You are on page 1of 8

Accident Analysis and Prevention 36 (2004) 231–238

The Manchester Driver Behaviour Questionnaire: a cross-cultural study


Timo Lajunen a,∗ , Dianne Parker b , Heikki Summala c
a Department of Psychology, Middle East Technical University, 06531 Ankara, Turkey
b Department of Psychology, University of Manchester, Manchester, UK
c Traffic Research Unit, Department of Psychology, University of Helsinki, Helsinki, Finland
Received 8 August 2002; accepted 26 November 2002

Abstract
The aim of the present study was to investigate if the original factorial structure of the Manchester Driver Behaviour Questionnaire
(DBQ) was replicated in Finland and The Netherlands. A postal questionnaire survey of drivers was carried out in Britain, Finland and
The Netherlands. Exploratory factor analysis together with target (Procrustes) rotation and factorial agreement indexes were calculated to
investigate the applicability of Finnish and Dutch versions of DBQ. Results of the factor comparisons showed that the DBQ four-factor
structures found in Finland and The Netherlands were congruent but not perfect with the target structure found in Britain. Reliabilities of
the scales were around the same level as in the British data. In addition to the four first-order factors, two second-order factors (deliberate
violations and unintentional errors) were found in all three countries which supports the original structure by Reason et al. [Ergonomics
33 (1990) 1315]. Issues related to cross-cultural use of traffic behaviour questionnaires are discussed.
© 2003 Elsevier Science Ltd. All rights reserved.
Keywords: Manchester Driver Behaviour Questionnaire; Factor structure; Cross-cultural; Second-order factors

1. Introduction attention and memory failures, which can cause embar-


rassment but are unlikely to have an impact driving safety
It has been estimated that in 90–95% of traffic crashes (Parker et al., 1995). Since errors and violations result from
human actions are a sole or a contributory factor (Rumar, different psychological processes, they should be treated
1985). Hence, most of the traffic crashes result from driver differently (Reason et al., 1990). Subsequently, Lawton et al.
malfunctioning rather than from a technical failure of the (1997) extended the violations scale by adding new items.
vehicle. Based on his model of human error, Reason et al. Factor analysis of this extended violations scale showed that
(1990) divided human risk behaviour to errors and viola- violations can be split into two distinctive scales according
tions, and developed a survey instrument, Driver Behaviour to the reason why drivers violate. Aggressive violations
Questionnaire (DBQ), to measure these concepts in driver contain an interpersonally aggressive component whereas
behaviour. “ordinary” violations are deliberate deviations from safe
In their first study about DBQ, Reason et al. (1990) driving without a specifically aggressive aim. In the DBQ
showed that driver errors and violations are two empiri- literature, mainly violations—not errors or lapses—have
cally distinct classes of behaviour. They defined errors as been related to crash involvement. Among elderly drivers,
‘the failure of planned actions to achieve their intended however, relatively high error and lapse scores have been
consequences’ and violations as ‘deliberate deviations from reported to predict involvement in an active accident, while
those practices believed necessary to maintain the safe op- passive accident involvement has been associated with high
eration of a potentially hazardous system’. Unlike errors, scores on the lapse factor (Parker et al., 2000).
violations were seen as deliberate behaviours, although both In addition to studies conducted in the UK, the DBQ has
errors and violations are potentially dangerous and could been used in several international studies. To date, DBQ
lead to a crash. Reason et al. also found a third DBQ factor, data have been collected in Australia (Blockey and Hartley,
which they named “slips and lapses”. This factor included 1995), China (Xie et al., 2003), Greece (Kontogiannis et al.,
2002), Finland and The Netherlands (Lajunen et al., 1999;
∗ Corresponding author. Tel.: +90-312-210-3126; Mesken et al., 2002), New Zealand (Sullman et al., 2000),
fax: +90-312-210-1288. Sweden (Åberg and Rimmö, 1998), and Turkey (Sümer
E-mail address: timo@metu.edu.tr (T. Lajunen). et al., 2002). The original three-factor (errors, violations,

