scispace - formally typeset
Search or ask a question
Journal ArticleDOI

Bus driver accident record: the return of accident proneness

28 Feb 2009-Theoretical Issues in Ergonomics Science (Taylor & Francis Group)-Vol. 10, Iss: 1, pp 77-91
TL;DR: In this paper, the authors investigated the stability of accident record over time in five samples of British bus drivers and found that the size of correlations between time periods increased with the increase in mean accident frequency.
Abstract: One of the assumptions of the theory of accident proneness is that drivers’ accident liability is stable over time, which was tested in the present paper. Previous investigations of this problem (or rather the conclusions) were found to be deficient because they did not take into account the statistical problem of low variance in the accident variable. However, by correlating the between time periods association coefficient and the mean number of accidents across several samples, this problem can be overcome. Therefore, the stability of accident record over time was investigated in five samples of British bus drivers. It was found that the size of the correlations between time periods increased with the increase in mean accident frequency. Furthermore, this increase could be described by a linear regression line, which fit the various points extremely well. Also, the size of correlations of ‘at fault’ accidents increased faster with the mean than did ‘all accidents’, although the latter had a higher initi...

Summary (2 min read)

1.1 History and theory of accident proneness

  • The beginning of safety research has often been traced to British studies about the distributions of accidents in various populations (mainly industrial workers) in the early 20th century (e.g. Greenwood & Woods, 1919; Newbold, 1927; Slocombe & Brakeman, 1930).
  • The conclusion drawn from this result is that the liability to have accidents is not only differently distributed in a population, but it is also surprisingly stable over time within each individual (note that most drivers in a year will have no accidents, as will they in the next year, an amazing stability that is not reflected by a correlation).
  • So, although most of the researchers who have used culpable accidents have found lower correlations between time periods, as compared to all accidents, and concluded that the effect of such restrictions are negative, they have not taken into account the effect of the lowered mean, or checked their culpability criterion for correctness.
  • First, it was investigated whether the intercorrelations between accidents in different time periods are determined by the mean level of accidents in the sample.

2.1 Subjects and data

  • A UK bus company made traffic accident data available for analyses from five different geographical regions for the time period 2001-2005.
  • For the present samples, the number of routes driven per working week ranged from 2 to 15 and these tend to be 10 driven by the same bus drivers operating from the same rotas week in and week out.
  • Given that there is a greater percentage of experienced drivers compared with novice drivers within a typical depot and swopping rotas tends to happen only occasionally, this eventuality is less common.
  • The basic variable was thus the number of accidents within a year for each driver.
  • For the present data, it has been shown that the criterion for culpability used by the company is too lenient (af Wåhlberg & Dorn, 2007), meaning that several percent of accidents that should have been classified as culpable have not been designated as such.

2.2 Methods of analysis

  • There are different ways of calculating the stability of accident record over time.
  • The simplest and most common is to take two time periods and correlate the numbers in each for the drivers.
  • For the next step two-year periods would be used, and so on until only two periods were left.
  • The only critical issue is whether there is enough difference in the accident means for an association to become evident.

2.3 Statistical methods

  • The preferred statistical tool for the analysis was Pearson correlation.
  • Results 13 First, accidents were correlated between single years for the two accident variables (All and culpable only) for all samples (see Table 2), and the means of these associations computed (first section of Table 3).
  • Third, mean correlations for single years against single years (Table 2) were computed by their proximity in time to the other year, with results that are shown in Table 4.
  • 14 Finally, experience was added as a factor in the analysis, by separating each sample into those drivers with less than a year’s experience in the beginning of 2001, and others.
  • It can be noted that despite the low numbers of drivers in the inexperienced groups (N=31-77) and the resulting possibility for chance deviations, the correlations still largely conformed to a straight regression line, with half the variation in correlation size explained by the mean number of accidents in the period.

4.1 Results

  • The results of the present analysis would seem to support the hypotheses stated in the introduction.
  • In the present study, exposure was only handled by arranging data per year.
  • This hypothesis is in principle testable in the present data, but the results would be hard to interpret, for several reasons.
  • Third, as exposure data was not available, it was not possible to calculate how strong the stability of exposure effect could be expected to be.
  • This would seem to indicate that some of the stability in accident record is indeed due to differential and stable exposure, but only for non-culpable crashes.

