rate the benefit driving affects them
for learner drivers to be taught aspects of driving. The scale developed 22 3 RF The learner driver knows about the risks of
by Bates et al. (2019) was aimed at understanding what young people driving while tired
thought should be a focus of driver education in a systematic and 23 3 RF The learner driver knows about how their
driving might be different if their mum or dad
theoretically underpinned manner. Scores on this scale were used as the was in the car with them
dependent variable in several hierarchical multiple regressions that 24 3 RF The learner driver knows that becoming
sought to identify the influence that other psychosocial variables (e.g., competitive with other drivers is dangerous
normlessness, attitudes towards speeding) had on perceptions of driver 25 3 SE The learner driver can identify if they are a risky
driver
education located at each of the GDE levels. The results suggested, for
26 3 SE The learner driver can recognize if they are
example, that participants with more positive attitudes towards risky showing off to their mates while driving
driving or those who reported greater levels of trait driver thrill-seeking 27 3 SE The learner driver can identify if they need to
had stronger beliefs that driver education should focus on mastery of plan their driving trips more
traffic situations. 28 4 KS The learner driver knows about how their
personal values affect their driving
However, it is noted that Bates et al. (2019) scale was deployed 29 4 KS The learner driver knows about how their level
without confirmatory psychometric investigation or validation. That is, of self-control affects their driving
it was assumed that the items conceptualised as capturing each GDE 30 4 KS The learner driver knows that what their friends
level and competency did so. The objective of the current study was to do affects their driving
31 4 RF The learner driver knows that risk is a part of
further develop and investigate the properties of a scale, based on the
driving
GDE, that can be used to gain the views of both novice drivers and their 32 4 RF The learner driver knows that society values safe
parents regarding their preferences about professional driver education. drivers
33 4 SE The learner driver can identify their personal
2. Method level of impulse control
34 4 SE The learner driver can identify their personal
level of risk-taking
The current study was developed using data collected as part of a 35 4 SE The learner driver has thought about how
larger wide-ranging project investigating multiple aspects of driver ed­ sometimes they might enjoy driving in an unsafe
ucation and training (Bates et al., 2019; Horswill et al., 2021; Rodwell way

3
D. Rodwell et al. Accident Analysis and Prevention 191 (2023) 107190

note. 1 = GDE level 1, 2 = GDE level 2, 3 = GDE level 3, 4 = GDE level 4; KS = 2.3. Procedure
Knowledge and skills, RF = Risk-increasing factors, SE = Self-evaluation skills.
Participants completed an online survey (approx. 40 mins) that
et al., 2021; Rodwell et al., 2018; Rodwell et al., 2020). Developing consisted of several scales focusing on different aspects of interest in the
strong scales is time consuming and difficult (Hinkin, 1995). As shown larger research program. First, participants were directed to an infor­
in Table 1, scale development involves nine steps undertaken in three mation page and provided informed consent by checking a box. Par­
phases (Boateng et al., 2018). The following sections describe the pro­ ticipants responded to demographic items with the responses to some
cesses undertaken each of the three phases described in Table 1 when items (e.g., age for the novice driver sample; if they were a parent of a
developing the PDES. novice driver or not for the parent sample) used to screen out ineligible
participants. After this, scales and items, including the GDE items, were
presented to participants in randomised blocks to guard against survey
2.1. Item development
fatigue and response bias. The use of skip logic meant that some ques­
tions were only presented to participants if they answered previous
The scale intends to measure the perceived importance of the twelve
items in a certain way. Ethical approval was received prior to collection
combinations of hierarchical level and competency identified within the
of data from each sample (Novice drivers: QUT 1700000220; Parents:
GDE. Thus the GDE conforms to an a priori ‘domain’ of focus as discussed
QUT 1700000720).
by Boateng et al. (2018). The items were generated using a deductive
approach (Hinkin et al., 1997) with the work of Bates et al. (2019) as a
2.4. Participants and context
foundation.
Building upon the items created for Bates et al. (2019), and drawing
A Graduated Driver Licensing (GDL) program operates in Queens­
from their experience as a road safety researcher and deep knowledge of
land, Australia where the study was conducted. In this program young
the GDE, the third author (LB) generated a large pool of items to be
people move through three licensing stages before being granted an
included in the scale. The items that were generated were independently
unrestricted (open) licence. Learner drivers must be accompanied by a
assessed and checked for consistency with the GDE by the first author
suitable driving supervisor, usually a parent, while provisional 1 and
(DR) and others in the research team of the larger project. Discrepancies
provisional 2 drivers can drive independently subject to some re­
and differences were discussed, and items were added, removed, or
strictions (e.g., restrictions driving with peer age passengers at night).
modified, until consensus was achieved regarding the items to be
All GDL restrictions are removed when a person obtains an Open licence.
potentially included in the PDES. It is noted, that as part of this
Table B1 in appendix B provides a comparison of the components
comprehensive reflective process all twelve of the items in Bates et al.
making up each GDL stage at the time of the study.
(2019) were amended or changed in some way, or removed from the
The novice driver sample consisted of 247 young people. Inclusion
final pool of items.
criteria were having either a current Queensland Provisional 1 (P1) or
The final pool consisted of 35 items which are presented in Table 2
Provisional 2 (P2) licence, being aged between 17 and 20 years, and
along with an indication of the GDE level and competency combination
living in Queensland, Australia. Due to the aims and objectives of the
they conceptually aim to investigate. Three items relating to knowledge
larger project, learner drivers were not recruited for the novice driver
and skills, risk-increasing factors and self-evaluation skills were created
sample. Novice drivers were recruited using paid social media adver­
for each GDE level with one exception. Only two items that the authors
tisements, mailing lists, word-of-mouth, and from a voluntary university
believed theoretically aligned with risk-increasing factors at GDE level 4
first year research participant pool. Incentives for novice drivers con­
were kept in the initial pool of items. Thus, the unexamined scale hy­
sisted of the opportunity to enter a prize draw for a $100 gift card or
pothetically consisted of 12 factors in total with a single factor corre­
university students could choose instead to receive course credit towards
sponding to each combination of GDE level and competency; made up of
an undergraduate course.
eleven 3-item factors and one 2-item factor.
The parent sample consisted of 518 parents. Inclusion criteria were
being the parent of a currently licensed novice driver (learner, P1, or
2.2. Scale development P2), having a current Open Queensland licence, and living in Queens­
land, Australia. Parents were recruited using paid social media adver­
After the set of items that represent the 12 aspects of the GDE tisements, mailing lists, and word-of-mouth. The incentive for parents
framework were generated, these needed to be administered to a sample was the chance to win one of five $100 gift cards to a local department
of respondents in order to assess the psychometric properties of the scale store.
(Hinkin, 1995). A key consideration for this process is the sample used.
Suggestions for the appropriate ratio of number of items to sample size 2.5. Measures
include 1:4 and 1:10 although a minimum sample of 150 participants is
considered appropriate for an Exploratory Factor Analysis (EFA) and a The main measure in this study is the GDE scale items noted in
minimum of 200 participants for a Confirmatory Factor Analysis (CFA) section 2.2. Participants were instructed to: Please advise your level of
(Hinkin, 1995). agreement with the following in regard to what you consider is most important
For the current study, data was collected from two independent that formal driver education focus on teaching (as opposed to other types of
samples who answered the items in separate online surveys deployed at driver education such as driving with parents). The rating scale ranged from
different times. The first sample consisted of parents of novice drivers (n 1) not important for formal driver education to teach to 5) very important for
= 518) and the data was collected between November 2017 and April formal driver education to teach with the midpoint identified as 3) some­
2018. The second sample consisted of novice drivers (n = 247) and the what important. Thus, higher scores indicate stronger agreement that a
data was collected between July and November 2017. The use of two specific element within the GDE should be the focus of driver education
samples allowed both EFA and CFA to be undertaken. It is noted that, in and training.
line with previous sample size recommendations for factor analysis Additionally, novice driver and parent participants answered ques­
(Hinkin, 1995), both the parent and novice driver samples were suffi­ tions relating to their demographics and driving history (with some
cient to conduct an EFA and/or CFA. The inclusion of both a novice minor differences in items between the two samples). Demographics and
driver and parent sample was important because while novice drivers driving history items included for the current study were age (years),
may undertake the education, frequently it is the parents who are gender (male, female, other), licence type (learner, provisional 1, pro­
involved in decisions regarding the education. visional 2, open, other), attendance at any type of professional driver