0001-4575/$ – see front matter © 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0001-4575(02)00152-5
232 T. Lajunen et al. / Accident Analysis and Prevention 36 (2004) 231–238

lapses) or four-factor (errors, lapses, aggressive and target rotations were performed to similar large samples of
“ordinary” violations) structure has been broadly repli- British, Dutch and Finnish drivers.
cated in most studies, although small differences in factor
structures have been reported. In their study conducted in
Western Australia, Blockey and Hartley (1995) extracted 2. Method
three factors, distinguishing between violations and dan-
gerous errors. The content of the factors differed slightly 2.1. Participants and procedure
from those of Reason et al. (1990), the factors being named
as general errors, dangerous errors and dangerous viola- The data reported here were collected as a part of a large
tions. Åberg and Rimmö (1998) factor analysed the DBQ survey of aggressive driving. Samples of 2000 drivers were
responses of a large sample of Swedish drivers, finding selected from the Finnish register of car owners, the electoral
the same error and violation factors as in Reason et al.’s register in the UK and the register of telephone users in
(1990) original study. The lapses, however, were split into The Netherlands. In Finland and Britain, an equal number
inattention and inexperience errors. Sullman et al. (2000) of men and women were selected. Since it was not possible
studied the DBQ responses of truck drivers in New Zealand, to get this kind of stratified sample in The Netherlands,
and found four factors, which they named as errors, lapses, the Dutch sample was a random sample of telephone users.
violations and aggressive violations. The most distinctive Samples were representative of the whole geographical area
factor structure has been found among Chinese drivers. Xie of the countries involved. A questionnaire together with a
et al. (2003) administered a 24-item version of the DBQ cover letter and a Freepost return envelope were sent to
to a sample of 363 Chinese drivers including both profes- participants. The participants replied to the questionnaire
sional and non-professional drivers. The exploratory factor anonymously. The questionnaires were each given a code
analysis resulted in three factors. The first factor included in order to assess who had responded. After 3 weeks, a
six error and six lapse items whereas the second factor had reminder letter was posted to increase the response rate. The
eight violation and one lapse item. The third factor con- response rate after the reminder was 42% in the UK, 35% in
sisted of two error and two lapse items. Hence, the main The Netherlands and 56% in Finland. The higher response
distinction between errors and violations was found also in rate for Finland was partly due to the fact that the sampling in
this study, although the factor structure was different from Finland was targeted only at drivers whereas in Britain and
the usual three- or four-factor structure found in Western The Netherlands the questionnaires were sent to electors and
countries. telephone users, respectively, some of whom may not have
Although the three- or four-factor structure of the DBQ had a driving licence. To estimate the “true” response rates,
seems to occur in various studies conducted in different the response rates gained in Britain and The Netherlands
countries and language areas, the cross-cultural stability of would have to be scaled in proportion to the percentage
the DBQ is far from that of “Big Five” (Costa and McCrae, of driving licence holders in the countries concerned. The
1985) or Eysenckian three personality factors (Eysenck and characteristics of the Finnish, British and Dutch samples are
Eysenck, 1991/1996). Most of the studies using the DBQ presented in Table 1.
have established “national scoring keys” for the DBQ. For
example, the Swedish DBQ makes a distinction between 2.2. Measures
inattention and inexperience errors. Although the specifica-
tion of a national factor solution may provide important in- 2.2.1. Driver Behaviour Questionnaire (DBQ)
formation for studies conducted in one particular country, In the present study, the extended 27-item Driver Be-
cross-national comparisons become difficult. haviour Questionnaire (Lawton et al., 1997; Parker et al.,
In addition to different factor structures and scoring keys, 1998) was translated into Dutch and Finnish. Lapses, errors,
the number of items has varied considerably between DBQ aggressive and “ordinary” violations items were included.
versions used in different studies. Since the number of items The items are listed in Table 2. Respondents were asked to
can influence the psychometric characteristics of the scale, indicate how often they themselves do each of the viola-
the comparisons between studies using, say, 24-item and the tions and errors when driving. Responses were recorded on
104-item scale used in Sweden by Åberg and Rimmö (1998) a six-point scale from “Never” to “Nearly all the time”.
are problematic. In conclusion, various versions of the DBQ,
different sampling strategies (e.g. postal survey, road-side 2.2.2. Demographic and exposure measures
surveys) and different target populations (e.g. professional Respondents answered questions about their age, sex, and
drivers, elderly drivers) all undermine the ease with which estimated annual mileage.
cross-cultural comparisons of the DBQ factor structure can
be made. The aim of the present study is to investigate 2.3. Statistical analyses
the equivalence of the DBQ factor structures in samples of
British, Dutch and Finnish drivers. Traditional psychometric The internal consistency of the DBQ scale scores
analysis (reliabilities) and exploratory factor analysis with was assessed by calculating Cronbach’s alpha reliability
T. Lajunen et al. / Accident Analysis and Prevention 36 (2004) 231–238 233

Table 1
Characteristics of the British, Finnish and Dutch samples
Britons (n = 831) Finns (n = 1123) Dutch (n = 703)

Mean age (S.D.) 39.55 (14.51) 37.52 (15.12) 45.90 (14.75)


Mean driving experience in years 18.35 (12.91) 16.93 (12.94) 23.62 (13.11)
Mean annual mileage 18612 (20894) 20435 (21730) 24637 (34635)
Males (%) 51.3 45.9 70.4
Accident involved in previous 3 years (%) 31.3 22.5 28.2

Table 2
Means and standard deviations of the DBQ answers in the British, Finnish and Dutch data
Britons Finns Dutch

Mean S.D. Mean S.D. Mean S.D.