4.2 Methodology and consequences

  • One important distinction between concepts needs to be made.
  • After the many decades of criticism of the accident proneness theory, the authors tentatively suggest that the statistical arguments that caused it to fall into disrepute may well have been rather prematurely based.
  • It might be noted here that there does not seem to be any effect of low means such that the correlation coefficients become smaller than expected at the low end, as could possibly be expected due to the increasing violation of the assumption of normality of the distribution.
  • It is not the theory that is deficient, but the previous interpretations of the results.
  • The authors are also grateful to Jenny Stannard (Cranfield University) for extracting the relevant data from the company database.

5. References

  • Accident statistics and the concept of accident proneness.
  • Driver behaviour and accident research methodology; unresolved problems. af Wåhlberg, A. E., & Dorn, L. (2007).

Did you find this useful? Give us your feedback

Content maybe subject to copyright    Report

1
Bus driver accident record; the return of accident proneness
Anders af Wåhlberg* (1)
and
Lisa Dorn (2)
1. Department of Psychology
Uppsala University
P. O. Box 1225
751 42 Uppsala
Sweden
e-mail: anders.af_wahlberg@psyk.uu.se
Homepage: http://www.psyk.uu.se/hemsidor/busdriver
Tel: +46-18-471 25 90, +46-18-33 90 95
2. Department of Human Factors,
Cranfield University, Cranfield, Bedfordshire, MK43 0AL
UK
* corresponding author
Theoretical Issues in Ergonomics Science, Volume 10, Issue 1, 2009, Pages 77-91

2
Abstract
One of the assumptions of the theory of accident proneness is that drivers’ accident liability is stable over time
which was tested in the present paper. Previous investigations of this problem (or rather the conclusions) were
found to be deficient, because they did not take into account the statistical problem of low variance in the
accident variable. However, by correlating the between time periods association coefficient and the mean
number of accidents across several samples, this problem can be overcome. Therefore, the stability of accident
record over time was investigated in five samples of British bus drivers. It was found that the size of the
correlations between time periods increased with the increase in mean accident frequency. Furthermore, this
increase could be described by a linear regression line, which fit the various points extremely well. Also, the
size of correlations of At fault accidents increased faster with the mean than did All accidents, although the
latter had a higher initial value. It was therefore concluded, in contrast to previous authors, that the accident
record of drivers is quite stable over time, and that the very low correlations which have often been found were
due to the samples and methods used (low-risk drivers and short time periods equalling low crash means), and
not of any inherent instability in drivers behaviour and/or accident record. It was also concluded that only
culpable accidents should be used for this type of calculation. No evidence was found for a decrease in
correlation size between single years' accidents when time periods between the years were lengthened, i.e.
accidents in one year predicted accidents in several other years equally well. However, the period used was
rather short. The results are discussed with reference to training intervention for accident –involved drivers,
especially for organizations with major fleets such as bus companies.
Key words: accident proneness, bus driver, crash, reliability, stability, accident record

3
1. Introduction
1.1 History and theory of accident proneness
The beginning of safety research has often been traced to British studies about the
distributions of accidents in various populations (mainly industrial workers) in the early 20th
century (e.g. Greenwood & Woods, 1919; Newbold, 1927; Slocombe & Brakeman, 1930).
One of the main hypotheses was that accidents were not random events for workers, but in
some way were due to stable individual difference in their behaviour. This was the origin of
the research tradition of accident proneness.
The early work concluded that some people are clumsy, or risk seeking, and therefore cause
more damage to themselves and their surroundings than their more safety-minded peers, a
notion with quasi-theoretical properties. For example, one major prediction that could be
drawn from this general line of thinking was that people would tend to be stable over time in
their accident proneness, meaning that their numbers (or rates) of accidents would tend to be
the same in different time periods, at least within the same environment. This reasonable way
of thinking did also yield some positive results when empirical testing was undertaken (e.g.
Newbold, 1927; Wong & Hobbs, 1949; Adelstein, 1952), before many researchers turned to
statistical quarrelling (Mintz & Blum, 1949; Maritz, 1950; Mintz, 1954; 1956), and progress
was halted.
The accident proneness concept was imported into the fledgling traffic research field (Farmer
& Chambers, 1929; 1939), along with the basic controversy of whether it actually exists.
While statisticians were mainly interested in how various theoretical distributions could be
fitted against data, and what the results could possibly mean, many traffic researchers studied
the stability of accidents over time (which had generated the positive results for industrial
accidents). At first, it was claimed that a small number of drivers within a time period were
responsible for a fairly large number of accidents (e.g. Baker, 1929; 1932; Tillman &