4
D. Rodwell et al. Accident Analysis and Prevention 191 (2023) 107190

Table 3 was born. Novices drove, on average, for fewer hours per week than the
Demographic and driving history characteristics of the novice driver and parent parents.
samples.
Variable Novice driver (N = 247) Parent of Novice driver (N = 3.2. Scale development
518)

Novice driver age (%) 3.2.1. Exploratory Factor Analysis – Parents of novice drivers
17 years 0 – An EFA (Promax rotations with Kaiser Normalisation) was conducted
18 years 42 – using the parents of novice driver’s dataset (N = 518). The original scale
19 years 38
consisted of 35 items. The final solution is displayed in Table 4 and in­

20 years 20 –
Parent age (years) – M = 48.67 (SD = 4.98), Min cludes nine factors made up of 32 items; items 11, 25, and 26 were
= 32, Max removed because they loaded below 0.4. The Kaiser-Meyer-Olkin was
Gender (%) 0.95 and the Bartlett’s Test of Sphericity was significant (p <.001). The
Male 23 11
final solution explained 80.50% of the variance in perceptions of driver
Female 76 88
Other 1 1
education goals.
Licence type (%)
Learner 0 0 3.2.2. Psychometric properties – Parents of novice drivers
Provisional 1 39 0 The internal consistency of the overall scale (α = 0.96) and each of
Provisional 2 61 0
the factors was assessed using Cronbach’s alpha with each displaying
Open 0 100
Other 0 0 acceptable reliability. Factor 1 (α = 0.95), labelled Goals for life and skills
Driver education attendance (%) for living accounted for 47.06% of the variance. This factor contained
Yes 61 33 eight items (items 33, 28, 29, 30, 31, 32, 34, 35) and represents per­
No 39 67
ceptions regarding all competencies (knowledge and skills, risk-
Driving exposure M = 7.41 (SD = 5.27), Min M = 10.70 (SD = 6.72), Min
(hours) = 0, Max = 30 = 0, Max = 30
increasing factors, and self-evaluation skills) at the highest level of the
Crash involvement (%) GDE matrix; that is, macro-contextual factors theoretically relevant but
Yes 25 69 distal to driver education and driving behaviour. The second factor (α =
No 75 31 0.89) accounted for 10.58% of the variance and was labelled Self-
Offence committed (%)
assessment of driving in traffic. Factor 2 contained four items (items 9, 16,
Yes 87 –
No 13 – 17, 18). This factor aligns with perceptions regarding the GDE’s com­
Driving experience – M = 31.27 (SD = 7.18), Min petency of self-evaluation skills at level 2 of the matrix hierarchy; that is,
(years) = 7, Max = 59 beliefs about the importance of having the capacity to self-reflect on
Demerit point accrual (%) personal characteristics that could influence novice driver’s ability to
None – 38
Some – 34
negotiate traffic. Factor 3, labelled Vehicle control risks (α = 0.85)
A few – 25 accounted for 5.66% of the variance. Factor 3 contains four items (items
Many – 2 4, 1, 5, 6). This factor captures level 1 of the hierarchy and the risk-
A lot, lost – 1 increasing factors competency thus representing perceptions regarding
licence
the importance of personal risks related to basic vehicle manoeuvring.
I prefer not to – 0
say Combined, the first three factors explained the majority of variance
(63%) and encompassed about half of the total number of items (16 of
32).
The next largest factor (α = 0.90) accounted for 4.62% of the vari­
education (several response options dichotomised into ‘yes’ or ‘no’) and ance and was labelled Risks of driving in traffic. Factor 4 contains three
driving exposure (hours per week). Both novice drivers and parents were items (items 13, 14, 15). Factor 4 is consistent with the GDE’s compe­
asked about their crash involvement and given the response options tency of risk-increasing factors at level 2 of the matrix hierarchy. This
‘yes’ or ‘no’; parents were also able to respond to this item with ‘I prefer factor represents perceptions regarding the importance of driver edu­
not to say’. Parents only were asked to report their driving experience cation to facilitate an understanding of personal risks associated with
(years since obtaining licence) and demerit point accrual (None, Some, negotiating traffic. Similar to Factor 4, the fifth factor contained three
A few, Many, A lot, Lost licence, I prefer not to say). Novice drivers only items (items 20, 19, 27). This factor was labelled Context of driving –
were asked to report if they had been caught by the police for knowledge and skills (α = 0.92) and accounted for 3.52% of the variance.
committing a driving offence (yes, no). Factor 5 aligns with knowledge and skills at level 3 of the GDE matrix
representing perceptions regarding the importance of equipping novice
3. Results drivers with basic capabilities to plan ahead before driving. These two
factors were made up by 6 of the 32 items. Combined, factor 4 and factor
3.1. Sample characteristics 5 explained just over 8% of the variance.
The remaining factors each explained less than 3% of the variance.
Both novice driver and parent samples were convenience samples. Factor 6 (α = 0.90) comprised of four items (items 24, 21, 22, 23) and
Table 3 provides the demographic and driving history characteristics of accounted for 2.57% of the variance. This factor was labelled Risks in the
each of the samples. Most of the novice driver sample was aged 18 or 19 context of driving as it aligns with the GDE competency of risk-increasing
years. A smaller proportion of the novice driver sample had P1 licences factors at level 3 of the hierarchy; individuals who rate these items
than P2 licences. There were more females in both the novice driver and highly believe it is important for driver education to help novices un­
parent samples than males or those who identified with another gender derstand there are risks associated with the driving context. Factor 7 (α
identity. A larger proportion of the novice driver sample reported that = 0.83), Factor 8 (α = 0.76), and Factor 9 (α = 0.71), accounted for
they had attended prior driver education compared to the parent sam­ 2.45%, 2.13%, and 1.91% of the variance, respectively. These factors
ple. Most parents reported they had been involved in a crash while most were labelled as Vehicle control – knowledge and skills, Vehicle control –
novice drivers had not been in a crash. Most novice drivers had not been self-assessment, and Driving in traffic – knowledge and skills. Factor 7
caught by police for a driving offence while just over a third of parents included items 2 and 3 and corresponds with knowledge and skills at
reported that they had not received any demerit points since their child level 1 of the GDE matrix representing a belief that driver education

5
D. Rodwell et al. Accident Analysis and Prevention 191 (2023) 107190

Table 4
Factor structure of the perceptions of driver education scale.
Item Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6 Factor 7 Factor 8 Factor 9
Goals for Self- Vehicle Risks of Context of Risks in Vehicle Vehicle Driving in
life and assessment of control driving in driving – the context control control – self- traffic –
skills for driving in risks traffic knowledge and of driving knowledge assessment knowledge and
living traffic skills and skills skills

