Accident Analysis and Prevention 73 (2014) 5364

Contents lists available at ScienceDirect

Accident Analysis and Prevention

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

The effect of family climate on risky driving of young novices: The

moderating role of attitude and locus of control
Aline Carpentier a, , Kris Brijs a,b , Katrien Declercq a , Tom Brijs a , Stijn Daniels a ,
Geert Wets a
a
Hasselt University, Transportation Research Institute (IMOB), Wetenschapspark 5, bus 6, BE-3590 Diepenbeek, Belgium
b
Faculty of Applied Engineering Sciences, Agoralaanbuilding H, BE-3590 Diepenbeek, Belgium

a r t i c l e i n f o a b s t r a c t

Article history: The aim of the study was to examine the relative importance of young novice drivers family climate
Received 19 March 2014 on their driving behavior. A sample of young novice drivers (N = 171) between the age of 17 and 24,
Received in revised form 28 July 2014 who held their permanent (or temporary) drivers license for no longer than one year, participated. The
Accepted 5 August 2014
questionnaire included items related to the participants family climate, 3 socio-cognitive determinants
(i.e., attitude, locus of control and social norm), and risky driving behaviors. We expected both fam-
Keywords:
ily climate and the socio-cognitive determinants to exert a direct effect on risky driving. Furthermore
Family climate
we hypothesized that the socio-cognitive determinants would moderate the impact of family climate
Parental inuence
Socio-cognitive factors
on risky driving. The results showed that the effect of family climate on risky driving only originated
Risky driving from one single factor (i.e., noncommitment). Besides that, the results conrmed the importance of the
three socio-cognitive determinants to the degree that attitude, locus of control, and social norm signif-
icantly predicted the self-reported risky driving. In line of what we hypothesized, attitude moderated
the relationship between noncommitment and risky driving. Lastly, we found an unexpected three-way
interaction which indicated that locus of control moderated the relation between noncommitment and
risky driving only when young drivers attitude was risk-supportive. We recommend scholars and practi-
tioners to take into account the interaction between external sources of inuence (such as an individuals
family climate) and more personally oriented dispositions (such as an individuals attitude, social norm
and locus of control) when trying to explain and change young novices risky driving.
2014 Elsevier Ltd. All rights reserved.

1. Introduction specically the inuence of parents on the risky driving of their

1.1. Parent-offspring socialization
As can be derived from Banduras social learning theory
(Bandura, 1986), behavior can be modied by learning from others.
Such learning can be done directly (for instance, through verbal per-
suasion), as well as indirectly (i.e., through vicarious experiences)
and thus observing the behavior of others. Over the last few years
we see an increase of interest in these social inuences, and more
Corresponding author.
Tel.: +32 0 11 26 91 {17, 29, 61, 55, 56, 58}/+32 0 11 37 07 77;
fax: +32 0 11 26 91 99.
E-mail addresses: aline.carpentier@uhasselt.be (A. Carpentier),
kris.brijs@uhasselt.be (K. Brijs), katrien declercq@uhasselt.be (K. Declercq),
tom.brijs@uhasselt.be (T. Brijs), stijn.daniels@uhasselt.be (S. Daniels),
geert.wets@uhasselt.be (G. Wets).
children.
Because of their educative responsibility, parents can inu-
ence their childrens driving behavior through family socialization
(Taubman-Ben-Ari et al., 2005). Parent-offspring socialization has
been explored for a variety of trafc safety related behaviors (Lam,
2001; Loubeau, 2000; Morrongiello and Barton, 2009), but mainly
in samples of young novice drivers (Bianchi and Summala, 2004;
Ferguson et al., 2001; Hartos et al., 2002; Miller and Taubman-
Ben-Ari, 2010; Scott-Parker et al., 2009; Shope et al., 2001;
Simons-Morton et al., 2002; Taubman-Ben-Ari and Katz-Ben-Ami,
2012). An important nding within this research domain is that
violation and driving records of parents translate into similar driv-
ing records of their children (Brookland et al., 2014; Ferguson et al.,
2001; Wilson et al., 2006). Ferguson et al. (2001) indicated that
the driving records of children are related to those of their par-
ents, since children were more likely to be involved in at least one
crash if their parents had more than 3 crashes. Not only crash and
violation records but also driving styles transmit from parent to

