You are on page 1of 17

Individual Differences in Accident Liability: A Review and

Integrative Approach

Rebecca Lawton and Dianne Parker, University of Manchester, Manchester, England

This paper reviews research since 1970 on the relationship between accident lia-
bility and individual differences, focusing specifically on accidents at work. The
history of research into accident liability and the methodological problems asso-
ciated with the research are considered. The review goes on to examine work on
the impact of personality factors, cognitive factors, and social factors on the like-
lihood of accident involvement at work. We suggest that research into individual
differences in accident liability should consider two possible routes to accident
involvement via errors and/or violations. Although errors are predominantly
associated with cognitive factors, violations have their origins in social psycho-
logical factors. We also consider the role of stress in mediating the personality-
accident association. It is contended that individuals differ in their reactions to
stress, so that although some respond by an increase in risk-taking behavior, the
effect on others is to increase the likelihood of suboptimal performance in terms
of information processing. Actual or potential applications of this research
include the development of a more sophisticated model of individual differences
in accident liability, which should be useful to organizations attempting to pro-
mote safety.

INTRODUCTION that the explanation of this susceptibility is to


be found in the personality of the individuals”
Historical Background to Accident (preface).
Liability Newbold (1927) and Farmer and Chambers
Two important changes in British industrial (1926) continued the work of Greenwood and
practices were held responsible for the dra- Woods, broadening the concept of unequal lia-
matic increase in accidents in the early part of bility for accidents. Newbold showed that
this century. The first was the increased pres- there was consistency in the individual’s ten-
sure of work and the speeding up of machines, dency to have an accident in various circum-
and the second was the drafting of younger stances (e.g., accidents at work and home),
and older men as well as women into the based on statistical evidence. In 1926, Farmer
workplace, while “able-bodied men” joined and Chambers took a further step, separating
the armed services. The pioneering statistical accident proneness as a special factor within
work of Greenwood and Woods (1919) was accident liability. However, the statistical meth-
carried out within this context. Their report to ods used were not sophisticated, and there was
the Industrial Fatigue Research Board stated no means to separate a tendency to report acci-
that the results “afford strong grounds for dents from a tendency to have accidents.
thinking that the bulk of the accidents occur Moreover, as Adelstein (1952) emphasized,
to a limited number of individuals who have a many of the early studies (Dunbar, 1943; Till-
special susceptibility to accidents and suggest man & Hobbs, 1949; Wong & Hobbs, 1949) had

Requests for reprints should be sent to Rebecca Lawton, Department of Psychology, University of Manchester,
Manchester, M13 9PL, England. HUMAN FACTORS, Vol. 40, No. 4, December 1998, pp. 655–671. Copyright © 1998,
Human Factors and Ergonomics Society. All rights reserved.
Downloaded from hfs.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016
656 December 1998 – Human Factors

serious methodological flaws. There were two More recent studies have also failed to pro-
main problems: First, differences in the level vide convincing evidence for the accident-
of exposure to risk of the so-called accident- proneness concept. Mayer, Jones, and Laughery
prone group were not considered. Second, (1987) examined the reports of more than
most of the studies were retrospective, and 7000 accidents among oil industry workers.
the researchers had previous knowledge about Although they did find that within each of five
which of their participants were accident job categories there was unequal accident lia-
prone. In spite of these flaws, the findings bility, the reliability of the finding is under-
could not be dismissed outright because they mined by the fact that they failed to control
were supported by a more rigorous prospec- sufficiently for exposure or experience. More-
tive study of pupil pilots (Biesheuvel & White, over, they conceded, “One cannot be content
1949) that showed differences in intelligence, to conclude that those workers who have expe-
skill, and personality between an accident rienced many accidents are accident prone…
group of 200 pilots and a nonaccident group attributing multiple accidents to accident
of 400 pilots. proneness is like saying that a sore throat is
Following this early research, the incorrect due to chronic pharyngitis” (p. 198).
assumption was made that if chance factors
alone were acting, all persons in the population Methodological Problems with Accident
would have the same number of accidents. As Liability Research
a relatively small proportion of individuals The majority of research published in the
appeared to be responsible for a relatively large area of accident liability is retrospective.
proportion of accidents, the notion of accident Significant differences between the accident-
proneness gained a life of its own, particularly free and accident-involved groups are taken to
as it appeared to excuse factory managers from provide evidence that the measured criterion
the responsibility for removing or controlling characteristic is causally linked to accidents,
hazards in the workplace. In theory, the “solu- which may not necessarily be the case. Studies
tion” to the problem was to remove those peo- employing a prospective design are much
ple who, because of some “flaw” in their more informative in establishing causality.
personality, were unable to cope with these However, the difficulty and expense of carry-
hazards. ing out longitudinal prospective studies means
Consequently, the focus of research was to that they are uncommon. Correlation coeffi-
isolate those factors that make an individual cients above .4 are unusual in both retrospec-
accident prone. In fact, it proved impossible to tive and prospective studies, leading some to
produce an overall stable profile of the accident- argue that the whole correlational approach is
prone individual or to determine whether unhelpful. Others point out that effects sizes
someone had an accident-prone personality. of this magnitude are by no means unusual in
Moreover, people seemed to become accident field studies and, given the large random com-
prone for limited periods, rather than continu- ponent and the unreliability of accident rates
ally throughout their lives. Of accident over time, should not be dismissed. McBride,
repeaters, Reason (1974) stated that “exami- Peck, and Coppin (1965, as cited in Shaw &
nation of accident repeaters over a lengthy Sichel, 1971b) argued, based on the estimated
period indicates that they are members of a reliability ceiling (r = .29) for accidents over a
club which is continuously changing its mem- three-year period, that it is unreasonable to
bership” (p. 187). In the same vein, Porter expect that any predictive battery or model
(1988) suggested that accident proneness is will ever account for more than 9% (0.292) of
not an enduring trait that attaches itself to the the variability in accidents over a similar period.
individual but, rather, is more likely to arise Therefore, if even a small effect can be repli-
from a combination of predispositions and cir- cated, serious consideration should be given
cumstances (e.g., an anxious individual who is to the findings. McKenna, Duncan, and
overworked and is operating a new machine Brown (1986) also argued that many different
for the first time). types of human failures can cause an accident,
Downloaded from hfs.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016
ACCIDENT LIABILITY 657