The learner driver can 1.003


identify their personal
level of risk-taking
The learner driver can 0.999
identify their personal
level of impulse control
The learner driver has 0.920
thought about how
sometimes they might
enjoy driving in an unsafe
way
The learner driver knows 0.915
about how their level of
self-control affects their
driving
The learner driver knows 0.728
that what their friends do
affects their driving
The learner driver knows 0.670
that risk is a part of
driving
The learner driver knows 0.608
that society values safe
drivers
The learner driver knows 0.586
about how their personal
values affect their driving
The learner driver can see 0.918
their own driving
tendencies
The learner driver has an 0.903
appreciation of their
limitations and abilities
to respond to unexpected
traffic situations
The learner driver has a 0.859
realistic idea of their own
driving skill in traffic
The learner driver can 0.500
understand their
strengths and weaknesses
at driving a car
The learner driver knows 0.798
about the best seating
position
The learner driver knows 0.789
how to position the car in
the lane
The learner driver knows it 0.749
is risky to not use my
seatbelt properly
The learner driver knows it 0.679
is risky to drive when it
was raining
The learner driver knows 0.855
about the risks associated
with breaking the road
rules
The learner driver knows 0.842
about the risks associated
with poor decision-
making
The learner driver knows 0.781
about the risks associated
with trying to do too
much while driving
The learner driver knows 0.939
how to estimate their
travel time when driving
(continued on next page)

6
D. Rodwell et al. Accident Analysis and Prevention 191 (2023) 107190

Table 4 (continued )
Item Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6 Factor 7 Factor 8 Factor 9
Goals for Self- Vehicle Risks of Context of Risks in Vehicle Vehicle Driving in
life and assessment of control driving in driving – the context control control – self- traffic –
skills for driving in risks traffic knowledge and of driving knowledge assessment knowledge and
living traffic skills and skills skills

The learner driver knows 0.932


the best way to choose
their travel route while
driving
The learner driver can 0.645
identify if they need to
plan their driving trips
more
The learner driver knows 0.776
about how having
passengers of their age in
their car while driving
affects them
The learner driver knows 0.729
about the risks of driving
while tired
The learner driver knows 0.687
about how their driving
might be different if their
mum or dad was in the
car with them
The learner driver knows 0.666
that becoming
competitive with other
drivers is dangerous
The learner driver knows 0.918
about centrifugal force
when turning
The learner driver knows 0.726
about the surface grip of
the car
The learner driver can 0.909
identify their abilities at
manoeuvring the car in
dangerous situations
The learner driver can 0.765
identify their abilities at
manoeuvring the car in
everyday situations
The learner driver knows 0.831
about traffic rules
The learner driver knows 0.484
how much room is
needed to leave between
them and the car in front
to be a safer driver

Note. Items removed were 11, 25, and 26.

should assist to increase knowledge and skills in controlling the vehicle. self-evaluation at level 3 of the framework. That is, beliefs that self-
Factor 8 included items 7 and 8 and aligns with self-evaluation skills at identification of one’s ability to plan ahead, and to understand when
level 1 of the GDE, the perception that novice drivers should be taught to motives for driving contribute to driving risk should be taught.
self-assess their ability to manoeuvre the vehicle. Finally, Factor 9
comprised of items 10 and 12. This factor encompasses knowledge and 3.2.3. Bivariate correlations between the factors and socio-demographic
skills at level 2 of the GDE, high scores on these items indicate the variables
perception that driver education should have a focus on knowledge and Bivariate correlations were assessed using the non-parametric
skills of negotiating traffic. Spearman’s correlation coefficient (See Table C1 in Appendix C). Cor­
Overall, the internal consistency, construct validity, and factor relations between each factor and parent socio-demographic variables
structure indicate that the scale is a useful measure of perceptions of including age, gender, prior traffic offences, involvement in a crash, and
aspects to focus on during driver education and training. It is important previous experience with formal driver education were evaluated
to note that items 11, 25, and 26 were removed from the scale. The finding few significant associations. There were no significant correla­
exclusion of item 11 removed a crucial element to focus on during driver tions between the Goals for Driver Education scale and age nor traffic
education: understanding how to predict the behaviour of other road offences. All significant correlations were small. There was a significant
users. While Factor 9 encompasses knowledge and skills concerning positive relationship between identifying as female and Factor 4 (Risks of
negotiating traffic, a core component of these knowledge and skills is the driving in traffic) and Factor 8 (Vehicle control – self-assessment). There
ability to anticipate the behaviour of other drivers in real-time traffic. was a significant positive relationship between previous crash experi­
Items 25 and 26 were intended to capture perceptions about the ence and Factor 4 (Risks of driving in traffic) and Factor 5 (Context of
importance for driver education to focus on novice drivers’ capacity for driving – knowledge and skills). Most factors were negatively associated

7
D. Rodwell et al. Accident Analysis and Prevention 191 (2023) 107190

Table 5
Means, standard deviations, and distribution of parent responses to the PDES items (n = 518).
Item Self-reported frequency M SD

Not important 2 Somewhat 4 Very important


important

Count Row N Count Row N Count Row N Count Row N Count Row N
% % % % %

Factor 1 Goals for life and skills for living 3.88 1.02
34 The learner driver can identify their personal level 26 5.0% 35 6.8% 81 15.6% 153 29.5% 223 43.1% 3.99 1.15
of risk-taking
33 The learner driver can identify their personal level 30 5.8% 43 8.3% 80 15.4% 152 29.3% 213 41.1% 3.92 1.19
of impulse control
35 The learner driver has thought about how 36 6.9% 58 11.2% 111 21.4% 135 26.1% 178 34.4% 3.70 1.24
sometimes they might enjoy driving in an unsafe
way
29 The learner driver knows about how their level of 15 2.9% 40 7.7% 82 15.8% 155 29.9% 226 43.6% 4.04 1.08
self-control affects their driving
30 The learner driver knows that what their friends do 31 6.0% 31 6.0% 72 13.9% 157 30.3% 227 43.8% 4.00 1.17
affects their driving
31 The learner driver knows that risk is a part of 17 3.3% 24 4.6% 84 16.2% 138 26.6% 255 49.2% 4.14 1.06
driving
32 The learner driver knows that society values safe 33 6.4% 44 8.5% 100 19.3% 149 28.8% 192 37.1% 3.82 1.20
drivers
28 The learner driver knows about how their personal 52 10.0% 63 12.2% 129 24.9% 149 28.8% 125 24.1% 3.45 1.26
values affect their driving
Factor 2 Self-assessment of driving in traffic 4.53 0.64
16 The learner driver can see their own driving 4 0.8% 7 1.4% 69 13.3% 137 26.4% 301 58.1% 4.40 0.83
tendencies
17 The learner driver has an appreciation of their 0.6% 5 1.0% 37 7.1% 109 21.0% 364 70.3% 4.59 0.72
limitations and abilities to respond to unexpected
traffic situations 3
18 The learner driver has a realistic idea of their own 3 0.6% 5 1.0% 34 6.6% 120 23.2% 356 68.7% 4.58 0.71
driving skill in traffic
9 The learner driver can understand their strengths 1 0.2% 7 1.4% 44 8.5% 131 25.3% 335 64.7% 4.53 0.72
and weaknesses at driving a car
Factor 3 Vehicle control risks 4.38 0.76
6 The learner driver knows about the best seating 8 1.5% 15 2.9% 85 16.4% 154 29.8% 256 49.4% 4.23 0.93
position
1 The learner driver knows how to position the car in 7 1.4% 10 1.9% 67 12.9% 113 21.8% 321 62.0% 4.41 0.88
the lane
5 The learner driver knows it is risky to not use my 18 3.5% 12 2.3% 45 8.7% 69 13.3% 374 72.2% 4.48 0.99
seatbelt properly
4 The learner driver knows it is risky to drive when it 6 1.2% 7 1.4% 65 12.5% 137 26.4% 303 58.5% 4.40 0.84
was raining
Factor 4 Risks of driving in traffic 4.56 0.74
14 The learner driver knows about the risks associated 10 1.9% 8 1.5% 47 9.1% 104 20.1% 349 67.4% 4.49 0.87
with breaking the road rules
13 The learner driver knows about the risks associated 5 1.0% 4 0.8% 38 7.3% 97 18.7% 374 72.2% 4.60 0.74
with poor decision-making
15 The learner driver knows about the risks associated 8 1.5% 6 1.2% 44 8.5% 86 16.6% 374 72.2% 4.57 0.82
with trying to do too much while driving
Factor 5 Context of driving – knowledge and skills 3.10 1.16
20 The learner driver knows how to estimate their 85 16.4% 98 18.9% 175 33.8% 103 19.9% 57 11.0% 2.90 1.22
travel time when driving
19 The learner driver knows the best way to choose 80 15.4% 86 16.6% 169 32.6% 118 22.8% 65 12.5% 3.00 1.23
their travel route while driving
27 The learner driver can identify if they need to plan 54 10.4% 72 13.9% 141 27.2% 124 23.9% 127 24.5% 3.38 1.28
their driving trips more
Factor 6 Risks in the context of driving 4.12 0.97
21 The learner driver knows about how having 23 4.4% 33 6.4% 72 13.9% 138 26.6% 252 48.6% 4.09 1.13
passengers of their age in their car while driving
affects them
22 The learner driver knows about the risks of driving 18 3.5% 23 4.4% 51 9.8% 113 21.8% 313 60.4% 4.31 1.05
while tired
23 The learner driver knows about how their driving 33 6.4% 45 8.7% 104 20.1% 176 34.0% 160 30.9% 3.74 1.17
might be different if their mum or dad was in the
car with them
24 The learner driver knows that becoming 22 4.2% 21 4.1% 46 8.9% 95 18.3% 334 64.5% 4.35 1.08
competitive with other drivers is dangerous
Factor 7 Vehicle control knowledge and skills 4.14 0.90
3 The learner driver knows about centrifugal force 10 1.9% 36 6.9% 103 19.9% 144 27.8% 225 43.4% 4.04 1.04
when turning
2 The learner driver knows about the surface grip of 7 1.4% 11 2.1% 88 17.0% 155 29.9% 257 49.6% 4.24 0.90
the car
Factor 8 Vehicle control – self-assessment 4.67 0.56
(continued on next page)