http://dx.doi.org/10.1016/j.aap.2014.08.005
0001-4575/ 2014 Elsevier Ltd. All rights reserved.
54 A. Carpentier et al. / Accident Analysis and Prevention 73 (2014) 5364
child. Miller and Taubman-Ben-Ari (2010) demonstrated that the
driving style of parents and their children correlate signicantly.
Bianchi and Summala (2004) found indications for parent-offspring
behavior transmissions, since they concluded that the self-reported
driving behavior of parents explained the self-reported behavior of
their children. In similar vein, the study of Prato et al. (2010) found
an association between the risk taking behavior of young novice
drivers on the one hand and their parents driving behavior, the
amount of supervised driving and the level of parental monitoring
on the other hand. Besides supervision and monitoring, there are
several other factors that underlie a parent-child connection such
as for instance modeling of driving behavior or parental teaching
and communication skills (Schmidt et al., 2014; Goodwin et al.,
2014). A longitudinal study by Shope et al. (2001) discovered that
negative inuence of parents such as the levels of parental monitor-
ing, family connectedness, nurturing and lenient attitude toward
young peoples drinking, increased young drivers risk of collision
and driving violations. A study by Hartos et al. (2002) demonstrated
that risky driving of young drivers at the follow up measurement
was, among others, predicted by parental restrictions. In another
light, Scott-Parker et al. (2009) revealed that instead of parental
restrictions, it was anticipated parental reward that signicantly
explained youngsters risky driving.
Although the impact that parents have on their childrens driv-
ing behavior has been made evident, further research on how
these parental driving behaviors are transmitted is still required.
The major objective of this study is to examine the underly-
ing mechanisms of parental inuence and its impact relative to
other determining (intrinsic or environmental) factors, since these
aspects provide a better understanding of the process of parent-
child socialization in trafc safety.
1.2. Family safety climate
Social learning, intergenerational, and socialization theories
have placed signicant emphasis on familial processes of behavior
transmission and focused more particularly on parents as offspring
role models (Schneewind, 1999; Taubman-Ben-Ari et al., 2005).
Starting from the idea that parenting practices are strongly embed-
ded in a somewhat broader family context, Taubman-Ben-Ari and
Katz-Ben-Ami (2013) recently added the concept of family safety
climate to the literature on parenting. In an exploratory study,
Taubman-Ben-Ari and Katz-Ben-Ami (2013) proposed 7 domains
that shape the family climate for road safety (FCRS) concept.
Together, these 7 dimensions cover the 2 basic mechanisms of direct
and indirect parental behavior transmission. This follows from the
seminal work on social learning by Bandura (1986), in which these
two types of learning processes are elucidated. In case of the FCRS
concept, 2 dimensions (i.e., modeling and noncommitment) can
be related to indirect social learning processes, and 5 dimensions
(i.e., monitoring, limits, feedback, messages, and communication)
to direct social learning processes.
With regard to the direct learning processes it can be noted that
there are some factors that are to some extent related to each other.
Although the factors feedback and messages closely relate to each
other, they address two different aspects of parentchild interac-
tion. On the one hand, feedback refers to the encouragement and
positive comment parents provide on their childrens driving. Mes-
sages on the other hand, relates to parents explicitly verbalizing
their own personal opinion on the importance of road safety, also
outside the context of discussing or evaluating their childrens driv-
ing behavior. Two other related factors are monitoring and limits.
While monitoring relates to the level up to which parents supervise
their childrens driving behavior, the factor limits stands for the
extent to which parents really set clear restrictions. The relevance
of these two factors is derived from studies showing that parental
monitoring and imposing proper restrictions (e.g., night time driv-
ing, driving with peer passengers) both have an inuence on the
driving behavior of adolescents (Beck et al., 2001; Hartos et al.,
2000, 2002; Shope et al., 2001). Interestingly, for driving restric-
tions to be effective, it is important that parents and their children
are in agreement with each other (Beck et al., 2001). In addition, it
appears to be best to actively involve children in negotiating driving
limits since low autonomy-supportive parental environments have
been shown to associate with strong extrinsic aspirations, with
the latter being signicantly related to an increased propensity to
engage in high-risk behavior (Williams et al., 2000).
The factor communication in turn, stands for still another qual-
itative aspect of direct parentchild interaction, namely, the level
at which they maintain open and direct contact with each other.
The importance of open and direct parentchild communication for
the prevention of risky behavior has been assessed recurrently for
a multitude of impaired health-related behaviors such as smoking,
substance abuse, unsafe sex, etcetera (Harakeh et al., 2005). Also,
as demonstrated by Sherman et al. (2004), open communication is
important to avoid misunderstandings with respect to the appli-
cation of any eventual rules and restrictions related to childrens
driving.
The two remaining factors, i.e., modeling and noncommit-
ment both are to be considered as forms of indirect social learning.
Modeling is about the example parents set through their own atti-
tude and driving style. The study by Cestac et al. (2014) showed that
merely asking your teen not to take a risk is certainly not enough.
Parents must set an example for their children. The importance of
modeling in driving behavior is undeniable, as research shows that
driving behaviors of children mirror the driving behaviors of their
parents (Bianchi and Summala, 2004; Ehsani et al., 2014; Lahatte
and Le Pape, 2008; Miller and Taubman-Ben-Ari, 2010; Taubman-
Ben-Ari et al., 2005), and that crash rates and violations of parents
can predict those of their children (Brookland et al., 2014; Ferguson
et al., 2001; Wilson et al., 2006). The factor noncommitment refers
to parents tendency to be actively involved in and committed (or
not) to road safety. Research indicates that more outspoken engage-
ment of parents in road safety increases desirable attitudes and
driving behavior (Ginsburg et al., 2009; Laird, 2014).
In order to measure these 7 factors, Taubman-Ben-Ari and Katz-
Ben-Ami (2013) developed a questionnaire (i.e., the Family Climate
for Road Safety Survey). They were able to identify them as the
conceptual backbone of family safety climate by means of factor
analysis in a sample of Israeli young novice drivers. Moreover, they
explored the predictive validity of their FCRS-model, and found that
2 out of the 7 underlying dimensions signicantly predicted Israeli
young novice drivers commitment to safe driving. Additionally, 4
out of 7 dimensions also predicted a set of self-reported driving
violations. As will become clear, the current study uses this mea-
surement scale in order to assess the effect of family climate on the
behavior of young novice drivers in Flanders. However, before pass-
ing to the more precise aims and hypotheses of this study, we rst
elaborate on the potential of an individuals personally held dis-
positions to moderate the effect of family safety climate on risky
driving.
1.3. Moderation of family safety climate effects
In the eld of trafc safety as in other (health related) domains,
it is a widely acknowledged idea that the formation of behav-
ior is a process where external factors (such as FCRS) interact
with more personally held dispositions (Green and Kreuter, 2005).
Yet, despite the popularity of this contention, there is not much
empirical work available on the more precise nature of this inter-
action. Walker and Frimer (2007) together with Hardy and Carlo
(2005) also concluded this when reviewing the domain of moral
 A. Carpentier et al. / Accident Analysis and Prevention 73 (2014) 5364
behavior. Walker and Frimer (2007, p. 857) therefore pleaded in
favor of a more systemic exploration of the interaction of personal
and situational variables. Aquino et al. (2009) responded to this
plea and were among the rst to examine the joint inuence of
personal and situational variables on moral intentions and behav-
ior. One of the key premises behind the proposed model was that
the inuence of situational variables on (moral) behavior would
be moderated by personal variables such as the centrality of moral
identity. They were able to conrm this moderation hypothesis by
means of an online survey and an in-lab priming experiment.
In this study, we adopt the same logic but focus on a different
set of variables. More in detail, the external factor of interest will be
family safety climate. The personal variables functioning as poten-
tial moderators of the effect of family safety climate on risky driving
will be drawn from the literature on social cognition models. For
quite some time, such models have been examined for their pre-
dictive validity in research on risky driving (Elliott et al., 2013). As
discussed by Armitage and Conner (2000), different social cognition
models exist that focus on a variety of personal variables. Three of
these personal variables are often cited in studies on young novice
drivers, i.e., attitude, locus of control (LOC) and social norm.
Attitudes are tendencies to evaluate an entity with some degree
of favor or disfavor, ordinarily expressed in cognitive, affective, and
behavioral responses (Eagly and Chaiken, 1993). In the context of
road safety, the concept attitude can be seen as the expression of
favor or disfavor towards a variety of (un)safe behaviors. It has been
shown repeatedly that attitude toward road safety is an important
predictor of driving behavior (De Pelsmacker and Janssens, 2007;
Elliott and Thomson, 2010; Iversen, 2004; Paris and Broucke, 2008;
Ulleberg and Rundmo, 2003). Therefore, the variable attitude is
included in our study.
Another frequently studied personal factor is locus of control.
Locus of control stands for the belief with regard to the source
and the cause for personal behavior. Locus of control consists of
2 dimensions, (Aiken et al., 1991) internal locus of control: the per-
ception that outcomes are determined more by internal attributes
(such as knowledge, skills, will-power), and (Ajzen, 1991) exter-
nal locus of control: the perception that outcomes are determined
by non-behavioral, external factors (such as climatologic circum-
stances, time availability, cooperation of others) (Ajzen, 2002;
Hennessy, 2011). Different studies support that locus of control
is a solid and reliable predictor of motor vehicle crashes and that
it enables to distinguish drivers that are involved in fatal motor
crashes more accurately from drivers that are not (Gidron et al.,
2003; Huang and Ford, 2012). We therefore also incorporate locus
of control into this study.
Social norm is another socio-cognitive factor to take into
account when predicting driving behavior. Social norm can be
seen as a persons beliefs or understandings of a group about how
members of this group should behave in certain situations. Social
norms can inuence a persons decisions either implicitly or explic-
itly. When the inuence is implicit, it derives from the observed
behavior of others. When the inuence is explicit, it derives from
statements or encouragement by others. In the context of the
present study, the group unit from which the social norms can
be derived will be the young drivers peer group. This is because
different studies conrmed the inuence of social norms elicited
from peers on young peoples driving behavior (Fleiter et al., 2010;
Jongen et al., 2013; Scott-Parker et al., 2009; Sinclair, 2013). Social
norm was therefore included into the study.
1.4. Aims and hypotheses
The main objective of this study is to examine the relative impor-
tance of young novice drivers family safety climate for road safety
on their driving behavior. In order to provide an answer, three
55
Fig. 1. Conceptual model
hypotheses will be formulated. As portrayed in the conceptual
model (Fig. 1), we expect the underlying factors of the FCRS con-
cept to have a direct inuence on the risky driving of young drivers
(hypothesis 1). More in detail, we expect that the more negative
(i.e., risk supportive) the family climate for road safety is, the more
likely it will be that young novices self-declare to drive in a risky
manner.
Regarding the three socio-cognitive determinants (i.e., attitude,
locus of control and social norm) we hypothesize that they exert
a direct inuence on young novices risky driving (hypothesis 2).
More in particular we make the following assumptions: (Aiken
et al., 1991) more negative (i.e., risk supportive) attitudes will lead
to a higher probability of risky driving, (Ajzen, 1991) the more
young novice drivers attribute the control over their behavior to
certain elements other than themselves, the higher the chance of
risky driving and (Ajzen, 2002) the more young novices experience
negative (i.e., risk supportive) feedback from their peers, the higher
the chance of risky driving.
Finally, we hypothesize that the three socio-cognitive factors
will moderate the effect of family safety climate on risky driving
(hypothesis 3). To the best of our knowledge, such a moderation
hypothesis has not yet been investigated before.
As a secondary objective, since we reuse the family climate for
road safety scale as it was developed by Taubman-Ben-Ari and Katz-
Ben-Ami (2013) in a sample of young novice drivers that differs in
terms of nationality (i.e., Flemish vs Israeli), this study also allows us
to explore the extent to which this scale validates cross-nationally.
2. Method
2.1. Participants
Young novice drivers were recruited via convenience sampling:
an online survey was developed and distributed through the Uni-
versity of Hasselt and online boards. One hundred and seventy-one
young novice drivers (104 women; 67 men) completed the ques-
tionnaire. Participants ranged in the age from 17 to 24 (M = 19.71,
SD = 1.49) and were in possession of their temporary drivers license
(M = 15 months, SD = 7) or their permanent drivers license for no
longer than one year (M = 8.5 months, SD = 3.4). Respondents par-
ticipated on a voluntary basis.
2.2. Procedure and instruments
Participants were asked to complete an online question-
naire consisting of 5 components measuring (Aiken et al., 1991)
56 A. Carpentier et al. / Accident Analysis and Prevention 73 (2014) 5364

Table 1
Variables and itemsaverages and standard deviations.

Variable and items M SD

Attitude ( = 0.81)
You can still drive safely when you have exceeded the alcohol limit with only one glass 3.49 1.37
If you are extra careful and attentive, you can sometimes exceed the speed limit 3.08 1.29
Quickly listening to a message on your cellphone isnt that harmful 3.79 1.17
Even a small amount of drugs while driving can be life-threatening [] 4.22 1.23
By driving to close to the car in front of you, you put both yourself and others at risk [] 4.14 1.00
If it is for short distances, it is not always necessary to wear your seatbelt 4.32 1.12
Locus of control ( = 0.75)
When I am rushed, it can happen that I drive more recklessly 2.91 1.15
If it is really urgent, it can happen that I use my cellphone while driving 3.37 1.37
If I would go to a party close to home, I wouldnt mind drinking some alcohol before driving home 3.71 1.32
It mainly depends on other drivers whether or not I get involved in a car accident 2.92 1.01
When all others drive fast, I will start driving faster myself. 3.13 1.23
When there is not much trafc, it can happen that I take more risks 2.89 1.19
Social norm ( = 0.81)
My friends use their cellphones regularly while driving. 2.88 1.16
My friends wouldnt mind if you occasionally dont wear a seatbelt 3.37 1.15
My friends drive faster than the speed limit on a regularly basis 2.57 1.14
My friends wouldnt mind if you drank a glass of alcohol before driving 2.92 1.24
My friends sometimes drive to close to the driver in front of them 2.94 1.13
My friends wouldnt mind if you used a small amount of drugs before driving 4.24 1.07
Social desirability ( = 0.62)
I have never exceeded the speed limit 3.93 1.16
I always am sure how to act in trafc situations 2.77 0.89
I always obey trafc rules, even if Im unlikely to be caught 2.54 1.06
I never regret my decisions in trafc 3.22 0.75
I always keep sufcient distance from the car in front of my car 2.56 0.91
I always remain calm and rational in trafc 3.05 1.09
Risky driving ( = 0.86)
I usually wear my seatbelt in the car (front and backseat), even for short distances [] 4.26 1.20
It sometimes happens that I overtake a driver making use of the wrong lane 3.78 1.38
It sometimes happens that I drive the car when I am not really sure if I drank too much or not 4.07 1.20
It sometimes happens that I text while Im driving 3.57 1.33
It sometimes happens that I cross an intersection when the lights just turned red 3.95 1.18
It sometimes happens that I overtake another vehicle were this is not allowed 3.79 1.28