Aggressive violations 0.71 0.70 0.53 0.69 0.49 0.55


7. Sound your horn to indicate your annoyance to another road user 1.06 1.08 0.72 0.99 0.79 0.91
17. Become angered by another driver and give chase with the intention of giving 0.20 0.59 0.46 0.82 0.11 0.40
him/her a piece of your mind
25. Become angered by a certain type of a driver and indicate your hostility by 0.85 0.96 0.41 0.77 0.56 0.77
whatever means you can
“Ordinary” violations 0.98 0.70 0.94 0.62 0.94 0.60
10. Pull out of a junction so far that the driver with right of way has to stop and let 0.82 0.99 0.29 0.54 0.52 0.70
you out
11. Disregard the speed limit on a residential road 1.38 1.21 2.04 1.32 1.40 1.17
18. Stay in a motorway lane that you know will be closed ahead until the last 0.52 0.85 0.34 0.68 0.61 0.81
minute before forcing your way into the other lane
20. Overtake a slow driver on the inside 0.70 0.95 0.22 0.59 0.79 0.95
21. Race away from traffic lights with the intention of beating the driver next to you 0.99 1.15 1.07 1.16 1.36 1.15
23. Drive so close to the car in front that it would be difficult to stop in an emergency 0.80 0.86 1.13 0.95 0.66 0.78
24. Cross a junction knowing that the traffic lights have already turned against you 0.78 0.84 0.86 0.87 0.47 0.67
28. Disregard the speed limit on a motorway 1.87 1.61 1.55 1.38 1.73 1.38
Errors 0.50 0.38 0.51 0.38 0.63 0.47
5. Queuing to turn left onto a main road, you pay such close attention to the main 0.63 0.71 0.56 0.69 0.52 0.65
stream of traffic that you nearly hit the car in front
6. Fail to notice that pedestrians are crossing when turning into a side street from a 0.44 0.61 0.72 0.72 0.67 0.88
main road
8. Fail to check your rear-view mirror before pulling out, changing lanes, etc. 0.71 0.82 0.77 0.78 1.21 1.75
9. Brake too quickly on a slippery road or steer the wrong way in a skid 0.68 0.70 0.57 0.68 0.71 0.73
13. On turning left nearly hit a cyclist who has come up on your inside 0.31 0.55 0.22 0.47 0.47 0.57
14. Miss “Give Way” signs and narrowly avoid colliding with traffic having right of way 0.24 0.47 0.21 0.49 0.39 0.60
16. Attempt to overtake someone that you had not noticed to be signalling a right turn 0.24 0.47 0.23 0.49 0.37 0.57
27. Underestimate the speed of an oncoming vehicle when overtaking 0.74 0.70 0.77 0.69 0.64 0.67
Lapses 0.97 0.48 0.87 0.45 0.79 0.43
1. Hit something when reversing that you had not previously seen 0.45 0.62 0.49 0.62 0.65 0.68
2. Intending to drive to destination A, you “wake up” to find yourself on the road to 0.86 0.92 10.02 0.92 0.77 0.84
destination B
4. Get into the wrong lane approaching a roundabout or a junction 1.45 0.82 1.27 0.74 1.01 0.74
12. Switch on one thing, such as the headlights, when you meant to switch on 0.89 0.89 0.85 0.83 0.54 0.71
something else, such as the wipers
15. Attempt to drive away from the traffic lights in third gear 0.69 0.86 0.80 0.81 0.65 0.82
19. Forget where you left your car in a car park 1.16 1.09 0.48 0.75 0.78 0.88
22. Misread the signs and exit from a roundabout on the wrong road 1.03 0.78 0.93 0.76 0.86 0.71
26. Realise that you have no clear recollection of the road along which you have 1.17 1.03 1.14 0.99 1.03 0.92
just been travelling
234 T. Lajunen et al. / Accident Analysis and Prevention 36 (2004) 231–238