making it unlikely that any one measure will violations. A mediating role for stress is pro-
be able to predict very much of the variance posed. The review is not exhaustive but is a
reliably. Considering the practical implications brief summary of research carried out in an
of being able to predict some of the variability occupational context and published since
in the rates of potentially fatal accidents, the 1970. Those interested in more comprehensive
pursuit of causative factors appears worthwhile literature reviews are directed to Hansen (1988),
(Rosenthal, 1990). Elander, West, and French Porter (1988), Lester (1991), and Lawton and
(1993) argued that the relatively small amount Parker (1998). Finally, suggestions for future
of variance in traffic crash rates accounted for research directions and recommendations for
by the predictor variables in most studies – organizational action will be made.
commonly less than 10% – implies that tran-
sient factors must play an important role in RESEARCH EVIDENCE
accident causation, leading to an underes-
timation of the importance of stable character- Personality-Based Predictors of Accident
istics. Involvement
As Grayson and Maycock (1988) pointed Over the years, research on accident prone-
out, the combination of methodological prob- ness has indicated a plethora of different
lems and inconsistent findings outlined previ- personality characteristics that appear to be
ously meant that “it was almost inevitable that associated with accident repeaters. Glendon
the study of individual differences should go and McKenna (1995) argued that the search is
into a decline in the face of the emerging sys- almost certain to be fruitless as “being unable
tems approach” (p. 235). More recently, how- to produce any overall stable profile of the
ever, the system redesign strategy has begun to accident prone person, it is obviously not pos-
yield diminishing returns and, coupled with sible to use a reliable yardstick in establishing
progress in both cognitive psychology (e.g., whether someone has an accident prone per-
the development of error taxonomies; Ras- sonality.” Nevertheless, research in this area
mussen, 1982; Reason, 1987) and social psy- continues to proliferate.
chology (e.g., improved behavioral modeling; Personality is usually defined as those rela-
Ajzen, 1988; Fishbein & Ajzen, 1975), this tively stable and enduring features that distin-
has led to renewed interest in the study of the guish individuals from one another and which
individual differences underlying accident lia- form the basis for predicting people’s future
bility. Statistical methods have also developed behavior (Pervin, 1993). Both personality
which can cope with some of the problems traits and personality types have been investi-
that arise when dealing with accident data. gated in relation to accident involvement. In
Techniques for testing causal models, such as relation to traits, the results of studies using
path analysis and its extension, structural Cattell’s 16PF (Cattell, 1965) have been in-
equation modeling (see Hoyle, 1995), may, consistent. Whereas Quimby and Watts (1981)
through statistical control, allow researchers to found no significant relationship between road
move toward proposing causal paths in acci- traffic accident involvement and any of the 16PF
dent causation. factors, in a study of pilots Lardent (1991)
The remainder of this review will consider showed that the crash involvement could be
the range of human factors that have been significantly predicted by high scores on con-
studied in relation to accident liability, under scientiousness (beta = .50), trustfulness (beta =
three broad headings: personality characteris- .25), naïveté (beta = .27), self-sufficiency,
tics, cognitive factors, and attitudinal factors. (beta = .19) and low tension (beta = .18). In
In addition, the possible role of stress in acci- the discussion of these results, Lardent (1991)
dent liability is introduced at appropriate referred to them as “unexpected and rather
points throughout. The link will be made unique” (p. 18) and suggested that fighter
between these factors and the immediate pilots may constitute a special case. He also
behavioral precursors of accidents, errors, and suggested that the findings lend support to
Downloaded from hfs.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016
658 December 1998 – Human Factors

those researchers who are cautious about a internal locus of control (who feel in control
“generalized, unitary model of accident prone- of the outcomes of their behavior with respect
ness” (p. 18). In a study of young male con- to safety) have been shown to be less likely to
scripts, using logistic regression, Hilakivi et al. be accident involved than those, labeled exter-
(1989) also found that accident involvement nals, who feel that outcomes are beyond their
could be predicted by relatively high levels of personal control (Jones & Wuebker, 1985;
naïveté and trustfulness. However, in a review Wuebker, Jones, & Dubois, 1985). Using a
of the literature on personality characteristics general measure of locus of control, Clement
and accidents that involved a conceptual fac- and Jonah (1984) showed a significant rela-
tor analysis, Booysen and Erasmus (1989) tionship between external locus of control and
proposed that the two factors most relevant to number of accidents for men only (r = .17).
accident involvement were recklessness and Mayer and Treat (1977) have shown that
anxiety-depression. externals are more accident involved than
The relationship between Eysenck’s person- internals. Thus, individuals who attribute out-
ality types (Eysenck, 1947, 1970; Eysenck & comes to factors beyond their personal control
Eysenck, 1968) and road traffic accidents has were more likely to report being accident
been reviewed by Lester (1991), who conclud- involved. A study of chemical workers by
ed that neither introversion-extroversion nor Denning (1983) using cluster analysis discov-
neuroticism (which measures personality ered that one group of workers with high acci-
along a stability-instability dimension) was sig- dent and injury rates was characterized by an
nificantly predictive of accidents across the external locus of control. However, two crite-
group of nine studies considered. Several rion group studies (Janzen, 1983; Sims,
other studies have suggested a link between Graves, & Simpson, 1984), which looked at
extroversion and accidents, though much of sawmill workers and mine workers, respec-
this research concerns road traffic accidents tively, failed to replicate this finding. The evi-
only (Fernandez, 1978; Mozdzierz, Macchi- dence, again, is mixed. It must also be borne
telli, Planek, & Lottmann, 1975; Schenk & in mind that it is plausible that having an acci-
Rausche, 1979). To further confuse matters, dent has an effect on one’s perceptions of
using a sample of Indian bus drivers, Pestonjee locus of control, rather than the other way
and Singh (1980) found no-accident partici- round. Moreover, Foreman, Ellis, and Beavan
pants to be more extroverted than those who (1983) have shown that the difference be-
had been involved in an accident. Moreover, a tween externals and internals held for partici-
recent study of offshore oil workers (Suther- pants whose self-reports of their accidents
land & Cooper, 1991) found that although implied that they were culpable but not for
there was no significant difference between those whose self-reports implied that they
the two groups in terms of extroversion, with were blameless. Two explanations are possi-
respect to neuroticism, 37% of the high-N ble. It may be that internals are more likely
scorers reported accidents, compared with than externals to give accounts of accidents
24% of the low-N scorers (χ2 = 3.69, p = .05). that show themselves to be culpable, or it may
Caution when interpreting the results of these be that the experience of having an accident
studies is also necessary, given the finding of for which one is culpable alters one’s percep-
Powell, Hale, Martin, and Simon (1971) that tions of locus of control. Either explanation
the association between extroversion and acci- undermines confidence in the results of stud-
dents may be accounted for, at least in part, by ies linking locus of control with retrospective
the greater inclination of extroverts to report reports of accidents.
accidents. The link between accident involvement and
Rotter’s (1966) concept of locus of control, the propensity to take risks has been investi-
adapted by Jones (1984) to form the Safety gated in several settings. Beirness and Simpson
Locus of Control Scale, has been employed (1988), in a prospective study of personality
several times as a possible indicator of acci- and accident involvement, used analysis of
dent involvement at work. Those with an variance to reveal a significant main effect for
Downloaded from hfs.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016
ACCIDENT LIABILITY 659

accident involvement on two subscales of crew error, distinguished between unstable


Zuckerman’s (1979) Sensation-Seeking Scale. introverts and unstable extroverts. He report-
The results indicated that those people in the ed that during an emergency, the former suffer
accident-involved group scored higher on the over arousal, whereas the latter are character-
subscales “thrill and adventure seeking” and ized by deliberate risk taking and sensation
“experience seeking.” Meadows (1994) repli- seeking. We would not want to argue that
cated this finding, showing the same relation- there is an inevitable association between per-
ship between scores on a scale of thrill seeking sonality type and accident proneness, but it is
and accident involvement over a three-year arguable that personality characteristics such
period among 1500 drivers (r = .24). as sensation seeking or rigidity of thought can
Considered overall, the findings in this area have an impact on behavior that may make
might indicate that personality is not systemat- accident involvement more likely. Beirness
ically associated with accident likelihood, that (1993), in a review of the literature, conclud-
personality is not well measured, or even that ed that although there is ample evidence that
different personality factors are associated personality factors can have an influence on
with accidents in particular industries and the manner in which people drive and the
jobs. Furthermore, studies of personality and incidence of accidents, the contribution of
accidents have tended to be retrospective, personality factors to crash involvement is
based on self-reports of accidents, and many likely to be indirect, mediated by other factors
have failed to control for factors such as expo- and behaviors.
sure to risk. These methodological flaws make This idea is also supported by Farmer
it difficult to rely on the findings. (1984), who examined the relevance of per-
The association between personality and sonality differences to the flying task. He pro-
accidents is not straightforward, and the vided evidence that whereas extrovert pilots
research has produced a confusing array of are better able to cope with the demands of
results. However, further analysis – particularly difficult phases of the flight, introverts, who
of the more detailed studies in this area – exhibit superior performance on vigilance tasks,
reveals a pattern that suggests that individuals may show superior monitoring of instruments
with characteristics toward the extreme ends during undemanding phases. Farmer conclud-
of two personality trait continuua have an ele- ed that “it is probable that no unique accident
vated risk of accident involvement. Shaw and prone personality exists, and that progress will
Sichel’s (1971a) data suggest that two person- be made only by attempting to relate personal
ality types are most likely to have road traffic characteristics to specific types of human
accidents: sociopathic extroverts (self-centered, error.” (p. 177).
over confident, aggressive, irresponsible, It is widely accepted that 80% to 90% of
resentful, intolerant, impulsive, antisocial, accidents are the result of human error (Hale
antagonistic to authority) and anxious-neurotics & Glendon, 1987; Heinrich, Petersen, & Roos,
(tension ridden, unduly sensitive to criticism, 1980). Recently it has been suggested (Parker,
indecisive, unable to concentrate, easily fatigued, Reason, Manstead, & Stradling, 1995; Reason,
depressed, emotionally labile, easily intimidat- Parker, & Free, 1994) that a useful distinction
ed, and with feelings of inadequacy). can be made between errors and violations.
Hansen (1989), in a study of 362 chemical On the one hand, errors may be defined as
industry workers, demonstrated a link be- planned actions that fail to achieve the desired
tween both general social maladjustment (r = goal. The actions can either be unintended
.28) and distractibility (r = .31) and accidents. (slips and lapses) or intended (mistakes), but
By controlling for exposure to risk and consid- in both cases the poor outcome is unintended.
ering a number of other variables, Hansen’s Although slips and lapses are both failures in
study provided a powerful test of the relation- execution, they differ with respect to observ-
ship between personality and accidents and, as ability. Slips, which are actions not as planned,
such, the findings are noteworthy. Similarly, are potentially observable, whereas lapses,
Chappelow (1989), in an investigation of air- which involve memory, are often apparent
Downloaded from hfs.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016
660 December 1998 – Human Factors