8
D. Rodwell et al. Accident Analysis and Prevention 191 (2023) 107190

Table 5 (continued )
Item Self-reported frequency M SD

Not important 2 Somewhat 4 Very important


important

Count Row N Count Row N Count Row N Count Row N Count Row N
% % % % %

7 The learner driver can identify their abilities at 2 0.4% 2 0.4% 23 4.4% 95 18.3% 396 76.4% 4.70 0.61
manoeuvring the car in dangerous situations
8 The learner driver can identify their abilities at 1 0.2% 5 1.0% 28 5.4% 110 21.2% 374 72.2% 4.64 0.65
manoeuvring the car in everyday situations
Factor 9 Driving in traffic – knowledge and skills 4.68 0.60
10 The learner driver knows about traffic rules 6 1.2% 6 1.2% 28 5.4% 80 15.4% 398 76.8% 4.66 0.74
12 The learner driver knows how much room is 1 0.2% 4 0.8% 26 5.0% 81 15.6% 406 78.4% 4.71 0.61
needed to leave between them and the car in front
to be a safer driver

with having attended formal driver education. abilities at manoeuvring the car in dangerous situations) scored the next
highest (M = 4.70; 76.4%).
3.3. Scale evaluation Factor 5 (Context of driving – knowledge and skills) scored the lowest,
on average, (M = 3.10) with 10–17% of participants perceiving these
3.3.1. Scale descriptive statistics – Parents of novice drivers skills as not important to be taught. The three items comprising this
The means, standard deviations, and response frequencies for the 32 factor (items 20, 19, 27) scored the lowest overall means. Research has
items included in the final EFA structure are reported in Table 5. Factors shown that relevant planning skills (beyond knowing the route) are
8 (Vehicle control – self-assessment) and 9 (Driving in traffic – knowledge often not taught to learner drivers yet are critical to safety when driving
and skill) had the largest means of 4.67 and 4.68 respectively yet independently (Boccara et al., 2015; Watson-Brown et al., 2020b). For
explained the smallest amount of variance in the scale. Factor 1 (Goals example, a novice driver who loses their way can often panic with the
for life and skills for living) had one of the lowest means (M = 3.88) yet associated emotions reducing their situation awareness and leading to
explained the largest amount of variance. Factors 8 and 9 are concise, driving behaviours that increase crash risk.
each consisting of only two items with a narrow focus on perceptions of
a specific aspect of driver education. Thus, it may not be entirely un­ 3.3.2. Confirmatory factor analysis – novice drivers
expected that these factors only capture small amounts of variance. A CFA was conducted on the novice driver’s dataset (n = 247) using
Alternatively, Factor 1 has the largest number of items capturing per­ the 32-item factor structure of the EFA established with the parents of
ceptions about a relatively broad array of aspects of driver education novice driver’s dataset. The CFA was assessed for good model fit by chi-
and training, as would be expected given the level of abstraction away square (non-significant), the goodness of fit index (GFI ≥ 0.95), the
from concrete driving behaviours that items targeted at level 4 of the comparative fit index (CFI ≥ 0.95), and the root mean-square error of
GDE should have. As such, it is logical that a factor which includes approximation (RMSEA ≤ 0.08) (Hoyle, 1995). The model developed
perceptions about a greater number of possible influences on driver from the EFA using the parent’s data was not a good fit to the novice
education would capture a large amount of variance. driver’s data (Table 6). Modification indices were used to guide
Item 12 (The learner driver knows how much room is needed to leave improvement to model fit and iterative removal of items. Table 6 sum­
between them and the car in front to be a safer driver) was the item with the marises the items that were removed and the corresponding goodness-
highest mean (4.71) and the largest frequency (78.4%) of parents of-fit indices.
perceiving this to be a very important skill to be taught to learner All items had high loadings (>0.70) except for item 23 (The learner
drivers. Indeed, research has found learner drivers are poor at main­ driver knows about how their driving might be different if their mum or dad
taining a safe distance to the vehicle in front (Nicolls et al., 2022; was in the car with them) which was subsequently removed (Step 1).
Watson-Brown et al., 2021a). Item 7 (The learner driver can identify their Covariances between errors within the same factor were examined and
covaried when > 20 (Steps 2 and 3). Standardised residual covariances
Table 6 were next assessed at 0.40 to consider further item removal. Item 6 (The
Goodness-of-fit indices and items removed from the Goals for Driver Education learner driver knows about the best seating position) and item 27 (The
scale developed from the parent’s data. learner driver can identify if they need to plan their driving trips more) were
removed. After removing these three items, model fit was judged to be
χ2 p GFI CFI RMSEA
(95% CI) tolerable. The final model contained 29 items and is illustrated in Fig. 2.
Full model (n = 32) 1290.48 <0.001 0.75 0.88 0.09 (0.09
to 0.10)
3.3.3. Descriptive statistics and Cronbach’s alphas – Novice drivers
Step 1: Removed one item 1194.47 <0.001 0.77 0.88 0.09 (0.08 The final factor structure of the PDES is visually mapped to the GDE
with loading < 0.70 to 0.10) in Table A1 in appendix A. Overall, the PDES is a valid and reliable scale
The learner driver knows about how their driving might be different if their mum or dad was to measure perceptions regarding the importance of the different com­
in the car with them
binations of hierarchical level and competency contained within the
Step 2: Covaried errors > 60 1115.85 <0.001 0.78 0.89 0.09 (0.08
to 0.09) GDE. The reliability of the scale was strong (α = 0.96). The internal
Step 3: Covaried errors > 20 1052.33 <0.001 0.79 0.90 0.08 (0.08 consistency of each factor was analysed using Cronbach’s alpha and
to 0.09) varied from 0.80 to 0.95 (Table 7). Descriptive statistics and frequencies
Step 4: Standardised residual 807.99 <0.001 0.82 0.92 0.07 (0.07 for the 29 items in the final novice driver scale structure are provided in
covariances > 0.40 – to 0.08)
removed two items
Table 7. Consistent with the findings in the parent responses, Factors 8
The learner driver knows about the best seating position (Vehicle control – self-assessment) and 9 (Driving in traffic – knowledge and
The learner driver can identify if they need to plan their driving trips more skill) had the largest means of 4.48 and 4.51 respectively. Factor 5
Note. p = significance value; GFI = Goodness of Fit Index; CFI = Comparative Fit (Context of driving – knowledge and skills) had the lowest mean of 3.17.
Index; RMSEA = Root Mean Square Error of Approximation. Item 10 (The learner driver knows about traffic rules) had the largest