coefficients. The equivalence of the DBQ factor structures reliability coefficients were around the same level as those
in three countries involved were assessed first by deter- found in previous DBQ studies in general.
mining the number of factors present in each of the three
samples and then by comparing the rotated factor ma- 3.2. Exploratory factor analysis
trices by using Procrustes target rotation techniques and
factorial agreement coefficients. The techniques used to de- 3.2.1. Number of factors
termine the number of factors were the “eigenvalue greater The three methods (eigenvalues greater that one, Cattell’s
than one” rule (“mineigen criterion”), visual inspection of Scree test, parallel test) for deciding on the number of fac-
“Scree plots” and parallel analysis (Thompson and Daniel, tors produced slightly different results. In the Finnish and
1996). British data, five factors had eigenvalues greater than one
In order to investigate if the first-order factors constitute whereas in the Dutch data the mineigen criterion suggested
second-order factors, factor scores for the extracted factors six factors should be extracted. Obviously, the mineigen cri-
were calculated. Correlation matrices for these factor scores terion resulted in over-factoring. In all three datasets, visual
were then factor analysed (principal axis factoring, varimax inspection of Scree plots suggested the presence of only
rotation). three factors. The third method, parallel test, suggested four
In the target rotation procedure, one cultural group is ar- factors for the Finnish data, three for the British data and
bitrarily designated as the target group. In the present study, three for the Dutch data. However, the three-factor solution
the British drivers were used as a target group, because the was not readily interpretable in either dataset. Overall, then,
DBQ was developed originally using British samples. Sep- a four-factor solution seemed to be most feasible.
arate identical factor analyses (principal axis factoring with
oblimin rotation) were conducted for the British, Dutch and 3.2.2. Factor structure: first-order factors
Finnish DBQ correlation matrices. Then, the structure matri- Oblique rotation was applied, because factor correla-
ces obtained from the Finnish and Dutch data were target ro- tion matrices indicated that factors had relatively high
tated to match the British structure matrix. Finally, factorial inter-correlations in the Finnish (range: 0.11–0.50), British
agreement was evaluated by calculating identity, additivity, (range: 0.02–0.52) and Dutch data (range: 0.04–0.47).
proportionality (Tucker’s coefficient) and linearity (correla- Oblimin rotated factor matrices for all three data sets are
tion) agreement coefficients. In general, values higher than presented in Table 2. Four-factors accounted for 40.4% of
0.95 indicate factorial similarity, whereas values lower than variance in the British data, 42.4% in the Finnish data, and
0.90 (van de Vijver and Poortinga, 1994) or 0.85 (ten Berge, 39.0% in the Dutch data. Table 4 shows that in all three
1986) are taken as a sign of non-negligible incongruities. countries, aggressive violation items had the highest factor
loadings in the same factor. Among “ordinary” violations,
the situation was slightly less clear. Although the items
3. Results got the highest loadings in the same factor, especially in
the Finnish data the “ordinary” violation items also had
3.1. Reliability analysis relatively high factor loadings on the first factor, which
was the one labelled aggressive violations. For Finns, many
Alpha reliability coefficients for the DBQ scales for three of the “ordinary” violations have an aggressive flavour. It
nationalities are listed in Table 3. In all three samples, should be noted that item “forcing someone to give away”
“ordinary” violations scale seemed to be the most internally (number 10) loaded only on the “errors” factor, maybe
consistent (α = 0.75–0.80) whereas the “lapses” scale had because of some ambiguity of the Finnish translation. The
the lowest alpha values (α = 0.64–0.69). Reliability anal- behaviour described in the item could be interpreted either
ysis indicated, however, that removal of any of the items as deliberate or unintentional.
would have led to decrease in internal consistency. DBQ Table 4 shows that almost all error items had their high-
scale scores showed the highest reliability coefficients in est factor loading on one factor, which could be labelled
UK and lowest in The Netherlands. Although some al- “errors”. The item “not checking the mirror” (number 8),
pha reliability coefficients seemed unacceptably low, the however, had factor loadings lower than 0.40 in both the
Dutch and the British data. In addition, the item “misjudge
an overtaking gap” also had a relatively high cross loading
Table 3 on the “lapses” factor in all three countries. The “lapses”
Alpha reliability coefficients of the DBQ scales for the British, Finnish
items had the highest factor loadings in the same factor in
and Dutch data
all three data sets. The factor loadings of items 2, 4 and 12
Britons Finns Dutch were, however, lower than 0.40 in the Dutch data. In general,
Aggressive violations (three items) 0.73 0.68 0.65 the factor analysis with oblique rotation seemed to produce
“Ordinary” violations (eight items) 0.79 0.80 0.75 a relatively robust four-factor structure in all three countries.
Errors (eight items) 0.73 0.73 0.64 These factors could be interpreted as aggressive violations,
Lapses (eight items) 0.69 0.65 0.64
“ordinary” violations, errors, and lapses.
T. Lajunen et al. / Accident Analysis and Prevention 36 (2004) 231–238 235