only to the person who experiences them performance on a dual task and scores on an
(Reason, 1990). accident-proneness questionnaire among a
Violations, on the other hand, are defined small student sample. Lester (1991), reviewing
as deliberate deviations from the recognized studies of individual differences in accident lia-
safe method of working, which might be in bility on the road, concluded that there was no
the form of rules, procedures, or group norms. evidence of a significant relationship between
This distinction, which will be discussed in reaction time and accident rate.
detail later, informed a paper by Hofmann, Since the 1950s, a period during which the
Jacobs, and Landy (1995) reviewing the liter- ideas of cognitive psychology became domi-
ature pertaining to individual, micro-, and nant, there has been more interest in the rela-
macro-organizational influences on safety per- tionship between accidents and higher-order
formance in high-reliability process industries. cognitive abilities such as attention and percep-
There, the authors emphasized that a mean- tual style. The bulk of this work has involved
ingful consideration of safety at the individual road traffic accidents, where it has been pro-
level of analysis requires that one takes into posed that attention and perception are of par-
account both the cognitive factors implicated ticular importance when traveling at speed.
by human errors and the attitudinal and moti- Initial evidence of a relationship between field
vational factors that effect safe behavior. dependence and accidents (Harano, 1970; Loo,
In the following two sections we consider 1978) did not find support in more statistically
these factors, reviewing the research evidence sophisticated studies (Harano, Peck, &
from cognitive and social psychology. First, we McBride; 1975; McKenna et al., 1986). There
examine the studies linking cognitive perfor- is some evidence from studies of attention-
mance with accidents and posit errors as the switching ability (Gopher & Kahneman, 1971;
mediating factor. Second, we consider social Kahneman, Ben-Ishai, & Lotan, 1973; Mihal &
psychological studies associating attitudes and Barrett, 1976) of a negative correlation be-
accidents and suggest that the mediating fac- tween scores on a dichotic listening task and
tor here may be violations. accident rate. Indeed, in a meta-analytic study
of a range of predictors of driving accidents,
Cognitive Predictors of Accident Arthur, Barrett, and Alexander (1991) found
Involvement auditory selective attention to be the most use-
Tests designed to measure cognitive pro- ful predictor (mean r = .26). Others have
cesses, including thinking, reasoning, and per- argued that the focus should be on visual,
ceiving, have been employed by accident rather than auditory, attention in relation to
researchers to test the hypothesis that there is driving, though Evans’s (1991) review of the
a relationship between a particular cognitive literature concluded that there is no evidence of
skill and accident likelihood. Early in the 20th a systematic relationship between visual perfor-
century, when psychology was dominated by mance and driving ability. Avolio, Kroeck, and
the behaviorist school of thinking, the focus of Panek (1985), using a sample of 72 drivers,
research was stimulus and response and there found differences between the nonaccident and
was much study of a range of low-level cogni- the accident group on a range of visual and
tive skills (see Hale & Hale, 1972, for a re- auditory attention tasks. Arthur, Strong, and
view). The few studies that have been carried Williamson (1994) reported an overall correla-
out in this area since the 1970s have, for the tion of .31 between errors on visual attention
most part, provided equivocal evidence on the test scores and accidents. Porter’s (1988)
impact of such skills in accidents. For example, review concluded that poor visual attention and
whereas Kishore and Jha (1978) reported a the experience of recent major life events
relationship between errors on a two-handed appear to be the most consistent predictors of
coordination task and slower reaction times accident liability. Broadbent, Broadbent, and
and accidents among bus drivers but not in- Jones (1986) provided evidence that attentional
dustrial workers, Porter and Corlett (1989) re- facility is multifaceted, and it could be argued
ported strong correlations (.38 to .65) between that some of the studies mentioned earlier
Downloaded from hfs.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016
ACCIDENT LIABILITY 661

suffer from a rather simplistic conceptualiza- It has been suggested that one way in
tion of attention. which stress may be associated with accidents
A range of studies investigating the rela- is through an elevation in the rates of error.
tionships between various cognitive abilities However, Reason (1990) argued that stress
and accident involvement (the majority of does not cause errors but, rather, that certain
which have considered only road traffic acci- cognitive styles result both in higher rates of
dents) have produced conflicting evidence. absentmindedness and in coping strategies
Moreover, many of the published studies can that are not appropriate for dealing with
be criticized for lacking ecological validity. stressful situations. Broadbent et al. (1986)
McKenna and Crick (1991) attempted to provided evidence suggesting that high levels
avoid this criticism by using a driving simula- of self-reported cognitive failure may indicate
tor as a means of assessing hazard perception vulnerability to subsequent external stresses.
and found that expert drivers showed better Trainee nurses with higher scores on the
hazard perception skills. However, there is no Cognitive Failures Questionnaire (CFQ) showed
direct evidence to suggest that the link be- more symptoms of stress than those with low
tween hazard perception and accident involve- CFQ scores when working on stressful wards
ment reflects cognitive ability (but see Quimby but not when working on less-stressful wards.
& Watts, 1981). It could equally well be ar- Questionnaire studies of absentminded errors
gued that some individuals are skilled as per- do not consider directly the relationship
ceivers but fail to interpret what they perceive between cognitive failures and accidents, but
as hazardous. they do throw some light on the subject. Posit-
In the safety literature it is now widely ing a mediating role for stress may help to
accepted that human error contributes to explain the inconsistent findings of the labora-
many industrial accidents (Reason, 1990; tory studies, many of which did not take stress
Wagenaar, Hudson, & Reason, 1990). Con- into account. Hockey, Clough, and Maule
sideration of human error allows us to move (1996), discussing the effects of emotional
beyond the confines of cognitive performance state on decision making and risk behavior,
in general (i.e., what an individual is capable suggested that stress may directly affect per-
of doing) to a consideration of the links formance by encouraging the use of shortcuts
between various error types and accidents. in cognitive processing, thus reducing mental
Although there do not appear to be any pros- effort. Shortcuts such as attentional narrow-
pective studies in the industrial context evalu- ing, reduced cognitive complexity in problem
ating the predictive power of tendency to slips solving, and a shift toward speed at the ex-
and lapses, in a retrospective study of 178 pense of accuracy may work by reducing the
fatalities and 99 serious accidents in Finland demand on stretched information-processing
(Salminen & Tallberg, 1996), 84% to 94% resources. It is not inconceivable that such
were attributable primarily to human error. changes may lead to an increase in errors.
The majority of these were errors at the skill- Clearly the effects of stress on cognitive per-
based level of processing – that is, errors of formance and error, and thus on the possibili-
execution. Slips and lapses (Norman, 1981; ty of an accident, require further research. In
Reason, 1990) occur most frequently when particular, it would be useful to clarify
the task requires little conscious effort and whether it is subjective perceptions of stress
can be said to occur automatically. Such errors or reactions to the experience of stress that
are thought to be the result of mistimed or affect errors and accidents.
omitted attentional checks on the progress of an
action sequence. It is well established (Broad- Attitudinal and Motivational Predictors of
bent, Cooper, Fitzgerald, & Parkes, 1982) that Accident Involvement
some people are indeed prone to absentminded Recent years have seen a growing emphasis
errors and are also likely to report relatively within human factors research on studies that
high numbers of memory lapses and instances go beyond human error in the form of slips
of inattention. and lapses and that embrace social-psychological
Downloaded from hfs.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016
662 December 1998 – Human Factors