[] = reversed item.

background data, (Ajzen, 1991) family climate for road safety,
(Ajzen, 2002) attitude, social norm and locus of control, (Aquino
et al., 2009) risky driving, and (Armitage and Conner, 2000) the
tendency to answer in a socially desirable way.
2.2.1. Risky driving
The 6 items in the risky driving scale were derived from differ-
ent road user behavior questionnaires (Elliott and Baughan, 2004;
Iversen, 2004; Martinussen et al., 2013). The items were assembled
in such a way that they explored several forms of dangerous driving
(e.g., It sometimes happens that I drive to close to the vehicle in
front of me or It sometimes happens that I text while Im driving).
The items were formulated negatively and they were measured on
a 5-point Likert scale from 1 (I agree very much) to 5 (I agree not
at all). Lower values indicated a risk supportive inclination, while
higher values implied a risk aversive inclination. Table 1 contains
the details.
2.2.2. Family climate for road safety
The original scale measuring family climate for road safety was
used. This scale contained items which explore the participants
idea about the 7 factors of FCRS (i.e., modeling, noncommitment,
feedback, messages, limits, communication and monitoring). Par-
ticipants expressed their level of agreement with these items on a
Likert scale from 1 (very much) to 5 (not at all). As will be discussed
more in detail under Section 3.1, factor analysis indicated that the
most optimal solution retained 36 items. These items, together
with the mean values, standard deviations, and factor loadings are
reported in Table 2.
2.2.3. Attitude
The scale measuring the attitude of young drivers toward risky
driving consisted of 6 items that were derived from existing sur-
veys such as the Driver Attitude Questionnaire (Mann, 2010) and
the attitude scale developed by Iversen (2004). The items covered
the following topics: drinking and driving, drugs, speeding, close
following, seat belt use and using a cellphone while driving (e.g.,
If you are very careful it is acceptable to drive a little faster). The
items were formulated negatively and they were measured on a
5-point Likert scale from 1 (I agree very much) to 5 (I agree not
at all). Lower values indicated a risk supportive inclination, while
higher values implied a risk aversive inclination. Items had a nega-
tive formulation and were measured by means of a 5-point Likert
scale from 1 (very much) to 5 (not at all). Mean values, standard
deviations and Cronbachs alpha can be found in Table 1.
2.2.4. Locus of control
The locus of control scale measured the level of control par-
ticipants believe to have over a selection of risk facilitating trafc
situations. We borrowed 6 items from already existing LOC-scales
(Levenson, 1973; Ozkan and Lajunen, 2005). More in detail, we used
a 5-point Likert scale from 1 (very much) to 5 (not at all), with lower
values indicating a higher (external) level of control whereas higher
values indicate lower (external) level of control (e.g., participants
that agree with the statement When all others drive fast, I will
drive faster myself, tend to have a higher external LOC). We refer
to Table 1 for more detailed information on these items.
2.2.5. Social norm
The social norm scale contained 6 items tapping the level
up to which respondents experience both implicit as well as
 A. Carpentier et al. / Accident Analysis and Prevention 73 (2014) 5364 57

Table 2
FCRSS for Flanders: factor model coefcients, averages and standard deviations.

Factors and items Loading M SD

Factor 1communication ( = 0.87)

In my family we talk openly about anything related to driving 0.84 2.04 1.01
My parents tell me when they think Im driving dangerously 0.79 1.75 0.83
In my family we talk openly about mistakes on the road or near accidents so I can learn from them 0.77 2.16 1.08
I can talk freely with my parents about different driving situations 0.74 1.74 0.83
I know how my parents expect me to drive 0.67 1.91 0.87
My parents expectations from me about driving safely are very clear to me 0.54 1.81 0.89
We talk at home about how to prevent or avoid dangerous situations on the road 0.42 1.80 0.85
My parents talk to me about potential hazards on the road 0.32 2.22 1.04
Factor 2modeling ( = 0.88)
My parents set an example by obeying trafc laws 0.78 2.53 1.10
My parents drive safely even when theyre in a hurry 0.74 2.45 1.08
My parents tell me to drive carefully even though theyre not very careful drivers [] 0.72 2.61 1.19
My parents talk about safe driving but they dont drive so safely themselves [] 0.67 2.33 1.19
My parents obey the trafc laws even when theyre tired or feeling stressed 0.65 2.70 1.08
My parents serve as role models for safe driving 0.61 2.70 1.02
Factor 3feedback ( = 0.88)
My parents compliment me for driving safely 1.00 2.67 1.01
I get positive feedback from my parents whenever they see me drive safely 0.96 2.53 1.04
I feel that my parents are proud of me when I drive safely 0.87 2.29 0.92
My parents praise me when I drive safely and carefully 0.84 2.64 1.09
My parents take an interest in how I drive 0.63 2.44 0.89
My parents encourage me and applaud me when they see I make sure to drive safely 0.55 2.15 0.88
Factor 4monitoring ( = 0.83)
I have to get my parents permission every time I want to go out in the car 0.81 2.91 1.40
Whenever I take the car, I have to tell my parents when Ill be home 0.78 2.72 1.24
Whenever I take the car, I have to tell my parents where Im going 0.73 2.53 1.35
Whenever I take the car, I have to tell my parents who Im taking with me wherever I go 0.66 2.93 1.29
Whenever I take the car, I have to call my parents and tell them if Im going to be late 0.66 2.84 1.32
Factor 5limits ( = 0.82)
My parents made it clear to me that if I didnt obey the trafc regulations they would restrict my 0.90 2.63 1.27
driving
My parents take every trafc violation very seriously, even when it doesnt result in a crash 0.79 2.49 1.13
If my parents found out I wasnt driving safely, they would impose limits on my driving 0.60 2.22 1.12
My parents let me take the car more often when they feel I drive safely 0.56 2.27 0.99
My parents make it clear that driving safely is more important than getting somewhere on time 0.53 2.19 1.06
Factor 6noncommitment ( = 0.82)
My parents tell me when they think Im driving dangerously [] 0.75 4.14 0.94
My parents really care that I drive safely [] 0.71 4.39 0.94
My parents believe that driving safely is very important [] 0.66 4.20 0.89
My parents wouldnt let me take the car if I drove recklessly, even if it would make it easier for 0.60 3.74 1.21
them if I drove (to go to the store, to pick someone up) []
My parents are willing to accept it if I get home late because I didnt want to speed. [] 0.40 3.86 1.10
My parents dont spend time teaching me how to drive safely 0.35 4.00 1.15

[] = reversed item.

explicit inuences exerted by their peers. Items were drawn from
already existing scales such as the peer scale developed by Parker,
Manstead and Stradlin (As cited in Mann, 2010). Both implicitly
expressed peer norms (e.g., My friends regularly use their cell-
phones while driving) and explicitly expressed peer norms (e.g.,
my friends do not mind if you do not wear your seatbelt all the
time) were assessed. Participants gave their level of agreement
on 5-point Likert scales ranging from 1 (agree very much) to 5
(agree not at all), with lower values suggesting high presence of
risk supportive peer norms and higher values standing for low
presence of risk supportive peer norms. Table 1 contains more
detailed information with regard to the items and their descriptive
statistics.
2.2.6. Social desirability
Social desirability is a form of response bias driven by respon-
dents tendency to give answers that make them look good
(Paulhus, 1991). Social desirability can undermine the validity
of self-reported measures (Harrison, 2010; Paulhus, 1991) which
explains why it is important to probe (and control) for it (if
necessary). We adopted items from the Driver Social Desirabil-
ity Scale (Lajunen et al., 1997). This scale assesses both forms of
social desirability, i.e., Impression Management (the deliberate
tendency to give favorable self-descriptions to others) and Self
Deception (positively biased but subjectively honest self-
description) (Lajunen et al., 1997; Paulhus, 1991). Also in this case,
we used a 5-point Likert scale ranging from 1 (agree very much) to
5 (agree not at all). Lower values indicated a higher tendency to
answer in a socially desirable manner and higher values suggested
a lower tendency to answer in a socially desirable way. We refer to
Table 1 for more details.
2.3. Statistical analysis
We analyzed data with SPSS 20 and SAS 9.3. For the analy-
ses a summary measure of the items was made for each of the
variables. First, we conducted an exploratory principal component
factor analysis on the items enclosed by the FCRSS-scale. In order
to decide on the rotation technique (i.e., orthogonal vs oblique),
we checked the extent to which extracted factors were correlated
and whether multicollinearity was an issue or not. Since factors
were all correlated in the range of 0.3 or higher and given there
was no question of multicollinearity (VIF < 3), we opted for an
oblique Promax rotation, as it is recommended in the statistical
literature (Meyers et al., 2013). More precise detail on how the
nal factor solution was further determined can be found under
Section 3.1.
58 A. Carpentier et al. / Accident Analysis and Prevention 73 (2014) 5364

Table 3
Pearson correlation coefcients between (a) the FCRSS factors (b) the personal determinants and (c) risky driving.