Table 4
Results of the exploratory factor analysis with oblimin rotation for Britons, Finns and Dutch
Factor 1 Factor 2 Factor 3 Factor 4

UK FIN NL UK FIN NL UK FIN NL UK FIN NL

Aggressive violations
7. Annoyed, sounding your horn 0.67 0.65 0.63
17. Angry, giving chase 0.63 0.69 0.53
25. Angry, indicating hostility 0.69 0.74 0.80 0.43
“Ordinary” violations
10. Forcing someone to give way 0.47 0.48
11. Speeding on residential road 0.43 0.67 0.76 0.68
18. Forcing your way into the other lane 0.50 0.51 0.54
20. Overtaking on the inside 0.44 0.40 0.55 0.46
21. Racing away from traffic lights 0.47 0.50 0.63 0.56 0.49
23. Close following 0.51 0.59 0.45
24. Shooting lights 0.43 0.53 0.48 0.41
28. Speeding on motorway 0.44 0.74 0.71 0.80
Errors
5. Insufficient attention to vehicle ahead 0.49 0.47 0.52
6. Not noticing pedestrians crossing 0.56 0.59 0.50
8. Not checking mirror 0.43
9. Braking too quickly on slippery 0.43 0.46 0.47 0.41
13. Failing to notice a cyclist 0.57 0.50 0.59
14. Not noticing “Give Way” sign 0.62 0.60 0.65
16. Overtaking a right turner 0.52 0.54 0.51
27. Misjudge an overtaking gap 0.51 0.48 0.48 0.42 0.46 0.43
Lapses
1. Hitting something when reversing
2. Taking more usual route by error 0.41 0.48 0.42
4. Wrong lane at roundabout/junction 0.40 0.49
12. Going for the wrong switch 0.48 0.51
15. Driving away in third gear 0.42 0.38 0.50
19. Forgetting where the car is in a car park 0.52 0.43 0.54
22. Taking wrong exit from roundabout 0.43 0.42 0.47 0.52 0.53
26. No recollection of road travelled 0.52 0.45 0.57
Eigenvalues 2.35 3.29 2.27 3.98 2.89 2.92 3.51 4.01 3.10 2.91 3.31 2.64
For the sake of clarity, factor loadings less than 40 have been omitted.

3.2.3. Factor structure: second-order factors tor analysis (Table 4) are shown in Table 5. Table 5 shows
The high correlations among the four factors extracted that “errors” and “lapses” loaded on the first factor, labelled
in the exploratory factor analysis suggest that second-order “unintentional mistakes”, and that aggressive and “ordinary”
factors could emerge. It might be supposed that aggressive violations loaded on the second factor (“violations”). This
and “ordinary” violations would constitute a second-order factor solution accounted for 69.2% of the variance in the
“violations” factor and “errors” and “lapses” would form British data, 63.9% in the Finnish data, and 62.3% in the
a general “unintentional mistakes” factor. The results of a Dutch data.
factor analysis (principal axis extraction method with vari-
max rotation) of the factor scores from the first-order fac- 3.3. Target rotation and agreement coefficients

Table 5 Target rotations of the Finnish and Dutch factor matrices


Results of the second-order factor analysis with varimax rotation were carried out using the British factor matrix as a target.
Factor 1: mistakes Factor 2: violations As recommended by van de Vijver and Leung (1997), iden-
tity, additivity, proportionality and linearity agreement co-
UK FIN NL UK FIN NL
efficients were calculated. The values of these indexes are
Aggressive violations −0.80 0.72 0.87 listed in Table 6. Table 6 shows that the identity and pro-
“Ordinary” violations 0.44 0.69 0.73 0.64
portionality coefficients indicated high similarity (although
Errors 0.69 0.85 0.74
Lapses 0.96 0.78 0.83 not full identity) between the Finnish and British structures
as well as between the Dutch and British DBQ factor struc-
Eigenvalues 1.60 1.42 1.27 1.17 1.14 1.22
tures. The values for additivity and the correlation coefficient
236 T. Lajunen et al. / Accident Analysis and Prevention 36 (2004) 231–238