issues such as safety attitudes, culture, and cli- ceptions and attitudes about safety of more
mate, all of which are related to the commis- than 5000 employees at a nuclear reprocessing
sion of violations at work. For example, a study plant, focusing on the relationship between
of three separate plants with high, medium, scores on three “risk factors” and individual
and low accident records within a large indus- characteristics such as accident histories, age,
trial facility (Olearnik & Canter, 1988) revealed gender, job position, and type of job within
significant correlations between accident fig- the organization. The most important finding
ures and questionnaire items reflecting a range concerning accident liability was that a “posi-
of safety-related attitudes. The attitudes mea- tive” attitude toward risk was a significant
sured included commitment to the safety of predictor of the likelihood of being involved in
coworkers, accident inquiries, rules about safe accidents.
working, recommended precautions, under- Two large-scale studies looking at the rela-
standing safety, the job itself, and plant supervi- tionship between farm accidents and safety
sors. The attitudes of employees toward safety attitudes have produced contradictory find-
were shown to be less positive in the high- ings. The first, a study of 493 Pennsylvanian
accident plant than the low-accident plant. farmers (Murphy, 1981), found that partici-
A number of studies have been carried out pants indicated the same attitudes toward
by Donald and colleagues at the Safety Re- farm safety concepts regardless of their acci-
search Unit at the University of Surrey, look- dent involvement. The second study, involving
ing at the association between attitude and 683 Canadian farmers (Harrell, 1995), used a
accident rates (e.g., Canter & Donald, 1990; safety attitude questionnaire and found that
Donald & Young, 1996). Donald and Canter those with a high score on personal risk taking
(1994) described a measure of safety attitudes and fatalism were more likely to have had an
that they showed to be positively and signifi- accident than were low scorers. Furthermore,
cantly correlated with accident rates in the safe farming practices, such as safe equipment
chemical industry. Of the 10 subscales of the operation and the use of protective clothing,
Safety Attitude Scale, 9 correlated significant- also differentiated between those who were
ly with self-reported accident figures (r = –.45 more and those who were less likely to be
to –.83). Donald and Canter concluded that involved in an accident. Incompatible findings
“rather than dealing with unintended factors, such as these within similar populations are
i.e. the slips and errors, what is necessary is to perplexing to the safety practitioner. However,
create a climate where the dangerous actions such findings are not surprising, given the
of which people are aware are not carried out” unreliable nature of accidents.
(p. 207). Hurst, Young, Donald, Gibson, and Job satisfaction is another important aspect
Muyselaar (1996) used the Safety Attitude of attitude to work which has been opera-
Scale together with a Process Safety Manage- tionalized in safety research studies. Allodi
ment System audit tool, believing that these and Montgomery (1979) reported that although
two measures cover the two parts of safety job dissatisfaction was not predictive of acci-
culture. Attitudes and beliefs were thought to dent liability, it was related to the probability
reflect the underlying components of safety of excessive time off and of developing a neu-
culture, whereas safety management systems rotic complication after an accident.
were thought to represent the tangible compo- Job satisfaction is likely to be low when
nents. Although Hurst et al. did not give the employees experience unwanted levels of
values for correlations between safety attitudes stress in the workplace. The role of work stress,
and accidents, they claimed that the correla- subjective and/or objective, has been exam-
tions between safety attitudes and accidents at ined in some studies, the suggestion being that
six major hazard sites in their study were stress may play a role both in the tendency to
lower than those reported by Donald and report accidents and in the time lost as a
Canter (1994). result of the accident. Hartley and El Hassani
Lee, Macdonald, and Coote (1993) carried (1994) examined the relationship between
out a large-scale survey assessing the risk per- stress and accidents in a sample of 335 truck
Downloaded from hfs.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016
ACCIDENT LIABILITY 663

drivers and 420 car drivers. Based on the responses to a safety survey does not provide
assumption that drivers are aware of and can strong evidence for the person-environment fit
report the behaviors, cognitions, and affect theory. It does, however, point to the impor-
associated with driving stress, a questionnaire tance of attitudes toward work and suggests
was employed to measure stress. Regression particular areas worthy of investigation, such
analysis showed that age, gender, and stress as the importance of peer and team support
(both job related and general) accounted for a for safety practices. The measure of person-
total of 35% of the variance in accident or environment fit in this study has not been vali-
violation reports. The significant factors were dated elsewhere, and the study did not
“driving imperturbability,” “perceived stress- actually include a measure of stress, a factor
fulness of driving,” “life stresses,” and “fre- that Sherry (1991) suggested is likely to medi-
quency of driving.” These results must be ate the relation between P-E fit and accidents.
treated with caution, however, given that acci- Future research in this area will need to clarify
dents and convictions together form the what is being measured by P-E fit, as differ-
dependent variable and therefore it is not pos- ences in attitudes between supervisors and
sible to distinguish accident-causing from employees is a rather crude measure that may
violation-causing factors. Moreover, the report- simply reflect employee deviance. Finally, the
ing of the regression analysis did not allow for greater usefulness of the attitudinal question-
the relative impact of stress variables and naire in predicting accidents in the precomple-
demographic/exposure variables. tion injury group suggests that having an
Sherry (1991) proposed that workplace acci- accident may have a strong influence on how
dents are caused by poor person-environment participants responded to the questionnaire
fit, which gives rise to the experience of stress items.
and thus to an increased accident risk. In Melamed, Luz, Najenson, Jucha, and Green
addition to person-environment fit (linked (1989) investigated the role of stress more
with accidents for the first time), Sherry mea- directly in terms of both accident occurrence
sured employee attitudes toward the work and sick leave in a study of 729 male workers
environment. Both supervisory and nonsuper- employed in five factories in Israel. They used
visory staff in a large American rail transpor- an objective measure of stress, the ergonomic
tation company took part in a survey. A new stress level (E-S-L), which includes ratings of
instrument was developed for the purpose of body motion and posture, physical effort,
the study with 150 attitudinal items represent- active hazards, and environmental stressors.
ing such areas as participation, credibility, peer Environmental annoyance was also measured
support, and safety. Person-environment (P-E) via self-reports so that the effect of individual
fit was calculated as the difference between sensitivity to stressors could be taken into
the employee and his or her supervisor on account. Finally, stress reactions in the form of
23 scales contained in the safety survey. work heart rate, job dissatisfaction, and somatic
Significant differences were found between complaints were examined. The study was ret-
the accident and nonaccident groups in terms rospective and used the organization’s acci-
of P-E fit, such that nonaccident-involved dent statistics over a one-year period as the
employees had better P-E fit. More detailed dependent variable. The results showed a pos-
analysis revealed that the differences were in itive linear association between accident rate
relation to supervisor support, credibility, and E-S-L. Those individuals who showed
recognition, and peer safety practices. In a greater sensitivity to environmental stressors
second study, which used only responses also had higher accident incidence, and this
to the safety questionnaire, Sherry was able to effect increased with rising E-S-L.
correctly classify 68.8% of the postcompletion Melamed et al. (1989) suggested that “pre-
injury group and 95.5% of the precompletion occupation with disturbing job and work envi-
injury group. ronment characteristics . . . is an important
That the accident and nonaccident groups contributor to accident involvement. This pre-
could be differentiated simply on the basis of occupation may serve as a distracting factor
Downloaded from hfs.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016
664 December 1998 – Human Factors