4
Hobbes, 1949). However, it was soon found that there were actually very low correlations
between accidents in different time periods (e.g. Forbes, 1939; Kerr, 1957). Also, removing
the drivers with many accidents in one time period would actually have very little effect in
reducing crashes, because those with multiple accidents were few and far between. With
time, the traffic accident proneness concept fell into disrepute, with many critics (Kirchner,
1961; Cameron, 1975; McKenna, 1983) and hardly any defenders (for a review, see Porter,
1988). Most traffic researchers who studied the question empirically came to the conclusion
that there was very little stability over time (e.g. Harrington, 1972) as did accident
researchers in general (e.g. Arbous & Kerrich, 1951), due to the fact that most of the
correlations between time periods were extremely small. Today, there seem to be little
interest in the question of whether (traffic) accident record is a stable individual differences
variable (Gebers, 2003, is one of the exceptions).
In the present paper, some of the assumptions, methods and conclusions by previous
researchers regarding whether accident record is stable over time will be challenged as
erroneous, and alternative ways of investigating this question suggested and used. The
subjects covered are; how to analyse and interpret the sizes of accident correlations between
time periods, using all accidents versus culpable ones, and the effect of correlating accidents
in non-adjacent periods.
The notion of accident proneness used in the present study does not assume anything about
its relation to personality or other personal variables, but is simply the statement that people
are differently prone to causing accidents, in a similar environment, and that this is a trait that
is stable over time.
Methodological and statistical background

5
The rejection of the stability over time part of accident proneness did not take into
consideration a number of facts, which, if properly evaluated, point in quite another direction;
firstly, early studies on workers' accidents did often find sizeable (>.20) correlations between
time periods (e.g. Greenwood & Woods, 1919; Newbold, 1927; Farmer & Chambers, 1939);
secondly, a number of studies of professional drivers have reported fairly strong (>.40)
associations over time (e.g. Creswell & Froggatt, 1963; Bach, Bickel & Biehl, 1975;
Milosevic & Vucinic, 1975; Blasco, Prieto & Cornejo, 2003); thirdly, low correlations have
almost all been found in car driver populations (e.g. Forbes, 1939; Burg, 1970; Stewart &
Campbell, 1972; Peck, 1993); finally, culpability for accidents have seldom been included in
the analyses. The first three observations are all part of the same problem, and the solution
when it comes to reconciling them is that the weak associations have most often been
calculated on rather short time periods for low-risk groups, resulting in very low means and
standard deviations, which necessarily mean low statistical associations (Peck, 1993). The
very differing levels of correlations in different studies are therefore a natural result of
differences in mean number of accidents in the samples for the time periods used. In the
words of Arbous and Kerrich (1951): "In fairness to the theory of proneness, however, it must
be pointed out that as the successive exposure periods are increased, the correlation
coefficient will also increase..." (p. 369). Although this statement was about increasing the
variance within a sample by using longer time periods for calculations, the same principle is
valid for comparisons between samples; those with higher means/variances will yield
stronger associations between periods. So, if association data from several studies is gathered,
it can be shown that it is the time period used for calculation and/or the mean number of
accidents in the sample which almost exclusively (70-80 percent of the variance) determines
the size of the correlation between time periods (af Wåhlberg, forthcoming, see Figure 1).

Citations
More filters
Journal ArticleDOI
TL;DR: The authors found that self-reported crashes are negatively associated with a lie scale for driving, while recorded ones were not, as could be expected if the scale was valid and a self-report bias existed.
Abstract: The use of lie scales has a fairly long history in psychometrics, with the intention of identifying and correcting for socially desirable answers. This represents one type of common method variance (bias introduced when both predictors and predicted variables are gathered from the same source), which may lead to spurious associations in self-reports. Within traffic safety research, where self-report methods are used abundantly, it is uncommon to control for social desirability artifacts, or reporting associations between lie scales, crashes and driver behaviour scales. In the present study, it was shown that self-reports of traffic accidents were negatively associated with a lie scale for driving, while recorded ones were not, as could be expected if the scale was valid and a self-report bias existed. We conclude that whenever self-reported crashes are used as an outcome variable and predicted by other self-report measures, a lie scale should be included and used for correcting the associations. However, the only existing lie scale for traffic safety is not likely to catch all socially desirable responding, because traffic safety may not be desirable for all demographic groups. New lie scales should be developed specifically for driver behaviour questionnaires, to counter potential bias and artifactual results. Alternatively, the use of a single source of data should be discontinued.