Table 6
Four identity indexes for DBQ scales
Aggressive violations “Ordinary” violations Errors Lapses

FIN NL FIN NL FIN NL FIN NL

Identity coefficient 0.96 0.94 0.97 0.96 0.98 0.96 0.98 0.96
Additivity coefficient 0.92 0.90 0.86 0.90 0.88 0.85 0.86 0.85
Proportionality coefficient 0.96 0.94 0.97 0.97 0.98 0.96 0.98 0.97
Correlation coefficient 0.92 0.90 0.87 0.91 0.88 0.88 0.86 0.86

were somewhat lower, especially in relation to the error and degree of factorial agreement is established by comparing
lapses factor, but still acceptable. It should be noted that the the rotated factor structure obtained in one culture with a
most widely applied index, i.e. coefficient of proportionality target structure (the British factor structure in this case) and
(also known as Tucker’s phi), and the most stringent index calculating indices of agreement. Unlike structural equa-
(identity coefficient) both indicated a high degree of simi- tion techniques, target rotation techniques, used together
larity between the factor structures. with agreement indices, are optimal exploratory techniques
when we do not have a strong a priori theory from which to
formulate hypotheses about cross-cultural differences and
4. Discussion similarities (McCrae et al., 1996; van de Vijver and Leung,
1997). In the present study, the objective was to investi-
Self-reports can be a very useful and efficient means for gate if the DBQ functioned in the same way in Finland
studying aberrant driving behaviour. At their best, anony- and The Netherlands as in Britain where it was originally
mous surveys can provide reliable in-depth information developed.
about behaviour, as well as about the motives and attitudes The results of the factor comparisons show that the DBQ
leading to risky driving. One of the widely used instruments four-factor structures (aggressive violations, “ordinary” vi-
for measuring self-reported driving is Manchester Driver olations, errors, lapses) found in Finland and The Nether-
Behaviour Questionnaire (DBQ). In addition to many studies lands were fairly congruent with the one found with the
conducted in UK, the DBQ has been used in several studies British data. It should be noted, however, that the factorial
in other countries. In these studies, the DBQ has been trans- agreement was satisfactory rather than perfect. This find-
lated at least to Chinese, Dutch, Finnish, Greek, Swedish, ing indicates that the four-factor structure of the DBQ can
and Turkish. Hence, the DBQ could potentially provide a be used in studies using Finnish and Dutch versions of the
means for international comparisons of driving behaviour. DBQ. On the other hand, the fact that the agreement between
Most of these DBQ studies have investigated the factor the structures was not perfect emphasises the importance of
structure of the DBQ by using exploratory factor analysis. taking cultural factors into account when applying an in-
Based on these analyses, it has been concluded that the fac- strument in another country and language. As Blockey and
tor structure based on the distinction to errors and violations Hartley’s (1995) study shows, an instrument may function
is robust, although some individual items may not behave differently even in two countries using the same language.
as expected. In some cases, a new “national” DBQ scor- In the present study, all three countries were industrialised
ing key has been proposed. Despite these cultural nuances, Western European countries with relatively safe traffic and
the distinction between errors as unintentional mistakes superficially similar traffic cultures. The small differences
and violations as deliberate acts has been supported by all in DBQ structure show that different behaviours in gen-
international studies. eral, and questionnaire items in particular, can be interpreted
However, although several international studies have slightly differently even in countries as culturally similar as
supported the universal nature of the DBQ factor struc- Finland, The Netherlands and the UK. This finding empha-
ture based on errors and violations, these conclusions have sises the importance of thorough knowledge of the countries
often been based solely on exploratory factor analysis of and cultures involved and great care in the translation pro-
data obtained in one country. Comparisons with the orig- cedure. In the present study, the Finnish translation of the
inal structure have been based mostly on visual inspection DBQ was checked by a group of experienced traffic psychol-
of the rotated factor matrix. Although visual inspection ogists, who were fluent in both Finnish and English. Even
helps in the identification of problematic items, it is not so, the Finnish translation of the item “forcing someone to
sufficiently rigorous for drawing conclusions about the give way” (“ordinary” violations item) was ambiguous and
similarity between factor structures, because the degree of the item was open to interpretation either as an unintentional
congruence/incongruence cannot be measured. One of the error or as a deliberate violation. The use of a procedure
most powerful methods for investigating the cross-cultural such as back translation by an independent translator might
feasibility of an instrument’s factor structure is exploratory have revealed the small but important difference between
factor analysis with target rotation. Using this method, the the original item and translation.
T. Lajunen et al. / Accident Analysis and Prevention 36 (2004) 231–238 237