Communication Modeling Feedback Monitoring Limits Non- Attitude Locus of Social Risky
commitment toward risky control norms driving
driving

Communication 1
Modeling 0.55** 1
feedback 0.55** 0.49** 1
Monitoring 0.33** 0.36** 0.35** 1
Limits 0.64** 0.54** 0.43** 0.49** 1
Noncommitment 0.73** 0.62** 0.59** 0.46** 0.68** 1
Attitude 0.45** 0.44** 0.34** 0.40** 0.48** 0.58** 1
toward risky
driving
Locus of 0.35** 0.39** 0.30** 0.32** 0.36** 0.40** 0.69** 1
control
Social norms 0.33** 0.32** 0.28** 0.18* 0.24** 0.28** 0.41** 0.37** 1
Risky driving 0.49** 0.42** 0.35** 0.34** 0.50** 0.56** 0.75** 0.72** 0.43** 1
**
Correlation is signicant at the 0.01 level.
*
Correlation is signicant at the 0.05 level.

Subsequently, we carried out a Spearmans correlation analysis
on risky driving and the various external and personal determi-
nants (i.e., the different family safety climate factors, attitude, locus
of control, and social norm) in order to explore the extent to which
these are associated with each other. Finally, a stepwise hierarchical
regression analysis was carried out to determine what the unique
contribution was in the prediction of risky driving. Based on the
strength of the associations that emerged from the correlation anal-
ysis, 4 types of explanatory variables were entered in the following
order: (Aiken et al., 1991) the background variables age, gender, and
driver license possession (entered in STEP 2), (Ajzen, 1991) the per-
sonal determinants attitude, LOC, and social norm (entered in STEP
3), (Ajzen, 2002) the FCRSS factors (entered in STEP 4), and (Aquino
et al., 2009) the interactions between signicant FCRSS factors and
signicant personal determinants (entered in STEP 5). Addition of
interactions between signicant FCRS factors and socio-cognitive
determinants was done in order to nd out whether these inter-
actions would signicantly increase the overall predictive power
of the model. The signicant interactions were further analyzed by
means of simple slopes analysis, following the procedure as it was
described in Aiken et al. (1991) and Hayes (2013). All was carried
out while controlling for social desirability (entered in STEP 1).
3. Results
3.1. Factor analysisFamily climate road safety survey
The results of the factor analysis revealed 12 factors (eigen-
value >1). Based on the scree plot, a 6-factor solution seemed most
preferable, explaining 54.1% of the variance. Altogether, this 6-
factor solution retained 36 items of the original scale. Interestingly,
the majority of these items loaded on the same factors as in the
study by Taubman-Ben-Ari and Katz-Ben-Ami (2013), except for
items related to the factor messages. In our analysis, this factor
did not emerge. Instead, its items spread over two other factors,
i.e., communication and noncommitment.
As presented in Table 2, Factor 1 consisted of 7 items loading
high (>0.32) on the factor and it explained 31.69% of the variance.
These items related to direct and indirect parent-child communi-
cation and open discussions with regard to safe driving. Therefore,
this factor was labeled communication. Factor 2 explained 5.86%
of the variance and consisted of 6 items. These items all related
to the example parents set to their children concerning driving
safely, obeying the law, etc. The factor therefore was labeled mod-
eling. Factor 3 explained 4.55% of the variance and consisted of
6 items which gave an indication of the feedback parents provide
and the interest they show in their offsprings safe driving. Factor
3 thus received the label feedback. Factor 4 explained 4.55% of
the variance and consisted of 5 items relating to parents supervi-
sion and control over their childrens car use and driving activities.
Accordingly, factor 4 was labeled monitoring. Factor 5 explained
3.90% of the variance and consisted of 5 items and could clearly
be labeled as limits. Factor 6 explained 2.80% of the variance and
consisted of 6 items which related to parents involvement, sup-
port or unsupportive and irresponsible behavior and therefore was
labeled noncommitment.
These 6 factors were retained for further analysis.
3.2. Correlation analysis
A Spearmans correlation was run to explore the relationship
between the variables. Signicant associations were found between
all variables, though some correlations were stronger than others
(see Table 3). The family climate factors correlated highest with
attitude and behavior. High associations were found (Aiken et al.,
1991) between attitude on the one hand and communication, mod-
eling, monitoring, limits and noncommitment on the other hand;
and (Ajzen, 1991) between behavior on the one hand and commu-
nication, modeling, monitoring, limits and noncommitment on the
other hand. Attitude and behavior both had the highest correlations
with noncommitment and the lowest with feedback. Besides com-
munication, limits, modeling and noncommitment, behavior itself
was most strongly associated with attitude and locus of control.
Thus, the stronger the self-reported negative driving, the higher
the levels of reported negative (i.e., risk supportive) attitude, the
lower the extent to which the young drivers believe they are in
control over their own behavior, the less positive (i.e., risk aver-
sive) communication and limits are, the poorer models parents are
perceived to be, and the higher the reported level of noncommit-
ment of parents.
3.3. Hierarchical regression analysis
As indicated previously, in the nal stage of the analysis we
carried out a 5-step hierarchical multiple regression (Table 4). The
regression analysis revealed that, in step 1, social desirability con-
tributed signicantly to the regression model (F(1, 169) = 88.78,
p < 0.001) accounting for 34.4% of the variance. Introducing the
background variables gender, age and drivers license explained
an additional 2.7%, though the change in R2 was not statisti-
cally signicant (F(3, 166) = 2.41, p = 0.07). In this second step, the
signicant effect for gender did show that young novice male
 A. Carpentier et al. / Accident Analysis and Prevention 73 (2014) 5364 59