90 citations

Journal ArticleDOI
TL;DR: The DBQ is probably the most widely used self-report questionnaire in driver behaviour research and shows that DBQ violations and errors correlate moderately with self-reported traffic accidents.
Abstract: This article synthesises the latest information on the relationship between the Driver Behaviour Questionnaire (DBQ) and accidents. We show by means of computer simulation that correlations with accidents are necessarily small because accidents are rare events. An updated meta-analysis on the zero-order correlations between the DBQ and self-reported accidents yielded an overall r of .13 (fixed-effect and random-effects models) for violations (57,480 participants; 67 samples) and .09 (fixed-effect and random-effects models) for errors (66,028 participants; 56 samples). An analysis of a previously published DBQ dataset (975 participants) showed that by aggregating across four measurement occasions, the correlation coefficient with self-reported accidents increased from .14 to .24 for violations and from .11 to .19 for errors. Our meta-analysis also showed that DBQ violations (r = .24; 6353 participants; 20 samples) but not DBQ errors (r = - .08; 1086 participants; 16 samples) correlated with recorded vehicle speed. Practitioner Summary: The DBQ is probably the most widely used self-report questionnaire in driver behaviour research. This study shows that DBQ violations and errors correlate moderately with self-reported traffic accidents.

58 citations

Journal ArticleDOI
TL;DR: The instrument can be used by bus companies for driver stress and fatigue management training to identify at-risk bus driver behaviour and reduce the tendency to engage in avoidance coping strategies, improve evaluative coping strategies and hazard monitoring when under stress may improve bus driver safety.
Abstract: There are likely to be individual differences in bus driver behaviour when adhering to strict schedules under time pressure. A reliable and valid assessment of these individual differences would be useful for bus companies keen to mitigate risk of crash involvement. This paper reports on three studies to develop and validate a self-report measure of bus driver behaviour. For study 1, two principal components analyses of a pilot questionnaire revealed six components describing bus driver behaviour and four bus driver coping components. In study 2, test-retest reliability of the components were tested in a sub-sample and found to be adequate. Further, the 10 components were used to predict bus crash involvement at three levels of culpability with consistently significant associations found for two components. For study 3, avoidance coping was consistently associated with celeration variables in a bus simulator, especially for a time-pressured drive. STATEMENT OF RELEVANCE: The instrument can be used by bus companies for driver stress and fatigue management training to identify at-risk bus driver behaviour. Training to reduce the tendency to engage in avoidance coping strategies, improve evaluative coping strategies and hazard monitoring when under stress may improve bus driver safety.

46 citations


Cites methods from "Bus driver accident record: the ret..."

  • ...Also, the Pearson correlation is very robust, showing no tendency to being affected by distributions that are more skewed than those of the variables used in the present study (af Wåhlberg and Dorn 2009)....

    [...]

Journal ArticleDOI
TL;DR: In this article, the reliability of self-reported driver mileage, violations, and crashes was evaluated using test-retest reliability, and the correlation between self reports of crashes in different time periods was found to be much larger than expected in one case, indicating a report bias.

39 citations


Cites background from "Bus driver accident record: the ret..."

  • ...For example, actual, recorded, traffic collision involvement has a certain between-time periods correlation, which increases strongly with the extension of the periods (af Wåhlberg & Dorn, 2009; see also the meta-analysis in af Wåhlberg, 2009)....

    [...]

Journal ArticleDOI
TL;DR: A meta-analysis of studies which had used measures of personality which could be converted into Big Five dimensions, and traffic accidents as the dependent variable, was undertaken by as mentioned in this paper, who found that tests of personality are weak predictors of traffic accident involvement, compared to other variables such as previous accidents.
Abstract: Problem The association between personality and traffic accident involvement has been extensively researched, but the literature is difficult to summarise, because different personality instruments and statistics have been used, and effect sizes differ strongly between studies. Method A meta-analysis of studies which had used measures of personality which could be converted into Big Five dimensions, and traffic accidents as the dependent variable, was undertaken. Analysis Outlier values were identified and removed. Also, analyses on effects of common method variance, type of instrument, dissemination bias and restriction of variance were undertaken. Results Outlier problems exist in these data, which prohibit any certainty in the conclusions. Each of the 5 personality dimensions were predictors of accident involvement, but the effects were small ( r Conclusions Tests of personality are weak predictors of traffic accident involvement, compared to other variables, such as previous accidents. Research into whether larger effects of personality can be found with methods other than self-reports is needed.

38 citations


Cites background from "Bus driver accident record: the ret..."

  • ...The effect size in accident prediction studies is mainly dependent upon the variation in the accident variable (af Wåhlberg, 2009; af Wåhlberg & Dorn, 2009; af Wåhlberg, Barraclough & Freeman, 2015; Barraclough, af Wåhlberg, Freeman, Watson & Watson, 2016)....