9
D. Rodwell et al. Accident Analysis and Prevention 191 (2023) 107190

Fig. 2. The revised Perceptions of Driver Education scale model.

mean (4.60) and the greatest frequency of novice drivers asserting that acknowledging novices perceive that instruction about this skill is very
this skill was very important to be taught (73.3%). Given the tendency of important.
current on-road training to prioritise vehicle manoeuvring and traffic Conversely, item 20 (The learner driver knows how to estimate their
rules (Watson-Brown et al., 2020b) it is unsurprising that novice drivers travel time when driving) had the lowest mean (3.00) and frequency
would deem being taught traffic rules to be most important. Indeed, item (17.8%) of novice drivers responding that it was very important to
7 (The learner driver can identify their abilities at manoeuvring the car in include this skill in driver education. This was a similar finding in the
dangerous situations) scored the next highest (M = 4.52; 66.8%) parent data and, as mentioned, research has shown that navigation

10
D. Rodwell et al. Accident Analysis and Prevention 191 (2023) 107190

Table 7
Means, standard deviations, and distribution of novice driver responses to the PDES items (n = 247).
Items Self-reported frequency M SD

Not important 2 Somewhat 4 Very important


important

Count Row N Count Row N Count Row N Count Row N Count Row N
% % % % %

Factor 1 Goals for life and skills for living (α = 0.95) 3.78 0.98
34 The learner driver can identify their personal level 11 4.5% 24 9.7% 47 19.0% 80 32.4% 85 34.4% 3.82 1.14
of risk-taking
33 The learner driver can identify their personal level 8 3.2% 23 9.3% 56 22.7% 77 31.2% 83 33.6% 3.82 1.10
of impulse control
35 The learner driver has thought about how 23 9.3% 31 12.6% 64 25.9% 60 24.3% 69 27.9% 3.49 1.27
sometimes they might enjoy driving in an unsafe
way
29 The learner driver knows about how their level of 10 4.0% 20 8.1% 58 23.5% 83 33.6% 76 30.8% 3.79 1.09
self-control affects their driving
30 The learner driver knows that what their friends do 7 2.8% 21 8.5% 48 19.4% 78 31.6% 93 37.7% 3.93 1.08
affects their driving
31 The learner driver knows that risk is a part of 3 1.2% 14 5.7% 44 17.8% 79 32.0% 107 43.3% 4.10 0.97
driving
32 The learner driver knows that society values safe 6 2.4% 23 9.3% 49 19.8% 70 28.3% 99 40.1% 3.94 1.09
drivers
28 The learner driver knows about how their personal 25 10.1% 43 17.4% 59 23.9% 62 25.1% 58 23.5% 3.34 1.29
values affect their driving
Factor 2 Self-assessment of driving in traffic (α = 0.87) 3.97 0.77
16 The learner driver can see their own driving 1 0.4% 9 3.6% 52 21.1% 81 32.8% 104 42.1% 4.12 0.89
tendencies
17 The learner driver has an appreciation of their 0 0.0% 7 2.8% 42 17.0% 77 31.2% 121 49.0% 4.26 0.84
limitations and abilities to respond to unexpected
traffic situations
18 The learner driver has a realistic idea of their own 1 0.4% 6 2.4% 53 21.5% 82 33.2% 105 42.5% 4.15 0.87
driving skill in traffic
9 The learner driver can understand their strengths 0 0.0% 11 4.5% 41 16.6% 71 28.7% 124 50.2% 4.25 0.89
and weaknesses at driving a car
Factor 3 Vehicle control risks (α = 0.80) 4.22 0.81
1 The learner driver knows how to position the car in 2 0.8% 11 4.5% 50 20.2% 73 29.6% 111 44.9% 4.13 0.94
the lane
5 The learner driver knows it is risky to not use my 6 2.4% 14 5.7% 22 8.9% 53 21.5% 152 61.5% 4.34 1.02
seatbelt properly
4 The learner driver knows it is risky to drive when it 2 0.8% 15 6.1% 30 12.1% 88 35.6% 112 45.3% 4.19 0.93
was raining
Factor 4 Risks of driving in traffic (α = 0.90) 4.41 0.72
14 The learner driver knows about the risks associated 0 0.0% 7 2.8% 25 10.1% 73 29.8% 142 57.5% 4.42 0.79
with breaking the road rules
13 The learner driver knows about the risks associated 0 0.0% 6 2.4% 25 10.1% 71 28.7% 145 58.7% 4.44 0.77
with poor decision-making
15 The learner driver knows about the risks associated 0 0.0% 8 3.2% 31 12.6% 65 26.3% 143 57.9% 4.39 0.83
with trying to do too much while driving
Factor 5 Context of driving – knowledge and skills (α = 0.91) 3.17 1.24
20 The learner driver knows how to estimate their 39 15.8% 49 19.8% 75 30.4% 40 16.2% 44 17.8% 3.00 1.31
travel time when driving
19 The learner driver knows the best way to choose 28 11.3% 38 15.4% 55 22.3% 77 31.2% 49 19.8% 3.33 1.27
their travel route while driving
Factor 6 Risks in the context of driving (α = 0.86) 4.13 0.87
21 The learner driver knows about how having 5 2.0% 27 10.9% 48 19.4% 79 32.0% 88 35.6% 3.88 1.08
passengers of their age in their car while driving
affects them
22 The learner driver knows about the risks of driving 0 0.0% 11 4.5% 32 13.0% 72 29.1% 132 53.4% 4.31 0.86
while tired
24 The learner driver knows that becoming 5 2.0% 14 5.7% 33 13.4% 74 30.0% 121 49.0% 4.18 1.00
competitive with other drivers is dangerous
Factor 7 Vehicle control knowledge and skills (α = 0.80) 3.56 1.02
3 The learner driver knows about centrifugal force 14 5.7% 36 14.6% 75 30.4% 64 25.9% 58 23.5% 3.47 1.16
when turning
2 The learner driver knows about the surface grip of 8 3.2% 27 10.9% 71 28.7% 78 31.6% 63 25.5% 3.65 1.07
the car
Factor 8 Vehicle control – self-assessment (α = 0.81) 4.48 0.72
7 The learner driver can identify their abilities at 0 0.0% 8 3.2% 20 8.1% 54 21.9% 165 66.8% 4.52 0.78
manoeuvring the car in dangerous situations
8 The learner driver can identify their abilities at 0 0.0% 7 2.8% 27 10.9% 65 26.3% 148 59.9% 4.43 0.80
manoeuvring the car in everyday situations
Factor 9 Driving in traffic – knowledge and skills (α = 0.84) 4.51 0.73
10 The learner driver knows about traffic rules 1 0.4% 6 2.4% 17 6.9% 42 17.0% 181 73.3% 4.60 0.76
12 The learner driver knows how much room is 0 0.0% 9 3.6% 23 9.3% 69 27.9% 146 59.1% 4.42 0.81
needed to leave between them and the car in front
to be a safer driver