Table 4 reveal that the interaction terms attitude noncommitment and

hierarchical regression analysis for the prediction of risky driving.
locus of control noncommitment had a statistically signicant
t R square R2 effect on the risky driving of young novice drivers. Together, the
Step 1 0.34 0.34*** independent variables accounted for 77.2% of the explained vari-
Social desirability 0.59 9.42*** ance in risky driving.
Step 2 0.37 0.03 The variables noncommitment attitude, noncommitment,
Social desirability 0.56 9.00*** attitude and LOC obtained the highest beta weights ( = 0.32,
Gender 0.13 2.01*
p < 0.001; = 0.27, p < 0.001; = 0.26, p < 0.001; = 0.24, p < 0.001),
Age 0.10 1.43
Drivers license 0.06 0.95 which demonstrates that these variables made the largest contribu-
Step 3 0.69 0.36*** tion to the regression equation with all other independent variables
Social desirability 0.18 3.20** held constant.
Attitudes toward 0.54 8.49***
Since the variance explained in risky driving signicantly
risky driving
LOC 0.16 2.46** increases by addition of the two interaction terms noncommit-
Social norms 0.09 1.90* ment attitude and noncommitment LOC, moderation of the
Step 4 0.73 0.05*** effect of noncommitment on risky driving by attitude and LOC
Social desirability 0.16 3.05** becomes a plausible idea. Therefore, these interactions were further
Attitudes toward 0.39 5.711***
analyzed by means of simple slopes analysis.
risky driving
LOC 0.18 2.882**
Social norms 0.10 2.126* 3.4. Simple slopes analysis
Modeling 0.05 0.80
Communication 0.01 0.08
The interaction terms which were found to be signicant in
Feedback 0.03 0.51
Monitoring 0.04 0.85
the regression analysis were further analyzed by means of sim-
Limits 0.12 1.95 ple slopes analysis for determining the nature of the interaction
Noncommitment 0.23 3.40** effects. The signicance of the regression lines were examined at
Step 5 0.77 0.04*** two levels of the hypothesized moderator: LOW (one standard devi-
Social desirability 0.15 3.08**
ation below the mean) and HIGH (one standard deviation above the
Attitudes toward 0.26 3.68***
risky driving mean).
LOC 0.24 4.06***
Social norms 0.10 2.28* 3.4.1. Attitude
Noncommitment 0.27 4.05***
Attitude was examined as a moderator of the relation between
Noncommit- 0.32 4.69***
ment attitude
noncommitment and risky driving. As illustrated in Fig. 2, more
Noncommit- 0.18 2.96** noncommitment was found to increase risky driving more, but
ment LOC only when attitude was low (b = 0.75, p < 0.0001). Noncommitment
Noncommit- 0.15 3.08 did not increase the level of risky driving when attitude was high
ment social
(b = 0.05, p = 0.59). Thus, when the young drivers had a nega-
norm
tive (i.e., risk supportive) attitude, the reported noncommitment
N = 171.
*
of their parents contributed more to their risky driving.
p < 0.05.
**
p < 0.01.
***
p < 0.001. 3.4.2. Locus of control
Locus of control was also examined as a moderator of the
relation between noncommitment and risky driving. Noncommit-
drivers agreed to commit risky driving more than the young ment was found to be a stronger predictor of risky driving under
female drivers. Introducing the three personal determinants in high internal LOC (b = 0.55, p < 0.0001) rather than under external
step 3 explained an additional 31.5% of the variance, with a sig- LOC (b = 0.15, p = 0.09). This indicates that more noncommitment
nicant change in R2 (F(3, 163) = 54.67, p < 0.001). A signicant increases risky driving as young novice drivers more strongly
contribution was found for each of the personal determinants: atti- believe that they are personally in control over the outcomes of
tude (b = 0.61, = 0.54, t = 8.49, p < 0.001), locus of control (b = 0.19, their behavior (Fig. 3).
= 0.16, t = 2.46, p < 01), and social norm (b = 0.1, = 0.09, t = 1.90, Interestingly, this result is not in line with what one would
p < 0.05). Thus, young drivers that reported (1) more negative expect. The fact that an external factor (i.e., noncommitment) has a
(i.e., risk supportive) attitudes, (2) lower personal control over stronger impact on ones personal behavior (risky driving) in case
events, and (3) more negative (i.e., risk supportive) inuence ema- one situates LOC more within the self, appears to be a counterin-
nating from peers, were more engaged in risky driving. Adding tuitive result. One would be inclined to believe that the impact of
the six family climate factors in step 4 explained an additional external factors would be weaker in such a case, because exter-
4.7% of the variation in risky driving, with a signicant change nal factors are considered as less controlling when LOC is situated
in R2 (F(6, 157) = 4.64, p < 0.001). Introducing these six factors in within the self. In our view, this leaves us with two options. We
the regression model only revealed a signicant contribution for could be confronted with a ceiling effect (i.e., the impact of exter-
noncommitment (b = 0.31, = 0.32, t = 3.40, p < 0.001). Thus, higher nal factors (like family safety climate) on risky driving is weaker
noncommitment to safe driving behavior contributed signicantly for individuals that situate LOC outside the self simply because for
to the young novice drivers higher reported risky driving. The fth those individuals, the available margin for external factors to have
and nal step, in which the interaction terms were introduced, an increased impact on behavior is smaller than the increase mar-
explained an additional 3.8% of the variation in risky driving, gin available for individuals that situate LOC inside the self), or we
with a signicant change in R2 (F(3, 154) = 8.60, p < 0.001). Two are facing a more complex (i.e., 3-way) interaction where attitude is
interactions had a signicant coefcient weight in the regression involved as an additional secondary moderating factor. This type of
equation for risky behavior: noncommitment attitude (b = 0.29, interaction has been described by Hayes (2013, p. 307308) as mod-
= 0.32, t = 4.69, p < 0.001) and noncommitment locus of con- erated moderation. For our study, moderated moderation implies
trol (b = 0.16, = 0.18, t = 2.96, p < 0.01). In other words, the results that attitude (i.e., the secondary moderator) would moderate the
60 A. Carpentier et al. / Accident Analysis and Prevention 73 (2014) 5364

Fig. 2. Simple slopes analysisnoncommitment attitude.

moderating effect of LOC (i.e., the primary moderator). Even though
there is a formal procedure to test for moderated moderation (e.g.,
Hayes, 2013), it must be said that the empirical evidence from
which such a hypothesis (with attitude as secondary moderator
and LOC as a primary moderator) could be derived is indirect and
scarce at best.
The only foundation we were able to nd to support a mod-
erating effect of attitude on effects exerted by LOC, comes from a
meta-analysis on the Theory of Planned Behavior (TPB) by Armitage
and Conner (2001, p. 742). In that study, the authors refer to Ajzens
(1991, p. 188) thoughts on the relative importance of the TPB-
variables (i.e., attitude, subjective norm, and perceived behavioral
control) which bring them to the contention that effects exerted by
the variable perceived behavioral control (we notice this variable
is conceptually very close to the variable LOC) can be moderated
depending on the strength of attitude and/or subjective norm. They
further argue there is only indirect evidence for this claim coming
from studies where it has been demonstrated that the relative pre-
dictive power of attitude and social norms varies in function of
differences in attitude (e.g., Notani, 1998; Sparks et al., 1992) and
individual sociability, respectively (e.g., Tramow & Finlay, 1996).
Despite the scarcity of supportive evidence, we decided to at least
explore the moderated moderation hypothesis according to the
procedure proposed by Hayes (2013).
3.4.3. Three-way interaction attitude LOC noncommitment
As a rst step, we introduced the 3-way interaction term
noncommitment attitude LOC into the regression model as an
additional predictor, resulting in a signicant increase in the
amount of variance explained in risky driving (R2 = 0.01, F(1,
153) = 7.93, p = 0.001) which indicates the presence of a three-way
interaction to be a plausible option.
Next, we conducted the simple slopes analysis as portrayed in
Fig. 4. Closer inspection of the plots seems to indicate that attitude
indeed moderates the (moderating) effect of LOC, but only when
attitude is negative (i.e., risk supportive). Paraphrased, a noncom-
mitted family climate will increase risky driving, and will do so
more in youngsters who situate LOC within themselves (b = 1.24,
p < 0.0001) than in youngsters who place LOC outside the self
(b = 0.57, p < 0.0001), but only if those youngsters attitude is risk
supportive. Different from that, a noncommitted family climate will
not increase risky driving more, neither for youngsters who locate

Fig. 3. Simple slopes analysisnoncommitment LOC.