Problems in translation procedure are not the only the level of reporting aberrant behaviours in general, socially
reason for lack of equivalence between the original question- desirable responding is likely to bias responses to different
naire and the translation. Often differences in factor struc- DBQ items in different cultures. In their quasi-experimental
tures reflect true cultural differences. Like general cultures, study, Lajunen and Summala (2003) compared the DBQ
traffic cultures vary from one region to another. The DBQ responses given in public and private settings. Self-reports
item “brake too quickly on slippery road” has very different of drinking and driving were biased by social desirability
meanings in countries with a long snowy winter and in coun- whereas drivers filling the DBQ form anonymously did not
tries where snow tires are never needed. Similarly, “failing report more speeding than those in the public settings. This
to notice cyclists” may be a very salient item in The Nether- reflects the fact that drinking and driving is one of the most
lands, a country with many cyclists, but less relevant, say, in socially reprehensible behaviours in Scandinavia whereas in
Turkey where cycling is not popular due to the hilly terrain. some other European countries drinking small doses of al-
Violation items seem to be especially problematic, because cohol might be considered as less condemnable. According
the distinction between “ordinary” violations and aggressive to SARTRE study, 71% of Swedish drivers report that they
violations is solely based on the intention behind the act never drive “after drinking even a small amount of alcohol”
(Lajunen and Parker, 2001; Lajunen et al., 1998). For exam- whereas the same figure for Italians was 8% (SARTRE,
ple, sounding one’s horn clearly reflects aggression in Scan- 1994). Hence, the DBQ item about drinking and driving can
dinavia whereas in Southern Europe the threshold for horn be expected to be very sensitive to social desirability bias
use is much lower and it does not necessarily indicate ag- and local (drinking) culture. At the present study, the DBQ
gression. These kind of culturally sensitive items may need item about drinking and driving was not used.
to be removed from the DBQ. One possibility is to construct In conclusion, the factors structures of the DBQ were
two versions of the DBQ for each country: one for national very similar, but not identical, in three Western European
use only (including culturally relevant items) and a shorter countries with relatively safe traffic and similar driving cul-
version for international comparisons. The items in the inter- ture. These results indicate that DBQ can be used and the
national version should form the core items of the longer na- scale scores compared with confidence at least in Western
tional version to secure international comparability of data. European countries. In non-European countries with less
Originally, the idea behind the DBQ was to differen- motorization, the DBQ might have worked very differently.
tiate deliberate violations from unintentional errors. Later Therefore, proper adaptation studies are needed before
versions of DBQ have usually had between three and five DBQ can be used outside the cultural group for which it
scales. Although these scales measure different types of er- was originally developed.
rors (lapses and errors) or violations (aggressive and or-
dinary violations) and in this way replicate the original
theoretical distinction, all previous studies have used only Acknowledgements
first-order factor analyses. In most of the studies, the scales
have had relatively strong inter-correlations. This finding We would like to thank Dr. Jolieke Mesken for her help
indicates that second-order factors could emerge. In the in data collection and translation of the questionnaires.
present study, factor analysis of the factor scores of the four This research was supported by the grants of the European
first-order factors showed that two second-order factors were Commission (contract no. ERBFMBICT972398), and the
present in the British, Dutch and Finnish data. This shows Finnish Organisation for Traffic Safety (Liikenneturva).
that the four factors (aggressive and “ordinary” violations, Ajoneuvohallintokeskus is thanked for providing the names
errors, lapses) reflect the original distinction between delib- and addresses of car owners in Finland.
erate violations and involuntary mistakes. However, despite
the theoretical importance of this finding, in every-day use
the scoring based on four scales may be more informative References
for road safety practitioners.
The results of the present study showed that the Dutch and Åberg, L., Rimmö, P.-A., 1998. Dimensions of aberrant driver behaviour.
Finnish DBQ factor structures are roughly equivalent to the Ergonomics 41, 39–56.
British one. This indicates that the translations were quite Blockey, P.N., Hartley, L.R., 1995. Aberrant driving behaviour: errors and
violations. Ergonomics 38, 1759–1771.
successful and that drivers interpreted most of the items in Costa, P.T., McCrae, R.R., 1985. The NEO Personality Inventory manual.
a similar way in all three countries. Although these results Psychological Assessment Resources, Odessa, FL.
show that the structure of the DBQ is about the same in three Eysenck, H.J., Eysenck, S.B.G., 1991/1996. Manual of the Eysenck
countries involved and, therefore, make international com- Personality Scales (EPS Adult). Hodder & Stoughton, London.
parisons possible, it does not indicate total absence of cul- Kontogiannis, T., Kossiavelou, Z., Marmaras, N., 2002. Self-reports of
aberrant behaviour on the roads: errors and violations in a sample of
tural bias. It is still possible that in one country aggressive
Greek drivers. Accid. Anal. Prev. 34, 381–399.
drivers score lower on the DBQ aggressive violations than in Lajunen, T., Parker, D., 2001. Are aggressive people aggressive drivers?: a
another country where expressions of aggression are socially study of the relationship between self-reported general aggressiveness,
more acceptable. In addition to these possible differences in driver anger and aggressive driving. Accid. Anal. Prev. 33, 243–255.
238 T. Lajunen et al. / Accident Analysis and Prevention 36 (2004) 231–238