11
D. Rodwell et al. Accident Analysis and Prevention 191 (2023) 107190

Table A1 Table B1
The twelve aspects of focus for driver education within the Goals for Driver Comparison of Queensland Graduated Driver Licensing system stages.
Education (GDE) with examples and indication of factor structure in the PDES. Learner (L) Provisional 1 (P1) Provisional 2 (P2) Open
Knowledge and Risk-increasing Self-evaluation
Minimum 16 Minimum 17 Minimum 18 Minimum 20
skills factors skills
years age years age years age years age
Level 4: Goals Knowledge of Cultural norms about Ability to
for life and GDL restrictions driving/non-driving recognise and Pass theory test Pass practical Pass online hazard Successful
skills for control impulses to progress to driving test to perception test to progression
living learner licence progress from progress from P1 through Learner,
Factor 1 Factor 1 Factor 1 learner licence licence P1 and P2 licence
Level 3: Goals Trip planning Risks associated with Awareness of phase
and contexts knowledge; driver condition (e. personal planning
for driving route choice g., mood, skills/time Decreased Decreased number Decreased number Open license
medication) management number of of demerit points. of demerit points. drivers can accrue
skills demerit points. Maximum of 4 Maximum of 4 a maximum of 12
Factor 5 Factor 6 Maximum of 4 points can be points can be Demerit points in
Level 2: Knowledge of Driving skill in Awareness of points can be accrued in a accrued in a three-year
Mastery of road rules; safety relation to weather attitudes towards accrued in a continuous one- continuous one- period
traffic margins conditions road rules one-year year period year period
situations period
Factor 9 Factor 4 Factor 2 Display licence Display licence Display licence Not applicable
Level 1: Vehicle Psychomotor Insufficient Realistic self- plates: Black plates: Red ‘P’ on plates: Green ‘P’
manoeuvring driving skills (e. automatization of evaluation about ‘L’ on yellow white background on white
g., changing psychomotor skills vehicle handling background background
gears) for operating the capability
vehicle Must complete Unsupervised Unsupervised Unsupervised
Factor 7 Factor 3 Factor 8 100 h fully independent independent independent
Note. PDES = Perceptions of Driver Education Scale; table adapted from Rodwell et al. supervised driving allowed; driving allowed; driving allowed;
(2018) practice no logbook no logbook no logbook
recorded in a required. required. required.
logbook
skills, including the importance of pre-trip planning, is often the last skill
Restrictions on Restrictions on Restrictions on Restrictions on
to be transferred from a driver trainer to the learner driver (Boccara
BAC (0.0), driving with peer BAC (0.0); limits BAC (<0.05);
et al., 2015; Watson-Brown et al., 2020b). additional passengers at on vehicle restrictions on
restrictions on night, BAC (0.0), horsepower; all mobile phone use;
4. Discussion mobile phone additional other conditions other conditions
use; all other restrictions on applicable to open may be applied (e.
conditions mobile phone use, licence g., requirement to
The PDES was built using the GDE (Hatakka et al., 2002) and a applicable to limits on vehicle wear corrective
previous study (Bates et al., 2019) as its foundation. The purpose of the open licence horsepower; all lenses)
PDES is to gain the views of individuals regarding their preferences other conditions
about aspects of professional driver education. Subsequently, this pref­ applicable to open
licence
erence data may lead to driver education that is more likely to be
accepted and therefore more effective. 12-month 12-month 24-month Ongoing
The initial pool of 35 items was developed conceptually to capture minimum minimum minimum
the cross-section of each competency at each hierarchical level of the duration duration duration
GDE. In developing the items this way, a theoretical 12-factor solution
Note. BAC = Blood/Breath Alcohol Concentration. Further information about the
was suggested. The fact that no items from Bates et al., (2019) were Queensland GDL can be found here: https://www.qld.gov.au/transport/licensin
included in the PDES without amendment suggests that developing and g/getting/young-drivers
validating a scale is critical. The final 29-item 9 factor solution was
found to have tolerable model fit and displayed a high degree of internal
consistency, both overall and within each of the factors. As visualised in investigating perceptions of the risk increasing factors competency
Table A1 in appendix A, the structure of the final 9-factor solution was loaded with one item investigating the knowledge and skills competency
judged to have a general alignment with the hypothetical 12-factor so­ at the same hierarchical level. This suggests there may be a strong
lution based on the intersection of competency and GDE level. connection in perceptions about what needs to be taught about risk and
In five of the nine factors in the final solution, items measuring the other knowledge and skills; but little interplay between these concepts
same competency and GDE level loaded together (i.e., factors 4, 5, 7, 8, and what needs to be taught in driver education and training with
9). Only one factor had items load that were designed to capture per­ respect to self-assessment.
ceptions about driver education needs on different GDE levels. In this Two further aspects of the final solution deserve comment. First, all
case, one item aiming to capture perceptions of the importance of eight items aimed at measuring perceptions of driver education aspects
teaching novice drivers self-evaluation skills related to their vehicle on GDE level 4 loaded onto a single factor. As such, there is a lack of
manoeuvring abilities (level 1) loaded with three items that measured specificity in terms of what the PDES measures regarding perceptions of
the perceived importance of training self-evaluation skills related to what should be taught to novice drivers that is related to less concrete
their mastery of traffic situations (level 2). This suggests that there may influences on driving. That is, an individual who scores highly on factor
be some overlap in perceptions about driver education and training for 1 perceives that it is generally important in driver education and
the two most basic GDE levels, at least in relation to training to provide training to teach young people about Goals for life and skills for living; it
novices with self-evaluation skills. Two factors (i.e., 3, 6) had items that does not specify that a focus of the teaching at this level should be on
were intended to investigate perceived importance of teaching novices’ knowledge and skills, risk increasing factors, or self-evaluation skills.
aspects of the knowledge and skills competency load with items inten­ Second, all three items intended to measure perceptions concerning self-
ded to measure perceived importance of teaching novices’ aspects of the evaluation skills at GDE level three were removed from the analyses.
risk-increasing factors competency. In both factors, two items Because of this a crucial element to focus on in driver education and

12
D. Rodwell et al. Accident Analysis and Prevention 191 (2023) 107190

Table C1
Correlations among the GDE scale and self-reported demographics and driving measures (n = 518).
1 2 3 4 5 6 7 8 9 10 11 12 13 14
a
1. Factor 1 1.00
2. Factor 2a 0.55** 1.00
3. Factor 3a 0.52** 0.58** 1.00
4. Factor 4a 0.56** 0.63** 0.61** 1.00
5. Factor 5a 0.75** 0.39** 0.48** 0.43** 1.00
6. Factor 6a 0.80** 0.48** 0.54** 0.54** 0.67** 1.00
7. Factor 7a 0.42** 0.49** 0.64** 0.40** 0.35** 0.38** 1.00
8. Factor 8a 0.38** 0.63** 0.51** 0.52** 0.28** 0.31** 0.38** 1.00
9. Factor 9a 0.41** 0.55** 0.61** 0.62** 0.37** 0.48** 0.36** 0.52** 1.00
10. Age − 0.02 − 0.02 0.04 0.03 0.03 − 0.03 0.08 − 0.04 0.01 1.00
11. Genderb 0.00 0.04 − 0.01 0.10* − 0.01 0.01 − 0.02 0.11** 0.05 − 0.15** 1.00
12. Offences − 0.02 − 0.03 − 0.02 − 0.05 0.01 − 0.01 − 0.06 − 0.03 0.00 − 0.00 − 0.07 1.00
13. Crash – yes/ 0.05 0.05 0.11* 0.10* 0.05 0.07 0.03 0.05 0.07 0.01 − 0.00 − 0.05 1.00
no
14. Formal − 0.10* − 0.10* − 0.14** − 0.03 − 0.09* − 0.09* − 0.12* − 0.08 − 0.04 − 0.05 0.19** − 0.05 − 0.05 1.00
driver
education –
yes/no

Note. *p <.05; **p <.01; aFactor 1 - Goals for life and skills for living; Factor 2 - Self-assessment of driving in traffic; Factor 3 - Vehicle control risks; Factor 4 - Risks of driving in
traffic; Factor 5 - Context of driving – knowledge and skills; Factor 6 - Risks in the context of driving; Factor 7 - Vehicle control knowledge and skills; Factor 8 - Vehicle control –
self-assessment; Factor 9 - Driving in traffic – knowledge and skills; bMale = 0; Non-parametric Spearman’s correlation coefficient was used.