making the worker less attentive to danger were analyzed using structural equation mod-
cues” (p. 1108). They went on to say that the eling, a technique rarely used in the field of
E-S-L is a reliable predictor of work accidents accident research. The model provided a good
and should be developed for systematic use in fit to the data, with significant links between
industry. It could be argued, however, that the accident consistency and scores on the general
E-S-L measure is confounded with exposure social maladjustment scale and the dis-
to risk, given that one of the four components tractibility scale. Hansen controlled for the
of this aggregate measure is active hazards. effects of exposure to accidents and of involve-
Indeed the results suggest that where exposure ment in counseling. Among more experienced
to risk is high, those individuals who are more workers, risk exposure was shown to mediate
sensitive to their environment are more likely the effect of experience on accidents. Thus,
to have been involved in an accident. the experienced workers had more accidents,
but this was because they were exposed to
SYNTHESIZING THE EVIDENCE greater risk. Furthermore, one of the tests of
cognitive ability (Bennett; Aiken, 1979) corre-
As described, the vast majority of studies lated with accident risk (.31) but not acci-
concerning individual differences in accident dents (.10), suggesting that employees may
liability have focused on one or more predic- have been chosen for risky jobs on the basis of
tor variables drawn from a fairly narrow per- the types of mechanical reasoning skills mea-
spective. However, some studies have attempted sured by the Bennett test.
a partial synthesis of the disparate literature Sutherland and Cooper (1991), in their
and have included a consideration of the role study of offshore workers, concluded that cer-
of stress. For example, Steffy, Jones, Murphy, tain personality characteristics/behavior pat-
and Kunz (1986) proposed a model of the terns seem to mediate the response to stress
relationship between stress and accidents that and may be predictors of increased vulnerabil-
takes into account evidence from both cogni- ity to accident involvement. They were cau-
tive and social psychology. They suggested that tious about the findings, given that they were
stressors, either work- or nonwork-related, based on self-report data, and suggested that
cause acute reactions, which can be psycho- the relationship between these variables may
logical, physiological, and/or behavioral (e.g., in fact be circular, not linear. In fact, Suther-
anxiety, fatigue, and alcohol use, respectively). land (1993) reported that both operator and
These reactions to stressors have the effect contractor personnel who had been involved
of decreasing intellectual and performance in an accident perceived the environment to
capacities, such as reaction times and judg- be more stressful than did accident-free per-
ment. Decreased capacities serve to increase sonnel, suggesting that the author regarded
the probability of errors and can, as a result, the relationship between occupational stress
lead to accidents. The occurrence of the acci- and accident involvement to be bidirectional.
dent can function as a stressor that reactivates Evidence was then provided to suggest that
the cycle. Steffy et al. used three case studies personality style mediates the relationship
in different industries to demonstrate that between stress and accidents, with Type A
stress assessment and intervention programs personalities perceiving more stress and being
can significantly reduce insurance-related losses. involved in a greater percentage of accidents
Hansen (1989) also took a multiperspective at work.
approach, developing and validating a rela- A program of research on driver behavior
tively sophisticated causal model of the acci- at Manchester University in England has also
dent process. The variables measured included encompassed the approaches of cognitive and
cognitive ability, general social maladjustment, social psychology. In several large-scale studies
distractibility, involvement in counseling, job a clear distinction has been established
experience, and age. The participants were between driving errors and driving violations,
362 production and maintenance workers at a the latter being at least partly intentional (e.g.,
large chemical processing company. The data Reason, Manstead, Stradling, Baxter, &
Downloaded from hfs.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016
ACCIDENT LIABILITY 665

Campbell, 1990). Parker and colleagues have managers who, while acknowledging the
also shown, both prospectively and retrospec- importance of safety, have production as their
tively, a positive relationship between road primary objective.
traffic violations and accident involvement,
but they have shown no such relationship for A Multiperspective Approach to
Individual Differences in Accident Liability
errors (Parker, Reason, et al., 1995; Parker,
West, Stradling, & Manstead, 1995). These Given the inherent unreliability of acci-
findings suggest a multifaceted approach to dents as a dependent measure, it may be more
accident prevention. Skills training may be profitable to focus on the psychological pre-
successful in reducing errors, but it will not cursors of errors and violations, many of which
have a significant effect on the commission of do not in fact lead to an accident. As Taylor
violations. The reduction of violations can be (1981) argued, taking accidents as the focus
tackled by persuading drivers not to drive in a of psychological investigation is probably not
dangerous manner. In psychological terms this the most useful approach, because no individ-
means that safety research should move away ual ever intends to have an accident. However,
from an exclusive focus on cognitive psychology, if the precursors of errors and violations can
which emphasizes performance factors, and be identified, then it may be possible to design
should include the investigation of attitudinal interventions that reduce both their occur-
and motivational precursors of accidents, rence and the consequent elevated likelihood
which are the province of social psychology. of an accident.
There is support for the importance of vio- A sophisticated explanation of the role of
lations in an industrial context in Free’s (1994) human factors in accident causation demands
analysis of 21 British Rail accident reports. a multiperspective approach. On the basis of
Almost every one was found to involve an the research evidence reviewed here, we
error in conjunction with a violation, though would suggest that there are two main routes
to accident causation. The first is the cognitive
the two were not necessarily attributable to
route, which involves failures in information
the same individual. Violations appear to have
processing or skill, resulting in errors. This is
the effect of taking people into an area of
the traditional realm of the psychologist in
greater risk, thus making the situation less for-
safety research. The second, social-psychological,
giving of subsequent errors made by the viola-
route involves a consideration of attitudinal
tor or anyone else in the vicinity. Ranney
and behavioral factors. These two routes to
(1994) echoed this conjecture in a review of
accident involvement are by no means mutual-
models of driving behavior, suggesting that “if
ly exclusive. In fact, it seems that accidents
drivers adopt inappropriate safety margins,
commonly involve errors and violations in
lapses of attention may leave insufficient room
combination. We would also suggest that
for recovery. Identifying the situational deter-
stress has a critical role to play in relation to
minants of safety margins is also a potential both routes, linking personality factors with
focus for roadway safety research” (p. 747). accidents. Thus personality factors may pre-
Although the immediate cause of a particu- dispose individuals to perceive and cope with
lar accident may appear to be an error in the stressors in ways likely to lead to particular
form of a slip or a lapse, the necessary factor types of unsafe acts. On the basis of the
in the accident sequence may be a violation of research evidence reviewed here, a series of
a rule or a norm. Battman and Klumb (1993) tentative hypotheses may be formulated:
discussed the importance of rule violations for
organizational safety. They claim that viola- 1. The occurrence of errors and violations, the
tions in an industrial context arise from a con- immediate precursors of accidents, increases
flict between organizational goals and under stressful conditions.
individual goals – for example, when work 2. Individuals at the extremes of the instability
pressures give rise to employees taking short- scale, either extroverts or introverts, have high-
cuts. Such actions may be disregarded by er rates of accident involvement than others,
Downloaded from hfs.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016
666 December 1998 – Human Factors