Lajunen, T., Summala, H., 2003. Can we trust self-reports of driving? Rumar, K., 1985. The role of perceptual and cognitive filters in observed
Transport. Res. Part F, in press. behaviour. In: Evans, L., Schwing, R.C. (Eds.), Human Behaviour and
Lajunen, T., Parker, D., Stradling, S., 1998. Dimensions of driver anger, Traffic Safety. Plenum Press, New York, pp. 151–165.
aggressive and highway code violations and their mediation by safety SARTRE, 1994. European Drivers and Traffic Safety. Presses de l’Ecole
orientation in UK drivers. Transport. Res. Part F 1, 107–121. Nationale des Ponts et Chaussées.
Lajunen, T., Parker, D., Summala, H., 1999. Does traffic congestion Sullman, M., Meadows, M., Pajo, K., 2000. Errors, lapses and violation
increase driver aggression? Transport. Res. Part F 2, 225–236. in the drivers of heavy vehicles. In: Proceedings of the Paper Presented
Lawton, R., Parker, D., Manstead, A.S.R., Stradling, S., 1997. The role of at the International Conference on Traffic and Transport Psychology,
affect in predicting social behaviours: the case of road traffic violations. Berne, Switzerland.
J. Appl. Soc. Psychol. 27, 1258–1276. Sümer, N., Lajunen, T., Özkan, T., 2002. Sürücü davranislarinin kaza
McCrae, R.R., Zoderman, A.B., Costa, P.T., Bond, M.H., Paunonen, S.V., riskindeki rolü: ihlaller ve hatalar (The role of driver behaviour in
1996. Evaluating replicability of factors in the revised NEO personality accident risk: violations and errors). In: International Traffic and Road
inventory: confirmatory factor analysis versus Procrustes rotation. J. Safety Congress & Fair, Ankara, Turkey.
Pers. Soc. Psychol. 70, 552–566. ten Berge, J.M.F., 1986. Rotatie naar perfecte congruentie en de Multipele
Mesken, J., Lajunen, T., Summala, H., 2002. Interpersonal violations, Groep Methode. Nederlands Tijdschrift voor Psychol. 41, 218–225.
speeding violations and their relation to accident involvement in Thompson, B., Daniel, F.G., 1996. Factorial analytic evidence for the
Finland. Ergonomics 7, 469–483. construct validity of scores: a historical overview and some guidelines.
Parker, D., Reason, J.T., Manstead, A.S.R., Stradling, S., 1995. Driving Educ. Psychol. Meas. 56, 197–208.
errors, driving violations and accident involvement. Ergonomics 38, van de Vijver, F., Leung, K., 1997. Methods and Data Analysis for
1036–1048. Cross-cultural Research. Sage, London.
Parker, D., Lajunen, T., Stradling, S., 1998. Attitudinal predictors of van de Vijver, F., Poortinga, Y.H., 1994. Methodological issues in
aggressive driving violations. Transport. Res. Part F 1, 11–24. cross-cultural studies on parental rearing behaviour and psychopat-
Parker, D., McDonald, L., Rabbitt, P., Sutcliffe, P., 2000. Elderly drivers hology. In: Perris, C., Arrindell, W.A., Eisemann, M. (Eds.), Parental
and their accidents: the Aging Driver Questionnaire. Accid. Anal. Prev. Rearing and Psychopathology. Wiley, Chicester, UK, pp. 173–
32, 751–7599. 197.
Reason, J.T., Manstead, A.S.R., Stradling, S.G., Baxter, J.S., Campbell, K., Xie, C.-Q., Parker, D., 2003. A social psychological approach to
1990. Errors and violations on the road: a real distinction? Ergonomics driving violations in two Chinese cities. Transport. Res. Part F,
33, 1315–1332. in press.

You might also like