training, the need to assist novice drivers to develop personal insight test–retest reliability of the PDES and its ability to predict safe driving
about important driving-related contexts and skills (e.g., trip planning), behaviour.
is not captured within the PDES. Future research should focus on
development of items that specifically capture perceptions of this 5. Conclusion
element of driver education and training.
The findings thus suggest that the PDES is a valid and reliable tool for The GDE is considered the optimal guide for developing appropriate
use with both parents and novice drivers. Further testing will be education and training interventions for the novice driver group (Bailey
required to assess if the PDES is similarly valid for use in investigating et al., 2022). Despite the important role this framework has played in
the perceptions of other important stakeholders such as professional numerous research studies (e.g., Molina et al., 2014; Tronsmoen, 2010;
driving trainers or those involved in developing novice driver govern­ Watson-Brown et al., 2018), there has been no standardised develop­
ment policy. Further refinement of items or the addition of new ment of a scale that can be used to operationalise the GDE, an issue noted
conceptually developed items may assist to improve the model fit of by Rodwell et al. (2018). The present study addresses this gap by
future versions of the PDES. Finally, while the focus of the PDES in the following best practice approaches for scale development (Boateng
current article was on perceptions of the GDE that could be collected et al., 2018; Hinkin, 1995; Hinkin et al., 1997) to conceptualize and test
prior to the design of driver education and training and used to aid the a scale with two population groups that are closely associated with
development of such training programs, it is also speculatively possible novice drivers. Overall, the findings suggest that the PDES is a valid and
that the PDES could be used to evaluate perceptions of existing driver reliable scale to assist providers and other interested stakeholders to
education programs. For example, after completing a driver education design and develop driver education and training initiatives that will be
program, novice drivers could be asked to reflect on their experiences accepted by key stakeholder groups. In turn, this may help to make
and complete the PDES. However, it is noted that the validity and reli­ driver education and training more acceptable to young people and
ability of using the PDES in such a way would need to be examined in effective at reducing road crashes in this cohort.
future research.
Declaration of Competing Interest
4.1. Limitations
The authors declare that they have no known competing financial
It is important to note that there are some limitations to the study. interests or personal relationships that could have appeared to influence
First, it needs to be acknowledged that the scale development relied on the work reported in this paper.
convenience samples of novice drivers and parents of novice drivers. As
such, no claim to representativeness of the novice driver or parent Data availability
populations can be made. Additionally, it is noted that participants
identifying as female gender were over-represented in both samples. Data will be made available on request.
However, few relationships were found between gender, the other socio-
demographic and driving history variables, and the factors in the PDES. Acknowledgements
Second, while this study followed the steps required to develop a
valid and reliable scale, time and resourcing constraints prevented The research team would like to acknowledge the assistance of the
further validation of the scale. For instance, a limitation of this study was following researchers who assisted with the larger research project that
that there was no formal assessment of content validity. Ideally, an this study was conducting as a part of: Assoc Prof. Gregoire Larue, Prof
assessment of content validity should occur to ensure that the items Narelle Haworth, Prof Barry Watson, Prof Ashleigh Filtness, Ms Alana
adequately measure the scale domain. This assessment of face validity Hawkins, and Mr Levi Anderson.This research was supported by the
could occur through the use of a Delphi approach with experts or in­ Australian Government through the Australian Research Council’­
terviews with a sample of those from the population who will complete s Linkage Projects funding scheme (project LP140100409). The
the survey (Boateng et al., 2018). Future research could also examine the Australian Research Council was not involved in conducting the