particularly in response to stress. Unstable of stress as a mediating factor. The tentative


extroverts may be more likely than normals to hypotheses outlined in the previous section
respond to stress by showing risk-taking behav- would provide a good starting point for such
ior (i.e., committing violations), whereas unsta- research. In particular, the roles of subjective
ble introverts exposed to stress may show a and objective stress and their impact on rates of
tendency to make more errors. This is not to errors and violations should be considered in
suggest that unstable extroverts never suffer future research.
from the performance decrements that give rise
to errors or that unstable introverts never have 2. Studies should always control for exposure to
the sorts of undesirable attitudes and motiva- risk of accident. In the area of driving this
tions that promote the commission of viola- means asking respondents to record their aver-
tions. age annual mileage and the types of roads on
which they drive most. For industrial research,
3. Stress mediates the relationship between per- factors such as task requirements, risk assess-
sonality and accidents in two different ways. ments, and hours worked should be taken into
First, a given situation may be differentially per- account. Age and experience are also variables
ceived as stressful according to personality likely to have an effect, direct or indirect, on
type. Second, different personality types may accident likelihood and therefore should be
respond to experienced stress in different ways. considered.

3. A number of researchers have suggested a dis-


The suggestion that different personality
tinction between a temporary and a permanent
factors may be associated with different types state of accident proneness (when considered as
of accidents and that personality characteris- a statistical fact). The use of longitudinal studies
tics are likely to be mediated in their effect by that collect data over a number of years would
performance and behavioral factors may serve allow this distinction to be validated. In addition
it would be useful to investigate whether certain
to account, at least in part, for the confusing
factors such as stress or life events might predict
array of results in this area of research. In temporary states of increased accident liability
fact, Sutherland and Cooper (1991) went so and whether other, more enduring characteris-
far as to suggest that the jobs people seek out tics of the individual show an association with
may reflect their personality characteristics. the more permanent state.
They found a relatively high proportion of 4. Most studies of accident liability consider all
extroverts on drilling rigs, and Farmer (1984) sorts of accidents. Minor and major accidents
found that pilots were more than normally are regarded as essentially the same, as are acci-
aggressive and dominant. The risks associated dents for which the individual was responsible
with particular jobs, and the safety character- and those where the individual had a passive
role. The development of a good taxonomy of
istics important for them, vary. In a setting accidents is crucial to this field of research.
where vigilance is required, errors may consti- West (1993) has begun to develop such a tax-
tute the primary risk to system breakdown. In onomy with respect to driving accidents, mak-
another setting, where the system is well ing the distinction between passive and active
defended against error, the primary risk may accidents and then going on to describe various
common accident types (e.g., passive right-of-
come from people adopting unofficial work way violation, active loss of control). Using this
practices in violation of safety procedures. taxonomy, Parker, West, et al. (1995) found
self-reported violations to be a better predictor
Directions for Future Research of certain accident types, particularly those
A number of directions for future research, where the accident-involved driver has an
active role. Studies of this kind necessitate the
concerning both areas of study and methods collection of large amounts of data, a factor
of study, can be suggested: that may account for their scarcity. However,
future research might, at the very least, consid-
1. The primary objective of future research in the er the active-versus-passive distinction.
field of accident liability should be to overcome
the limitations of previous studies and to devel- 5. Accidents are rare and have multiple causes. As
op a model of the accident causation process, a result, chance is the scourge of accident
rather than to examine specific individual char- research. The same unsafe behavior may in one
acteristics in isolation. Models of the accident instance go unpunished yet in another result in
causation process should also consider the role a fatal accident. All sorts of external factors
Downloaded from hfs.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016
ACCIDENT LIABILITY 667

may influence the outcome (weather, cowork- may include organizational factors such as
ers, noise, mechanical failure of a tool, etc.), poor training, incorrect procedures, and work-
making the prediction of large amounts of vari-
place stressors, the focus of research moves
ance in accident likelihood difficult, if not
impossible. Future research, having demon- away from the individual in isolation toward
strated a relationship between an unsafe behav- the individual in the organizational context.
ior and accidents, should then focus on the Taking the multiperspective approach out-
investigation of factors that predict the unsafe lined previously enables us to make a number
behavior. This change of focus has already hap-
pened to some extent in relation to driving acci-
of points that are important in terms of organi-
dents. It is well established that driving above zational safety. First, it should be stressed that
the posted speed limit is predictive of road traf- although personality characteristics may make
fic accidents in the long run. However, any some people more accident liable in some cir-
attempt to demonstrate a direct link between cumstances, the organization can do much in
speeding as measured in a single study and the
occurrence of accidents within that study is mitigation. The research evidence available sug-
unlikely to meet with success. Most speeding gests that selection on the basis of future acci-
goes unpunished by negative consequences. dent liability is not a useful approach, given
However, that does not mean that speeding is that the causal relationships involved are com-
not important in accident causation. Therefore, plex. Glendon (1979) reported that although
much research is now dedicated to determining
the attitudinal and motivational characteristics managers recognize their role in improving
that are associated with this dangerous driving safety, they are reluctant to accept their role in
behavior. accident prevention, considering this to be the
responsibility of the workers. It is important to
6. Traditionally, researchers into accident liability
remember that exposure to a hazardous situa-
have focused almost exclusively on those fac-
tors that predict inclusion in the accident tion is a necessary step in the accident causa-
group, which in most populations is much tion process. Second, it is important to monitor
smaller than the no-accident group and is sub- the stress, both subjective and objective, that
ject to a high chance factor. Perhaps future employees experience, as unwanted stress levels
research should concentrate on those people
who manage over a long period of time to avoid
are likely to increase the chance of an accident.
an accident. Another sensible strategy would be Third, organizations must always keep in mind
to shift the focus of intervention toward the that accidents are often determined by multiple
positive benefits of safety and away from the factors, and even at the sharp end (i.e., the
negative effects of accidents, as research has human-machine interface) may involve people
shown that most people do not believe an acci-
dent will actually happen to them (Taylor, (either actively or passively), errors, violations,
1981). and technical failures.
The multiperspective approach to accident
Recommendations for Organizational investigation indicates that investigators should
Action consider a range of possible contributing fac-
Traditionally, the focus of human factors in tors rather than focus solely on the perpetrators
safety has been the individual. One aim of this and their actions. Although high levels of skill
review has been to demonstrate that accidents may lead to low levels of errors, this very level
themselves are largely meaningless as outcome of skill may foster complacency and, so, have
measures, being unreliable and stochastic in an adverse effect on attitudes and risk percep-
nature. What the more recent research of both tions, making violations more likely. Thus, the
cognitive and social psychologists emphasizes accidents of experienced and inexperienced
is the need to study the individual in context. workers may have quite different causes. This
To focus on the specific actions leading to an relationship is particularly interesting and
accident and to adopt remediation strategies would benefit from further investigation.
on the basis of the accident analysis is likely to It is also important to remember that
be as unreliable as accidents themselves. although human factors may be the immediate
Investigations should be aimed at the precur- precursors to accidents, inadequacies of the
sors to unsafe behavior. As these precursors system commonly underlie these failures. In a
Downloaded from hfs.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016
668 December 1998 – Human Factors