 A. Carpentier et al. / Accident Analysis and Prevention 73 (2014) 5364
Fig. 4. Simple slopes analysisnoncommitment LOC attitude.
LOC outside the self (b = 0.09, p = 0.23), nor for youngsters who situ-
ate LOC within the self (b = 0.01, p = 0.94), in case those youngsters
attitude is positive (i.e., risk aversive).
4. Discussion
The current study had two objectives. The main objective was
to examine the relative importance of young novice drivers fam-
ily safety climate for road safety on their driving behavior. As
a secondary objective the validity of the scale was explored. In
addressing the rst objective, we formulated three hypotheses. As
a rst hypothesis, we expected the underlying factors of the FCRS
concept to have a direct inuence on the risky driving of young
novices. The results of the stepwise multiple hierarchical regres-
sion analysis indicate that the direct inuence of FCRS (dened as a
multidimensional concept) on risky driving is limited to one single
factor only, i.e., noncommitment. Put differently, it was found that
young novice drivers whose parents are to a lesser extent commit-
ted to road safety appear to engage more in risky driving. We should
be cautious in evaluating whether (or not) this result supports the
rst hypothesis. Given only one out of seven factors underlying the
FCRS concept signicantly predicted risky driving, we conclude that
the hypothesis is only partially supported at best. The fact that non-
commitment more specically comes out as a signicant predictor,
is not totally surprising and in line with the studies of Taubman-
Ben-Ari and Katz-Ben-Ami (2013) and Desrichard et al. (2007) in
which a comparable concept (i.e., parental involvement) predicted
the intention to violate driving rules.
The second hypothesis was formulated in light of the widely
accepted idea that external factors (such as family safety climate)
seldom inuence the formation of behavior in complete isola-
tion from more personally oriented dispositions. The literature
on social cognition models largely supports the argument that
socio-cognitive variables (such as attitude, social norms, LOC) con-
tribute to the prediction of a wide range of behaviors. We therefore
expected the variables attitude, social norms and LOC to have a
direct inuence on young novices risky driving. This hypothe-
sis was conrmed by the data we collected. Attitude as well as
social norms and LOC all contributed signicantly to the predic-
tion of risky driving. In more specic terms, we established that
(Aiken et al., 1991) more negative (i.e., risk supportive) attitudes,
(Ajzen, 1991) less perceived internal control over the outcome of
ones behavior, and (Ajzen, 2002) more perceived negative (i.e., risk
supportive) inuence emanating from peer groups, increase risky
driving. These results are in line with several studies where the
same socio-cognitive concepts have been tested for their predictive
61
validity toward risky driving (Gidron et al., 2003; Huang and Ford,
2012; Iversen, 2004; Jongen et al., 2013; Scott-Parker et al., 2009;
Ulleberg and Rundmo, 2003). The fact that social norms comes out
as the least important socio-cognitive predictor, closely resembles
what was found by Sheppard et al. (1988), Van den Putte (1991),
and Godin and Kok (1996). Interestingly, the three socio-cognitive
variables contribute substantially more to the prediction of risky
behavior than the external FCRS factors. This somehow indicates it
is indeed best not to consider external factors in isolation from more
personally oriented dispositions when studying the formation of
behavior.
We formulated the third hypothesis to carry out a more in-
depth exploration of the interaction between external factors and
internally held dispositions. More in particular, we expected atti-
tude, LOC and social norms to operate as moderators of the direct
effect of the FCRS-factors on risky driving. The interaction terms
between noncommitment (i.e., the only signicantly predicting
factor underlying FCRS) and each of the three socio-cognitive vari-
ables were added to the regression. This revealed that interactions
between noncommitment on the one hand and attitude and LOC
on the other hand, signicantly increased the models overall pre-
dictive power. This at least already suggests that the assumption of
interaction between FCRS-related factors and socio-cognitive vari-
ables contributes to a more complete prediction of risky driving.
In order to get a clearer view on the exact nature of the interac-
tion between noncommitment and attitude and LOC, we performed
a simple slopes analysis. It came out that both attitude and LOC
function as primary moderators of the effect of noncommitment
on risky driving. An additional (and rather unanticipated) nd-
ing was that attitude also operated as a secondary moderator (i.e.,
as a moderator of the moderation effect exerted by LOC). As dis-
cussed previously, the only evidence available to which we can
link the particular nding that attitude can potentially moderate an
effect exerted by LOC, comes from literature on the relative impor-
tance of TPB variables (e.g., Ajzen, 1991; Armitage and Conner,
2001). There it has been empirically documented that attitude (and
more in particular, attitude strength) can moderate effects exerted
by perceived behavioral control, a variable that is conceptually
very close to LOC (e.g., Notani, 1998; Sparks et al., 1992). In more
basic terms, the analyses to test the third hypothesis show us two
things: (Aiken et al., 1991) less commitment coming from young
novices family environment will increase their self-reported level
of risky driving more, but only when their attitude is risk support-
ive; (Ajzen, 1991) less commitment coming from young novices
family environment will increase the self-reported level of risky
driving more in youngsters who situate LOC within themselves
62 A. Carpentier et al. / Accident Analysis and Prevention 73 (2014) 5364
than in youngsters who place LOC outside the self, but only if those
youngsters attitude is risk supportive.
Before discussing the practical implications of these ndings,
we turn to the second objective of this study, i.e., to explore cross-
validity of the 7-factor model for FCRS as it was proposed by
Taubman-Ben-Ari and Katz-Ben-Ami (2013). Results for the factor
analysis show that the original 7-factor model obtained by means
of an Israeli sample is overall quite well preserved in a Flemish sam-
ple. Six out of the seven factors could be replicated similarly to the
study by Taubman-Ben-Ari and Katz-Ben-Ami (2013). The factor
labeled messages was absorbed by the factors communication
and noncommitment in the Flemish sample. Maybe, this is due to
the fact that, for Flemish respondents, items associated originally
with the factor messages are interpreted as being more closely
related to communication and/or noncommitment. This might
especially count for the factor communication which is indeed
conceptually close to the factor messages. Even though the over-
all factor structure converges quite well between the two samples,
a more detailed examination shows there are also some noticeable
differences. First, of the 54 items contained by the original scale,
only 36 were retained in the Flemish sample, which is a substantial
reduction. Second, while for the Israeli sample, the different fac-
tors explained comparable percentages of variance (scores were in
the range between 9.15% and 5.37%), for the Flemish sample, the
rst factor explained a substantially higher percentage of variance
(i.e., 31.69%) than the other ve factors for which scores ranged
from 5.86% to 2.80%. Third, also in terms of order of importance,
there are differences between the Israeli and the Flemish sample.
For instance, while in the former the factor modeling explains
most of the variance, followed by feedback and communication, in
the latter communication comes rst, followed by modeling and
feedback, respectively. In sum, the FCRS scale as it was developed
by Taubman-Ben-Ari and Katz-Ben-Ami (2013) appears to have
potential to cross-validate in other than Israeli samples, although
there are a few issues to be kept in mind. Even though the originally
proposed factors are all together well replicated, they are organized
and structured somewhat differently in a Flemish sample. Also,
when looking at individual items and the extent to which these
can be retained, it seems there is a quite pronounced difference
between an Israeli and a Flemish sample.
5. Implications
Practitioners and policy makers can learn three things from this
study. First of all, it gets reconrmed that familial commitment
toward road safety is an important determinant of young novices
tendency to engage in (or decline from) risky driving. What particu-
larly warrants our attention is that the mere absence of such a safety
supportive commitment can already contribute to increased levels
of risky driving. Put differently, for young novices to be stimulated
to behave dangerously while driving, it is not a prerequisite that
family members themselves are actively engaged in unsafe driv-
ing. The simple fact that family relatives do not sufciently approve
safe driving (and likewise, disapprove unsafe driving) could already
be a risk facilitating condition for young novices. This implies that
family members should not just try to avoid being a negative role
model, but they should also actively reinforce safe driving within
the family.
A second nding with important practical implications is conr-
mation of the idea that external factors (such as FCRS) inuence the
formation of individual behavior together with person oriented dis-
positions (such as attitude, social norms, and LOC). In other words,
even though factors situated outside an individual (i.e., in the social
and/or physical environment) can be identied as relevant inter-
vention targets for practitioners and policy makers, those factors
should not be addressed in isolation from socio-cognitive factors
situated within the individual, and vice versa. It is important for
intervention planners to keep this in mind in order to avoid the risk
of being too restrictive in the identication of target variables and
the setting of program objectives (Green and Kreuter, 2005). Rather
than a fragmentary approach, a holistic or systemic view with suf-
cient attention for the synergy between the individual and his/her
environment is therefore to be encouraged (e.g., Bartholomew et al.,
2011). One way to better exploit the synergy between parents and
their children might be to have them participate jointly to safety
promoting interventions, rather than addressing them by means of
separate initiatives. Practice shows that there is indeed a current