13
D. Rodwell et al. Accident Analysis and Prevention 191 (2023) 107190

research. This research was supported (partially or fully) by the Motor Ivers, R.Q., Hunter, K., Clapham, K., Helps, Y., Senserrick, T., Byrne, J., Martiniuk, A.,
Daniels, J., Harrison, J.E., 2016. Driver licensing: descriptive epidemiology of a
Accident Insurance Commission (MAIC) Queensland. The views
social determinant of Aboriginal and Torres Strait Islander health. Aust. N. Z. J.
expressed herein are those of the authors and are not necessarily those of Public Health 40 (4), 377–382. https://doi.org/10.1111/1753-6405.12535.
the MAIC. Ker, K., Roberts, I., Collier, T., Beyer, F., Bunn, F., Frost, C., 2005. Post-licence driver
education for the prevention of road traffic crashes: a systematic review of
randomised controlled trials. Accid. Anal. Prev. 37 (2), 305–313. https://doi.org/
Appendix A 10.1016/j.aap.2004.09.004.
Keskinen, E., 2007. What is GDE all about and what it is not. In: Henriksson, W.,
Table A1 Stenlund, T., Sundström, A., Wiberg, M. (Eds.), The GDE-model as a guide in driver
training and testing.: Proceedings from the conference, Umeå, May 7-8, 2007. Umea
Universitet.
Appendix B Lonero, L., Mayhew, D., 2010. Teen Driver Safety: Large-Scale Evaluation of Driver
Education Review of The Literature On Driver Education Evaluation 2010 Update.
AAA Foundation for Traffic Safety.
Table B1 Lund, A.K., Williams, A.F., Zador, P., 1986. High school driver education: Further
evaluation of the Dekalb County study. Accid. Anal. Prev. 18 (4), 349–357. https://
Appendix C doi.org/10.1016/0001-4575(86)90048-5.
Mayhew, D.R., 2007. Driver education and graduated licensing in North America: Past,
present, and future. J. Saf. Res. 38 (2), 229–235. https://doi.org/10.1016/j.
Table C1 jsr.2007.03.001.
Mayhew, D., Simpson, H., Williams, A., Ferguson, S., 1998. Effectiveness and role of
driver education and training in a Graduated Licensing System. J. Public Health
References
Policy 19 (1), 51–67. https://doi.org/10.2307/3343089.
Mayhew, D., Simpson, H., Robinson, A., 2002. The safety value of driver education an
Bailey, T., Wundersitz, L., O’Donnell, K., Rasch, A., 2022. Identifying best practices in a training. Inj. Prev. 8 (Suppl 2), ii3-ii8. https://doi.org/10.1136/ip.8.suppl_2.ii3.
process evaluation of a novice driver education program. Eval. Program Plann. 93, Molina, J., García-Ros, R., Keskinen, E., 2014. Implementation of the driver training
102105 https://doi.org/10.1016/j.evalprogplan.2022.102105. curriculum in Spain: An analysis based on the Goals for Driver Education (GDE)
Bates, L., Watson, B., King, M.J., 2014. The role of parents and non-parents in the framework. Transport. Res. F: Traffic Psychol. Behav. 26, Part A, 28–37. https://doi.
supervision of learner drivers in Australia. Accid. Anal. Prev. 70, 40–45. org/10.1016/j.trf.2014.06.005.
Bates, L., Hawkins, A., Rodwell, D., Anderson, L., Watson, B., Filtness, A.J., Larue, G.S., National Highway Traffic Safety Administration. (2021). Early eestimates of motor vehicle
2019. The effect of psychosocial factors on perceptions of driver education using the traffic fatalities and fatality rate by sub-categories in 2020. https://crashstats.nhtsa.dot.
goals for driver education framework. Transport. Res. Part F, Traff. Psychol. Behav. gov/Api/Public/ViewPublication/813118.
66, 151–161. https://doi.org/10.1016/j.trf.2019.09.004. Nicolls, M., Truelove, V., Watson-Brown, N., 2022. Self-regulation and tailgating
Beanland, V., Goode, N., Salmon, P.M., Lenné, M.G., 2013. Is there a case for driver behaviours among young drivers. Transport. Res. F: Traffic Psychol. Behav. 89,
training? A review of the efficacy of pre- and post-licence driver training. Saf. Sci. 51 347–354. https://doi.org/10.1016/j.trf.2022.07.007.
(1), 127–137. https://doi.org/10.1016/j.ssci.2012.06.021. Panou, M., Bekiaris, E., Papakostopoulos, V., 2007. Modelling driver behaviour in
Berg, H.Y., 2006. Reducing crashes and injuries among young drivers: what kind of European Union and international projects. In: Cacciabue, P.C. (Ed.), Modelling
prevention should we be focusing on? Inj. Prev. 12 (Suppl 1), i15–i18. https://doi. Driver Behaviour in Automotive Environments. Springer London, London, pp. 3–25.
org/10.1136/ip.2006.012062. Peden, A.E., Cullen, P., Francis, K.L., Moeller, H., Peden, M.M., Ye, P., Tian, M., Zou, Z.,
Boateng, G.O., Neilands, T.B., Frongillo, E.A., Melgar-Quiñonez, H.R., Young, S.L., 2018. Sawyer, S.M., Aali, A., Abbasi-Kangevari, Z., Abbasi-Kangevari, M.,
Best practices for developing and validating scales for health, social, and behavioral Abdelmasseh, M., Abdoun, M., Abd-Rabu, R., Abdulah, D.M., Abebe, G., Abebe, A.
research: A primer. Front. Public Health 6. https://doi.org/10.3389/ M., Abedi, A., Ivers, R.Q., 2022. Adolescent transport and unintentional injuries: a
fpubh.2018.00149. systematic analysis using the Global Burden of Disease Study 2019. Lancet Public
Boccara, V., Vidal-Gomel, C., Rogalski, J., Delhomme, P., 2015. A longitudinal study of Health 7 (8), e657–e669.
driving instructor guidance from an activity-oriented perspective. Appl. Ergon. 46, Peraaho, M., Keskinen, E., Hatakka, M., 2003. Driver Competence In A Hierarchical
21–29. https://doi.org/10.1016/j.apergo.2014.06.001. Perspective; Implications For Driver Education. University of Turku, Traffic
Bradshaw, C., Turner, B., Makwasha, T., & Cairney, P. (2015). Road fatalities and serious Research.
injuries in Australia and New Zealand 2001-10 (Austoads, Ed.). Austroads. Roberts, I., Kwan, I., 2001. School based driver education for the prevention of traffic
Christie, R. (2001). The effectiveness of driver training as a road safety measure: an crashes. Cochrane Database Syst. Rev. 3 https://doi.org/10.1002/14651858.
international review of the literature Australasian Road Safety Research, Policing and cd003201. Cd003201.
Education Conference Melbourne, Victoria. http://acrs.org.au/files/arsrpe/ Rodwell, D., Hawkins, A., Haworth, N., Larue, G.S., Bates, L., Filtness, A., 2018. A mixed-
RS010018.pdf. methods study of driver education informed by the Goals for Driver Education: Do
Collins, J.L., Thomas, L.J., Vugrin, M., 2021. Life is a highway: Driver’s license young drivers and educators agree on what was taught? Saf. Sci. 108, 140–148.
acquisition and social determinants for youth in foster care and their non-foster care https://doi.org/10.1016/j.ssci.2018.04.017.
peers. Nurs. Forum 56 (1), 95–102. https://doi.org/10.1111/nuf.12517. Rodwell, D., Larue, G.S., Bates, L., Haworth, N., 2020. What, who, and when? The
Cuenen, A., Wets, G., Ruiter, R.A.C., Brijs, T., Brijs, K., 2014. Evaluating the effectiveness perceptions that young drivers and parents have of driving simulators for use in
of a post-license education program for young novice drivers in Belgium. Accid. driver education. Safety (Basel) 6 (4), 46. https://doi.org/10.3390/safety6040046.
Anal. Prev. 66, 62–71. https://doi.org/10.1016/j.aap.2014.01.015. Rodwell, D., Alexander, M., Bates, L., Larue, G.S., Watson, B., 2021. Parents’ perceptions
Elvik, R., Hoye, A., Vaa, T., Sorenson, M., 2009. Handbook of Road Safety Measures, 2nd of driver education: A theoretically guided qualitative investigation. Transport. Res.
ed. Emerald Inc. Part F, Traff. Psychol. Behav. 77, 293–311. https://doi.org/10.1016/j.
Fisher, D.L., Dorn, L., 2016. The training and education of novice teen drivers. In: trf.2021.01.009.
Fisher, D.L., Caird, J., Horrey, W., Trick, L. (Eds.), Handbook of Teen and Novice Tronsmoen, T., 2010. Associations between driver training, determinants of risky driving
Drivers : Research, Practice, Policy, and Directions. CRC Press, pp. 270–289. behaviour and crash involvement. Saf. Sci. 48 (1), 35–45. https://doi.org/10.1016/j.
Glendon, A.I., McNally, B., Jarvis, A., Chalmers, S.L., Salisbury, R.L., 2014. Evaluating a ssci.2009.05.001.
novice driver and pre-driver road safety intervention. Accid. Anal. Prev. 64, Watson-Brown, N., Scott-Parker, B., Senserrick, T., 2018. Development of a higher-order
100–110. https://doi.org/10.1016/j.aap.2013.11.017. instruction coding taxonomy for observational data: Initial application to
Hatakka, M., Keskinen, E., Gregersen, N.P., Glad, A., Hernetkoski, K., 2002. From control professional driving instruction. Appl. Ergon. 70, 88–97.
of the vehicle to personal self-control; broadening the perspectives to driver Watson-Brown, N., Scott-Parker, B., Senserrick, T., 2019. Association between higher-
education. Transport. Res. F: Traffic Psychol. Behav. 5 (3), 201–215. https://doi.org/ order driving instruction and risky driving behaviours: Exploring the mediating
10.1016/S1369-8478(02)00018-9. effects of a self-regulated safety orientation. Accid. Anal. Prev. 131, 275–283.
Haworth, N., Kowadlo, N., Tingvall, C., 2000. Evaluation of pre-driver education Watson-Brown, N., Scott-Parker, B., Senserrick, T., 2020a. Higher-order driving
program. Monash University Accident Research Centre. instruction and opportunities for improvement: Exploring differences across learner
Henriksson, W., Stenlund, T., Sundström, A., & Wiberg, M. (2007). The GDE-model as a driver experience. J. Saf. Res. 75, 67–77.
guide in driver training and testing.: Proceedings from the conference, Umeå, May 7- Watson-Brown, N., Scott-Parker, B., Simons-Morton, B., Senserrick, T., 2020b. Exploring
8, 2007. the dimensions of driving instruction through naturalistic observation of formal
Hinkin, T.R., 1995. A review of scale development practices in the study of practical lessons with learner drivers. Transp. Res. Rec. 2674 (3), 219–231.
organizations. J. Manag. 21 (5), 967–988. https://doi.org/10.1016/0149-2063(95) Watson-Brown, N., Mills, L., Senserrick, T., Freeman, J., Davey, J., Scott-Parker, B.,
90050-0. 2021a. A complex system of learning to drive: The instructor’s perspective. Saf. Sci.
Hinkin, T.R., Tracey, J.B., Enz, C.A., 1997. Scale construction: developing reliable and 136, 105172 https://doi.org/10.1016/j.ssci.2021.105172.
valid measurement instruments. J. Hospital. Tour. Res. (Washington D.C.) 21 (1), Watson-Brown, N., Scott-Parker, B., Senserrick, T., 2021b. Higher order training
100–120. https://doi.org/10.1177/109634809702100108. supporting competence, autonomy, relatedness (HOT-CAR): A model to improve
Horswill, M.S., Hill, A., Rodwell, D., Larue, G.S., Bates, L., Watson, B., 2021. A brief and learner drivers’ higher order skills. Transport. Res. F: Traffic Psychol. Behav. 80,
unsupervised online intervention improves performance on a validated test of hazard 79–89.
perception skill used for driver licensing. Transport. Res. F: Traffic Psychol. Behav.
78, 130–136. https://doi.org/10.1016/j.trf.2021.02.003.

14

You might also like