study of aircrew error, Chappelow (1989) dence of deliberate deviations from safe prac-
reported that in 44% of the 149 accidents under tices. To increase the effectiveness of safety
investigation significant errors were made by initiatives, violations as well as errors should
people remote from the critical events. These be tackled. Over the last few years training
errors included poor design of the equipment, (e.g., cockpit training and then crew resource
inadequacies in training and briefing, and ad- management) has evolved in the aviation
ministrative failures. The distinction between industry (Weiner, Kanki, & Helmreich, 1993).
active and latent errors, first made by Reason This form of team training aims to foster
(1990), is now widely accepted. The majority effective crew coordination and team working
of research examining individual differences in through better communications, decision mak-
accident liability has focused on the active ing, and leadership. By focusing on social
errors of people at the sharp end of the system: skills as well as cognitive factors, such training
the drivers, the pilots, the machine operators. is likely to have an impact on both error and
However, as Reason explained, the failures of violation rates.
managers, planners, and decision makers The reduction of violations requires a range
(latent errors) can be equally catastrophic: of organizational interventions. Continued
“very few unsafe acts [errors and violations] compliance with procedures that do not repre-
result in actual damage or injury, even in rela- sent safe and efficient practice is unlikely, and
tively unprotected systems. And in highly pro- therefore organizations should strive to ensure
tected systems, the various layers of defence that all the procedures they have in place rep-
can only be breached by the adverse con- resent the optimal way of working. Having
junction of several different causal factors” developed good procedures, compliance with
(p. 208). Such causal factors are likely to them must be supported. Providing the cor-
include latent failures at the managerial levels, rect equipment, rewarding safe behavior, en-
which, though difficult to detect, are easily couraging ownership of rules, and reducing
recoverable. A poor management decision (e.g., unrealistic work demands are just some of the
poor scheduling) may cause accidents via an ways in which organizations can encourage
increase in errors made in response to time compliance with procedures. (For more on this
pressure. It would seem therefore that investi- issue see Reason, Parker, & Lawton, in press,
gations of the human factor in accident causa- and Lawton, 1998.)
tion should also consider the personality, The multiperspective approach also sug-
performance, and behavior of the decision gests an important role for occupational stres-
makers in organizations. sors in accident causation. Understanding the
Organizations should also consider careful- way in which stress arising from the work and
ly the ways in which they attempt to improve home environments affects both performance
safety. Sanders and McCormick (1987) identi- and behavior is crucial for accident preven-
fied three strategies for the reduction of tion, particularly as organizations can take an
human error: training, selection, and design. Of active role in reducing workplace stressors
these they consider the design solution to be such as unscheduled overtime, excessive work-
the most effective because it does not require load, and time pressure. As Glendon and
the maintenance and support of either selec- McKenna (1995) pointed out, “because a sig-
tion or training. Moreover, it is much easier to nificant amount of stress is caused at work as
limit the occurrence of human errors by mak- a result of organizational culture and relation-
ing them impossible, difficult, or inconsequen- ships, stress is a problem for organizations
tial through design than it is to rely on fallible and not merely individuals” (p. 204).
human beings. However, training is the mea- In summary, in the coming millennium the
sure most widely employed to foster behav- role of the psychologist in industrial safety is
ioral safety. Although traditional skills training unlikely to focus on the search for accident-
might, by increasing the skill level of the prone individuals. The problem of accidents at
workforce, cause a reduction in the frequency work will not be solved by selection and train-
of errors, it will do little to reduce the inci- ing alone but, in addition, requires a more
Downloaded from hfs.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016
ACCIDENT LIABILITY 669

thoughtful analysis of the way in which indi- Chappelow, J. W. (1989). Remedies for aircrew error (Report No.
664). Farnborough, UK: Royal Air Force Institute for Aviation
vidual, cognitive, attitudinal, and organiza- Medicine.
tional factors interact to generate the unsafe Clement, R., & Jonah, B.A. (1984). Field dependency, sensation
seeking and driving behavior. Personality and Individual
behaviors implicated in accidents. Differences, 5, 87–93.
Denning, D. L. (1983). Correlates of employee safety perfor-
mance. Presented at the Southeastern I/O Psychology
ACKNOWLEDGMENTS Association Meeting, Atlanta, Georgia.
Donald, I., & Canter, D. (1994). Employee attitudes and safety in
the chemical industry. Journal of Loss Prevention in the
The work described in this paper was fund- Process Industries, 7, 203–208.
ed by the Health and Safety Executive (UK). Donald, I., & Young, S. (1996). Managing safety: An attitudinal
based approach to improving safety in organisations.
Its contents, including any opinions and con- Leadership and Organisational Development Journal, 17,
clusions expressed, are those of the authors 13–20.
Dunbar, D. (1943). Psychosomatics diagnosis. New York: Hoeber.
alone and do not necessarily reflect HSE poli- Elander, J., West, R., & French, D. (1993). Behavioural correlates
cy. We gratefully acknowledge the constructive of individual differences in road-traffic crash risk: An exami-
nation of methods and findings. Psychological Bulletin, 2,
criticisms of the editor, William Howell, and 279–294.
the three anonymous reviewers who, together, Evans, L. (1991). Traffic safety and the driver. New York: Van
Nostrand Reinhold.
have considerably improved upon the first Eysenck, H. J. (1947). Dimensions of personality. London:
submission of this manuscript. Routledge & Kegan Paul.
Eysenck, H. J. (1970). The structure of human personality (3rd.
ed.). London: Methuen.
REFERENCES Eysenck, H. J., & Eysenck, S. B. G. (Eds.). (1968). Manual for the
Eysenck Personality Inventory. San Diego, CA: Educational
and Industrial Testing Service.
Adelstein, A. M. (1952). Accident proneness: A criticism of the Farmer, E. (1984). Personality factors in aviation. International
concept based upon an analysis of shunters’ accidents. Journal Journal of Aviation Safety, 2, 175–179.
of the Royal Statistical Society, 113, 354–400. Farmer, E., & Chambers, E. G. (1926). A psychological study of
Aiken, L. R. (1979). Psychological testing and assessment (3rd individual differences in accident liability (Industrial Fatigue
ed.). Boston: Allyn and Bacon. Research Board Report No. 38). London: Her Majesty’s
Ajzen, I. (1988). Attitudes, personality and behaviour. Milton Stationery Office.
Keynes, England: Open University Press. Fernandez, J. L. S. (1978). Psychology of the automobile driver:
Allodi, F., & Montgomery, R. (1979). Psychological aspects of Personality factors of drivers with multiple accidents. Revista
occupational injury. Social Psychiatry, 14, 25–29. De Psicologia General y Aplicada, 33, 217–228.
Arthur, W., Barrett, G. V., & Alexander, R. A. (1991). Prediction Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and
of vehicular accident involvement: A meta-analysis. Human behavior. Reading, MA: Addison-Wesley.
Performance, 4, 89–105. Foreman, E. I., Ellis, H. D., & Beavan, D. (1983). Mea culpa? A
Arthur, W., Strong, M. H., & Williamson, J. (1994). Validation of study of the relationships among personality traits, life-events
a visual attention test as a predictor of driving accident and ascribed accident causation. British Journal of Clinical
involvement. Journal of Occupational and Organisational Psychology, 22, 223–224.
Psychology, 67, 173–182. Free, R. (1994). The role of procedural violations in railway acci-
Avolio, B. J., Kroeck, K. G., & Panek, P.E. (1985). Individual dif- dents. Unpublished doctoral dissertation, University of
ferences in information processing ability as a predictor of Manchester, UK.
motor vehicle accidents. Human Factors, 27, 577–587. Glendon, A. I. (1979). Safety: Whose responsibility. Occupational
Battman, W., & Klumb, P. (1993). Behavioural economics and Health, 31, 31–37.
compliance with safety regulations. Safety Science, 16, 35–46. Glendon, A. I., & McKenna, E. F. (1995). Human safety and risk
Beirness, D. J. (1993). Do we really drive as we live? The role of management. London: Chapman and Hall.
personality factors in road crashes. Alcohol, Drugs and Gopher, D., & Kahneman, D. (1971). Individual differences in
Driving, 6, 129–143. attention and the prediction of flight criteria. Perceptual and
Beirness, D. J., & Simpson, H. M. (1988). Lifestyle correlates of Motor Skills, 33, 1342–1355.
risky driving and accident involvement among youth. Alcohol, Grayson, G., & Maycock, G. (1988). From proneness to liability.
Drugs and Driving, 4, 193–204. In J. A. Rothengatter & R. A. de Bruin (Eds.), Road user
Biesheuvel, S., & White, M. E. (1949). The human factor in flying behaviour (pp. 234–241). Assen, Netherlands: Van Gorcum.
accidents. South African Air Force Journal, 1, 25–36. Greenwood, M., & Woods, H. M. (1919). A report on the inci-
Booysen, A. E., & Erasmus, J. A. K. (1989). The relationship dence of industrial accidents with special reference to multiple
between some personality factors and accident risk. South accidents (Industrial Fatigue Research Board Report No. 4).
African Journal of Psychology, 19, 144–152. London: Her Majesty’s Stationery Office.
Broadbent, D. E., Broadbent, M. H. P., & Jones, J. L. (1986). Per- Hale, A. R., & Glendon, A. I. (1987). Individual behaviour in the
formance correlates of self-reported cognitive failure and of control of danger. Amsterdam: Elsevier.
obsessionality. British Journal of Clinical Psychology, 25, Hale, A. R., & Hale, M. (1972). A review of the industrial acci-
285–299. dent research literature. London: Her Majesty’s Stationery
Broadbent, D. E., Cooper, P. F., Fitzgerald, P., & Parkes, K. R. Office.
(1982). Cognitive Failures Questionnaire (CFQ) and its corre- Hansen, C. P. (1988). Personality characteristics of the accident
lates. British Journal of Clinical Psychology, 19, 177–188. involved employee. Journal of Business and Psychology, 2,
Canter, D., & Donald, I. (1990, July). Accident by design: 346–365.
Environmental, attitudinal, and organisational aspects of Hansen, C. P. (1989). A causal model of the relationship between
accidents. Presented at the 11th International Conference of accidents, biodata, personality and cognitive factors. Journal
the International Association for the Study of People and of Applied Psychology, 74, 81–90.
Their Physical Settings, Ankara, Turkey. Harano, R. M. (1970). Relationship of field dependence and
Cattell, R. B. (1965). The scientific analysis of personality. motor vehicle accident involvement. Perceptual and Motor
Harmondsworth, England: Penguin. Skills, 31, 272–274.
Downloaded from hfs.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016
670 December 1998 – Human Factors