trend to jointly include parents and their youngsters in programs
that can be situated in the pre-license stage of learning how to
drive and/or the post-license stage of solo driving (Chaudhary et al.,
2004; Goodwin et al., 2006; Simons-Morton et al., 2006; Simons-
Morton et al., 2004; Toledo et al., 2012). In light of the results we
obtained in this study, such initiatives should be aimed congru-
ently at the activation of commitment toward road safety for the
parents, and the promotion of (Aiken et al., 1991) safety supportive
attitudes, (Ajzen, 1991) skills to resist risk supportive peer pressure,
and (Ajzen, 2002) the ability to personally cope with risk facilitating
situational conditions for the youngsters.
A third nding of practical importance is the unprecedented
empirical observation that the FCRS-related factor noncommit-
ment not only affects risky driving independently of person
oriented dispositions but that there is also interaction between
the former and the latter. More in detail, we found that personally
held social cognitions such as attitude and LOC can moderate the
risk increasing tendencies generated by noncommitment. A simple
slopes analysis revealed more in-depth information on the exact
nature of the moderation. Of special interest is that attitude only
moderates the risk increasing effect generated by noncommitment
when it is negative (i.e., risk supportive). Indeed, irrespective of
whether LOC is situated within the self or outside, positive (i.e.,
risk aversive) attitudes do not mitigate the risk increasing ten-
dencies generated by noncommitment. Contrary to that, negative
(i.e., risk supportive) attitudes further strengthen the risk increas-
ing tendencies caused by noncommitment. Additionally, operating
as a secondary moderator, risk supportive attitudes give an extra
boost to the risk increasing effect generated by noncommitment
for young novices that situate LOC within the self. In sum, when
interacting with noncommitment, attitude always operates as a
negative (i.e., risk increasing) reinforcer, never as a positive (i.e.,
risk mitigating) one. In practical terms, this means that the counter-
measure of promoting risk aversive attitudes among young novices
will not weaken the risk increasing effect exerted by noncommit-
ment, but that additional reinforcement of this risk increasing effect
might be prevented.
6. Limitations and recommendations for future research
The current study is subject to some limitations that need to
be discussed. First, this study is based on a convenience sample
of university students. We should be cautious with student sam-
ples, especially in terms of external validity. Peterson for instance
(Peterson, 2001), concluded that effect sizes derived from college
student subjects frequently differ from those derived from nonstu-
dent subjects both directionally and in magnitude. In addition to
that, trafc safety research in general shows that (late) adolescent
research samples (as the one under study here), more than other
age groups, experience problems with risk perception. Indeed,
(late) adolescents have been frequently identied as high risk-
takers in their daily life-style, as more sensitive to peer inuences
in adopting inappropriate norms, and as more inclined to drive at
 A. Carpentier et al. / Accident Analysis and Prevention 73 (2014) 5364
high-risk conditions (Lee, 2007). Therefore, the results of this study
should be interpreted with care. Nonetheless, the merits of student
samples should not be overlooked either (Enis et al., 1972). To illus-
trate, meta-analyses stemming from various research areas within
social sciences have shown that effect sizes do not always differ sig-
nicantly between student and non-student samples (Verlegh and
Steenkamp, 1999). Also, in our specic case, there is something
to say in favor of selecting students as model type respondents,
because the target group is young novice drivers. Another limita-
tion of the study is that the view of parents on family climate for
road safety was not measured, whereby the answers of the young
drivers with regard to the family climate factors cannot be validated
against the vision of parents. It could be an added value to take this
up in future studies. Another issue that needs to be raised is that
the study was based on a cross-sectional design. There are different
shortcomings when using this type of design. One of the concerns is
also referred to as Neyman bias, which basically indicates that past
behavior was measured and a consequence of this type of design is
that, regardless of how objective your survey may be, it cannot be
expected that participants are able to answer question about past
events with complete accurateness. To end with, the study did not
take the different family structures and specic circumstances into
account, and the impact this could have on the way parents edu-
cate and communicate with their children. For instance, it is known
that mothers and fathers have different relationships and ways of
interaction with respectively their sons and daughters (Maximo
et al., 2011). In this perspective, investigating the different family
structures in future research could shed some light on the impact
this has on the way family climate for road safety impacts the risky
driving of young drivers.
7. Conclusion
We come to the following four conclusions. First of all, non-
commitment is the only FCRS-related factor that signicantly
contributes to the prediction of risky driving. Second, attitude,
social norms and locus of control contribute to the prediction
of risky behavior considerably more than the FCRS factors. This
indicates the importance of considering external- and internal fac-
tors together when studying risky driving. Third, the interaction
between the external factor noncommitment and the internal fac-
tors attitude and locus of control signicantly contributes to the
prediction of risky driving. A more in-depth exploration of these
interactions shows that (Aiken et al., 1991) less commitment com-
ing from young novices family environment will increase their
self-reported level of risky driving more, but only when their atti-
tude is risk supportive; (Ajzen, 1991) less commitment coming
from young novices family environment will increase the self-
reported level of risky driving more in youngsters who situate LOC
within themselves than in youngsters who place LOC outside the
self, but only if those youngsters attitude is risk supportive. Finally,
even though the originally proposed factors underlying the FCRS
concept seem to replicate quite well in samples of different nation-
alities, there are some noticeable differences as to how these factors
are structured and organized.
References
Aiken, L.S., Reno, R.R., West, S.G., 1991. Multiple Regression: Testing and Interpreting
Interactions. SAGE Publications, Newbury Park, CA.
Ajzen, I., 1991. The theory of planned behavior. Organ. Behav. Hum. Decis. Processes
50, 179221.
Ajzen, I., 2002. Perceived behavioral control, self-efcacy, locus of control, and the
theory of planned behavior. J. Appl. Soc. Psychol. 32 (4), 665683.
Aquino, K., Freeman, D., Reed, A., Felps, W., Lim, V.K.G., 2009. Testing a social-
cognitive model of moral behavior: the interactive inuence of situations and
moral identity centrality. J. Pers. Soc. Psychol. 97 (1), 123141.
63
Armitage, C.J., Conner, M., 2000. Social cognition models and health behaviour: a
structured review. Psychol. Health 15 (2), 173189.
Armitage, C.J., Conner, M., 2001. Efcacy of the theory of planned behaviour: a meta-
analytic review. Br. J. Soc. Psychol. 40 (4), 471499.
Bandura, A., 1986. Social Foundations of Thought and Action: A Social Cognitive
Theory. Prentice-Hall, Englewood Cliffs, NJ.
Bartholomew, L.K., Parcel, G.S., Kok, G., Gottlieb, N.H., Fernandez, M.E., 2011.
Planning Health Promotion Programs: An Intervention Mapping Approach.
Jossey-Bass, San Francisco, CA.
Beck, K.H., Shattuck, T., Raleigh, R., 2001. Parental predictors of teen driving risk.
Am. J. Health Behav. 25 (1), 1020.
Bianchi, A., Summala, H., 2004. The genetics of driving behavior: parents driv-
ing style predicts their childrens driving style. Accid. Anal. Prev. 36 (4),
655659.
Brookland, R., Begg, D., Langley, J., Ameratunga, S., 2014. Parental inuence on ado-
lescent compliance with graduated driver licensing conditions and crashes as
a restricted licensed driver: New Zealand drivers study. Accid. Anal. Prev. 69,
3039.
Cestac, J., Paran, F., Delhomme, P., 2014. Drive as I say, not as I drive: inuence of
injunctive and descriptive norms on speeding intentions among young drivers.
Transp. Res: F: Trafc Psychol. Behav. 23, 4456.
Chaudhary, N.K., Ferguson, S.A., Herbel, S.B., 2004. Tennessees novice driver safety
project: a program to increase parental involvement. Trafc Inj. Prev. 5 (4),
356361.
De Pelsmacker, P., Janssens, W., 2007. The effect of norms, attitudes and habits
on speeding behavior: scale development and model building and estimation.
Accid. Anal. Prev. 39 (1), 615.
Desrichard, O., Roch, S., Bgue, L., 2007. The theory of planned behavior as mediator
of the effect of parental supervision: a study of intentions to violate driving rules
in a representative sample of adolescents. J. Saf. Res. 38 (4), 447452.
Eagly, A.H., Chaiken, S., 1993. The Psychology of Attitudes. Harcourt Brace Jovanovich
College Publishers, Orlando, FL.
Ehsani, J.P., Simons-Morton, B., Xie, Y., Klauer, S.G., Albert, P.S., 2014. The asso-
ciation between kinematic risky driving among parents and their teenage
children: moderation by shared personality characteristics. Accid. Anal. Prev. 69,
5661.
Elliott, M.A., Baughan, C.J., 2004. Developing a self-report method for investigat-
ing adolescent road user behaviour. Transp Res: F: Trafc Psychol. Behav. 7 (6),
373393.
Elliott, M.A., Thomson, J.A., 2010. The social cognitive determinants of offending
drivers speeding behaviour. Accid. Anal. Prev. 42 (6), 15951605.
Elliott, M.A., Thomson, J.A., Robertson, K., Stephenson, C., Wicks, J., 2013. Evidence
that changes in social cognitions predict changes in self-reported driver behav-
ior: causal analyses of two-wave panel data. Accid. Anal. Prev. 50, 905916.
Enis, B.M., Cox, K.K., Stafford, J.E., 1972. Students as subjects in consumer behavior
experiments. J. Mark. Res. 9 (1), 72.
Ferguson, S.A., Williams, A.F., Chapline, J.F., Reinfurt, D.W., De Leonardis, D.M., 2001.
Relationship of parent driving records to the driving records of their children.
Accid. Anal. Prev. 33 (2), 229234.
Fleiter, J.J., Lennon, A., Watson, B., 2010. How do other people inuence your driv-
ing speed? Exploring the who and the how of social inuences on speeding
from a qualitative perspective. Transp. Res: F: Trafc Psychol. Behav. 13 (1),
4962.