    [...]

  • ...Variance in the accident variable has been shown to strongly affect effect sizes in accident prediction studies (af Wåhlberg, 2009; af Wåhlberg & Dorn, 2009; af Wåhlberg, Barraclough & Freeman, 2015; Barraclough, af Wåhlberg, Freeman, Watson & Watson, 2016)....

    [...]

References
More filters
Journal ArticleDOI
TL;DR: The findings indicate that high-SELECTIVE TREATMENT programs can never be expected to result in a reduction in the overall accident rate, and greater effort should be invested in developing in-situ driver improvement techniques.

81 citations


"Bus driver accident record: the ret..." refers result in this paper

  • ...…accident data, there was really no alternative, as it was imperative that the present results could be compared to those of other studies on the stability of accident record, almost all of which have used Pearson correlations (e.g. Häkkinen 1958, Peck et al. 1971, Peck and Kuan 1983, Gebers 2003)....

    [...]

Journal ArticleDOI
TL;DR: Although both territory and prior driving record proved to have some validity in predicting a driver's accident risk, the accuracy of the prediction was low, and the relatively small unique predictive contribution of territory suggests that territory may be less important than previously believed.

74 citations

Journal ArticleDOI
TL;DR: Five cognitive ability tests were administered to a sample of 153 bus-driver trainees and the embedded figures test (EFT) and the dichotic listening test (DLT) showed no correlations with driver performance measures, thus failing to replicate earlier findings.
Abstract: Five cognitive ability tests were administered to a sample of 153 bus-driver trainees. The embedded figures test (EFT) of Witkin (1950) and the dichotic listening test (DLT) of Gopher and Kahneman (1971) were chosen on the basis of previously reported correlations with driving accident rate. The remainder were designed to cast light on what cognitive processes the EFT and DLT measure, and hence why they should relate to driving ability. The EFT correlated only marginally with success in driver training and with accident rate in a follow-up period of two years. There was no support for the hypothesis that this test measures a general ability to resist the influence of dominant stimuli. Instead a substantial correlation (0·64) was obtained with a typical ‘intelligence’ test. The DLT showed no correlations with driver performance measures, thus failing to replicate earlier findings. There was no support for the hypothesis that this test measures a general ability to switch from one task or mental set to anot...

67 citations

Journal Article
TL;DR: In this article, a number of studies have attempted to isolate driver accident risk correlates through multivariate and multiple regression techniques, but no single variable or set of variables can accurately predict the subsequent accident involvement rates of individual drivers.
Abstract: This paper reviews a number of studies which have attempted to isolate driver accident risk correlates through multivariate and multiple regression techniques. Multivariate approaches evaluate the interrelation among multiple risk factors and attempt to establish the relative predictive power of each variable within the context of a large set of variates. The review is limited to general (non-commercial) drivers. It is shown that no single variable, or set of variables, can accurately predict the subsequent accident involvement rates of individual drivers. However, a large number of driver characteristics affect the likelihood of a driver's accident involvement and these relationships can be used to make acturial predictions. Among the most consistent predictors of increased risk are: poor history of accidents and traffic convictions; being young; being male; being inexperienced; being from a lower socioeconomic status background; increased exposure; poor social adjustment; and certain attitudinal and personality traits. A hypothetical scheme is presented for integrating the preceding relationships into a comprehensive explanatory model.

62 citations


"Bus driver accident record: the ret..." refers background in this paper

  • ...…1963, Bach et al. 1975, Milosevic and Vucinic 1975, Blasco et al. 2003); third, low correlations have almost all been found in car driver populations (e.g. Forbes 1939, Burg 1970, Stewart and Campbell 1972, Peck 1993); finally, culpability for accidents has seldom been included in the analyses....

    [...]

  • ...It might be added that Peck (1993) presented calculations on the maximum attainable correlation between accidents and a ‘perfect predictor’, given the variance of the former....

    [...]

  • ...…the same problem and the solution when it comes to reconciling them is that the weak associations have most often been calculated on rather short time periods for low-risk groups, resulting in very low means and standard deviations, which necessarily mean low statistical associations (Peck 1993)....

    [...]

Journal ArticleDOI

59 citations


"Bus driver accident record: the ret..." refers result in this paper

  • ...This reasonable way of thinking did also yield some positive results when empirical testing was undertaken (e.g. Newbold 1927, Wong and Hobbs 1949, Adelstein 1952), before many researchers turned to statistical quarrelling (Mintz and Blum 1949, Maritz 1950, Mintz 1954, 1956) and progress was halted....

    [...]