Gidron, Y., Gal, R., Syna Desevilya, H., 2003. Internal locus of control moderates
the effects of road-hostility on recalled driving behavior. Transp. Res: F: Trafc
Psychol. Behav. 6 (2), 109116.
Ginsburg, K.R., Durbin, D.R., Garca-Espana, J.F., Kalicka, E.A., Winston, F.K., 2009.
Associations between parenting styles and teen driving, safety-related behav-
iors and attitudes. Pediatrics 124 (4), 10401051.
Godin, G., Kok, G., 1996. The theory of planned behavior: a review of its applications
to health-related behaviors. Am. J. Health Promot.: AJHP 11 (2), 8798.
Goodwin, A.H., Foss, R.D., Margolis, L.H., Harrell, S., 2014. Parent comments and
instruction during the rst four months of supervised driving: an opportunity
missed? Accid. Anal. Prev. 69, 1522.
Goodwin, A.H., Waller, M.W., Foss, R.D., Margolis, L.H., 2006. Parental supervi-
sion of teenage drivers in a graduated licensing system. Trafc Inj. Prev. 7 (3),
224231.
Green, L.W., Kreuter, M.W., 2005. Health Program Planning: An Educational and
Ecological Approach. McGraw-Hill, New York, NY.
Harakeh, Z., Scholte, R.H.J., de Vries, H., Engels, R.C.M.E., 2005. Parental rules and
communication: their association with adolescent smoking. Addiction (Abing-
don, England) 100 (6), 862870.
Hardy, S.A., Carlo, G., 2005. Identity as a source of moral motivation. Hum. Dev. 48
(4), 232256.
Harrison, W.A., 2010. Is it nally time to kill self-report outcome measures in road
safety? An investigation of common method variance in three surveys of a cohort
of young drivers that included the Driver Behaviour Questionnaire. In: Presented
at the Australasian Road Safety Research Policing Education Conference, 2010.
Hartos, J., Eitel, P., Simons-Morton, B., 2002. Parenting practices and adolescent risky
driving: a three-month prospective study. Health Educ. Behav. 29 (2), 194206
(The Ofcial Publication of the Society for Public Health Education).
Hartos, J.L., Eitel, P., Haynie, D.L., Simons-Morton, B.G., 2000. Can I take the car?
Relations among parenting practices and adolescent problem-driving practices.
J. Adolesc. Res. 15 (3), 352367.
Hayes, A.F., 2013. Introduction to Mediation, Moderation, and Conditional Process
Analysis: A Regression-Based Approach. Guilford, New York, London.
64 A. Carpentier et al. / Accident Analysis and Prevention 73 (2014) 5364
Hennessy, D., 2011. Social, personality, and affective constructs in driving. In:
Porter, B.E. (Ed.), Handbook of Trafc Psychology. Academic Press, San Diego, pp.
149163, Retrieved from http://www.sciencedirect.com/science/article/pii/
B9780123819840100128. (Chapter 12).
Huang, J.L., Ford, J.K., 2012. Driving locus of control and driving behaviors: inducing
change through driver training. Transp. Res: F: Trafc Psychol. Behav. 15 (3),
358368.
Iversen, H., 2004. Risk-taking attitudes and risky driving behaviour. Transp. Res: F:
Trafc Psychol. Behav. 7 (3), 135150.
Jongen, E.M.M., Brijs, K., Brijs, T., Wets, G., 2013. Inhibitory control and peer pas-
sengers predict risky driving in young novice driversa simulator study. In:
Proceedings of the Seventh International Driving Symposium on Human Factors
in Driver Assessment, Training, and Vehicle Design, WA.
Lahatte, A., Le Pape, M., 2008. Is the way young people drive a reection of the way
their parents drive? An econometric study of the relation between parental risk
and their childrens risk. Risk Anal. 28 (3), 627634 (An Ofcial Publication of
the Society for Risk Analysis).
Laird, R.D., 2014. Parenting adolescent drivers is both a continuation of parenting
from earlier periods and an anticipation of a new challenge. Accid. Anal. Prev.
Lajunen, T., Corry, A., Summala, H., Hartley, L., 1997. Impression management and
self-deception in trafc behaviour inventories. Pers. Individ. Differences 22 (3),
341353.
Lam, L.T., 2001. Factors associated with parental safe road behaviour as a pedestrian
with young children in metropolitan New South Wales, Australia. Accid. Anal.
Prev. 33 (2), 203210.
Lee, J.D., 2007. Technology and teen drivers. J. Saf. Res. 38 (2), 203213.
Levenson, H., 1973. Reliability and validity of the I,P, and C scalesa multidi-
mensional view of locus of control. In: Presented at the Proceedings from the
American Psychological Association Convention, Montreal, Canada, Retrieved
from http://www.eric.ed.gov/ERICWebPortal/detail?accno=ED087791.
Loubeau, P.R., 2000. Exploration of the barriers to bicycle helmet use among 12 and
13 year old children. Accid. Anal. Prev. 32 (1), 111115.
Mann, H., 2010. Predicting young driver behaviour from pre-driver attitudes, inten-
tions and road behaviour. In: Thesis. Heriot-Watt University, Retrieved from
http://www.ros.hw.ac.uk/handle/10399/2396.
Martinussen, L.M., Hakamies-Blomqvist, L., Mller, M., zkan, T., Lajunen, T., 2013.
Age, gender, mileage and the DBQ: the validity of the Driver Behavior Question-
naire in different driver groups. Accid. Anal. Prev. 52, 228236.
Maximo, S.I., Tayaban, H.S., Cacdac, G.B., Cacanindin, M.J.A., Pugat, R.J.S., Rivera, M.F.,
Lingbawan, M.C., 2011. Parents communication styles and their inuence on
adolescents attachment, intimacy and achievement motivation. Int. J. Behav.
Sci. 6 (1), 6074.
Meyers, L.S., Gamst, G., Guarino, A.J., 2013. Applied multivariate research: design
and interpretation (second.). SAGE Publications, Thousand Oaks, CA.
Miller, G., Taubman-Ben-Ari, O., 2010. Driving styles among young novice
driversthe contribution of parental driving styles and personal characteristics.
Accid. Anal. Prev. 42 (2), 558570.
Morrongiello, B.A., Barton, B.K., 2009. Child pedestrian safety: parental supervision,
modeling behaviors, and beliefs about child pedestrian competence. Accid. Anal.
Prev. 41 (5), 10401046.
Notani, A.S., 1998. Moderators of perceived behavioral controls predictiveness in
the theory of planned behavior: a meta-analysis. J. Consum. Psychol. 7 (3),
247271.
Ozkan, T., Lajunen, T., 2005. Multidimensional trafc locus of control scale (T-LOC):
factor structure and relationship to risky driving. Pers. Individ. Differences 38
(3), 533545.
Paris, H., Broucke, S.V. den., 2008. Measuring cognitive determinants of speeding: an
application of the theory of planned behaviour. Transp. Res: F: Trafc Psychol.
Behav. 11 (3), 168180.
Paulhus, D.L., 1991. Measurement and control of response bias. In: Robinson, J.P.,
Shaver, P.R., Wrightsman, L.S. (Eds.), Measures of Personality and Social Psycho-
logical Attitudes. Academic Press, New York, NY, pp. 1759.
Peterson, R.A., 2001. On the use of college students in social science research: insights
from a second-order meta-analysis. J. Consum. Res. 28 (3), 450461.
Prato, C.G., Toledo, T., Lotan, T., Taubman-Ben-Ari, O., 2010. Modeling the behavior
of novice young drivers during the rst year after licensure. Accid. Anal. Prev.
42 (2), 480486.
Schmidt, S., Morrongiello, B.A., Colwell, S.R., 2014. Evaluating a model linking
assessed parent factors to four domains of youth risky driving. Accid. Anal. Prev.
69, 4050.
Schneewind, K.A., 1999. Impact of family processes on control beliefs. In: Ban-
dura, A. (Ed.), Self-efcacy in Changing Societies. Cambridge University Press,
Cambridge; NY.
Scott-Parker, B., Watson, B.C., King, M.J., 2009. Exploring how parents and peers
inuence the behaviour of young drivers. In: In Proceedings of the 2009 Aus-
tralasian Road Safety Research, Policing and Education Conference: Smarter,
Safer Directions, Sydney Convention and Exhibition Centre, Sydney, Retrieved
from http://www.rsconference.com.
Sheppard, B.H., Hartwick, J., Warshaw, P.R., 1988. The theory of reasoned action:
a meta-analysis of past research with recommendations for modications and
future research. J. Consum. Res. 15, 325343.
Sherman, K., Lapidus, G., Gelven, E., Banco, L., 2004. New teen drivers and their
parents: what they know and what they expect. Am. J. Health Behav. 28 (5),
387396.
Shope, J.T., Waller, P.F., Raghunathan, T.E., Patil, S.M., 2001. Adolescent antecedents
of high-risk driving behavior into young adulthood: substance use and parental
inuences. Accid. Anal. Prev. 33 (5), 649658.
Simons-Morton, B.G., Hartos, J.L., Beck, K.H., 2004. Increased parent limits on teen
driving: positive effects from a brief intervention administered at the motor
vehicle administration. Prev. Sci. 5 (2), 101111.
Simons-Morton, B.G., Hartos, J.L., Leaf, W.A., 2002. Promoting parental management
of teen driving. Inj. Prev. 8 (suppl 2), ii24ii31.
Simons-Morton, B., Hartos, J.L., Leaf, W.A., Preusser, D.F., 2006. Do recommended
driving limits affect teen-reported trafc violations and crashes during the rst
12 months of independent driving? Trafc Inj. Prev. 7 (3), 238247.
Sinclair, M., 2013. Attitudes, norms and driving behaviour: a comparison of young
drivers in South Africa and Sweden. Transp. Res: F: Trafc Psychol. Behav. 20,
170181.
Sparks, P., Hedderley, D., Shepherd, R., 1992. An investigation into the relation-
ship between perceived control, attitude variability and the consumption of
two common foods. Eur. J. Soc. Psychol. 22 (1), 5571.
Taubman-Ben-Ari, O., Katz-Ben-Ami, L., 2012. The contribution of family climate for
road safety and social environment to the reported driving behavior of young
drivers. Accid. Anal. Prev. 47, 110.
Taubman- Ben-Ari, O., Mikulincer, M., Gillath, O., 2005. From parents to
childrensimilarity in parents and offspring driving styles. Transp. Res: F: Trafc
Psychol. Behav. 8 (1), 1929.
Taubman-Ben-Ari, O., Katz-Ben-Ami, L., 2013. Family climate for road safety: a new
concept and measure. Accid. Anal. Prev. 54, 114.
Toledo, T., Lotan, T., Taubman - Ben-Ari, O., Grimberg, E., 2012. Evaluation of
a program to enhance young drivers safety in Israel. Accid. Anal. Prev. 45,
705710.
Tramow, D., Finlay, K.A., 1996. The importance of subjective norms for a minority
of people: between subjects and within-subjects analyses. Pers. Soc. Psychol.
Bull. 22 (8), 820828.
Ulleberg, P., Rundmo, T., 2003. Personality, attitudes and risk perception as predic-
tors of risky driving behaviour among young drivers. Saf. Sci. 41 (5), 427443.
Van den Putte, B., 1991. 20 Years of the Theory of Reasoned Action of Fishbein and
Ajzen: A Meta-analysis. University of Amsterdam, Amsterdam (Unpublished
manuscript).
Verlegh, P.W.J., Steenkamp, J.-B.E.M., 1999. A review and meta-analysis of country-
of-origin research. J. Econ. Psychol. 20 (5), 521546.
Walker, L.J., Frimer, J.A., 2007. Moral personality of brave and caring exemplars. J.
Pers. Soc. Psychol. 93 (5), 845860.
Williams, G.C., Hedberg, V.A., Cox, E.M., Deci, E.L., 2000. Extrinsic life goals and
health-risk behaviors in adolescents. J. Appl. Soc. Psychol. 30 (8), 17561771.
Wilson, R.J., Meckle, W., Wiggins, S., Cooper, P.J., 2006. Young driver risk in relation
to parents retrospective driving record. J. Saf. Res. 37 (4), 325332.