Harano, R. M., Peck, R. C., & McBride, R. S. (1975). The predic- Meadows, M. L. (1994). Psychological correlates of road crash
tion of accident liability through biographical data and psy- types. Unpublished doctoral dissertation, University of
chometric tests. Journal of Safety Research, 7, 16–52. Manchester, UK.
Harrell, W. A. (1995). Accident history and perceived risk of Melamed, S., Luz, J., Najenson, T., Jucha, E., & Green, M. (1989).
injury as factors influencing fatalism about occupational acci- Ergonomic stress levels, personal characteristics, accident
dents. Perceptual and Motor Skills, 81, 665–666. occurrence and sickness absence among factory workers.
Hartley, L. R., & El Hassani, J. (1994). Stress, violations, and acci- Ergonomics, 32, 1101–1110.
dents. Applied Ergonomics, 25, 221–230. Mihal, W. L., & Barrett, G. V. (1976). Individual differences in
Heinrich, H. W., Petersen, D., & Roos, N. (1980). Industrial acci- perceptual information processing and their relation to auto-
dent prevention (5th ed.). New York: McGraw-Hill. mobile accident involvement. Journal of Applied Psychology,
Hilakivi, I., Veilahti, J., Asplund, P., Sinivuo, J., Laitinen, L., & 61, 229–233.
Koskenvuo, K. (1989). A sixteen-factor personality test for Mozdzierz, G. J., Macchitelli, F. J., Planek, T. W., & Lottman, T. J.
predicting automobile driving accidents of young drivers. (1975). Personality and temperament differences between
Accident Analysis and Prevention, 21, 413–418. alcoholics with high and low records of traffic accidents and
Hockey, G. R. J., Clough, P. J., & Maule, A. J. (1996, September). violations. Journal of Studies on Alcohol, 36, 395–399.
Effects of emotional state on decision making and risk behav- Murphy, D. J. (1981). Farm safety attitudes and accident involve-
iour. Presented at the Risk and Human Behaviour (Economic ment. Accident Analysis and Prevention, 13, 331–337.
and Social Research Council) Conference, York. Newbold, E. M. (1927). Practical applications of the statistics of
Hofmann, D. A., Jacobs, R., & Landy, F. (1995). High reliability repeated events. Journal of the Royal Statistical Society, 92,
process industries: Individual, micro, and macro organisation- 487–535.
al influences on safety performance. Journal of Safety Norman, D. A. (1981). Categorisation of action slips. Psycho-
Research, 26, 131–149. logical Review, 88, 1–15.
Hoyle, R. H. (Ed.). (1995). Structural equation modeling: Olearnik, H., & Canter, D. (1988, December). Empirical valida-
Concepts, issues and applications. London: Sage. tion of the relationship between safety attitudes and industrial
Hurst, N. W., Young, S., Donald, I., Gibson, H., & Muyselaar, A. accidents. Presented at the British Psychological Society
(1996). Measures of safety management performance and atti- London Conference.
tudes to safety at major hazard sites. Journal of Loss Parker, D., Reason, J. T., Manstead, A. S. R., & Stradling, S.
Prevention in Process Industries, 9, 161–172. (1995). Driving errors, driving violations and accident
Janzen, J. M. (1983). A study of the relationship of locus of con- involvement. Ergonomics, 38, 1036-1048.
trol, age and work experience used to discriminate individuals Parker, D., West, R., Stradling, S., & Manstead, A. S. R. (1995).
in the saw mill industry. Dissertation Abstracts International, Behavioural characteristics and involvement in different types of
44, 438A. traffic accident. Accident Analysis and Prevention, 27, 571–581.
Jones, J. W. (1984). The safety locus of control scale (Tech. Pervin, L. A. (1993). Personality: Theory and research (6th ed.).
Report). St. Paul, MN: St. Paul Companies. New York: Wiley.
Jones, J. W., & Wuebker, L. J. (1985). Development and validation Pestonjee, D. M., & Singh, U. B. (1980). Neuroticism-extrover-
of the Safety Locus of Control (SLC) scale. Perceptual and sion as correlates of accident occurrence. Accident Analysis
Motor Skills, 61, 151–161. and Prevention, 12, 201–204.
Kahneman, D., Ben-Ishai, R., & Lotan, M. (1973). Relation of a Porter, C. S. (1988). Accident proneness: A review of the concept.
test of attention to road accidents. Journal of Applied In D. J. Oborne (Ed.), International reviews of ergonomics:
Psychology, 58, 113–115. Current trends in human factors research and practices (Vol.
Kishore, G. S., & Jha, S. (1978). Psycho-motor efficiency as a fac- 2, pp. 177–206). London: Taylor & Francis.
tor in traffic accidents. Psychologia: An International Journal Porter, C. S., & Corlett, E. N. (1989). Performance differences of
of Psychology in the Orient, 21, 107–110. individuals classified by questionnaire as accident prone or
Lardent, C. L. (1991). Pilots who crash: Personality constructs non-accident prone. Ergonomics, 32, 317–333.
underlying accident prone behavior of fighter pilots. Powell, P. I., Hale, M., Martin, J., & Simon, M. (1971). 2000 acci-
Multivariate Experimental Clinical Research, 10, 1–25. dents: A shop floor study of their causes (Report No. 21).
Lawton, R. (1998). Not working to rule: Understanding procedur- London: National Institute of Industrial Psychology.
al violations at work. Safety Science, 28, 77–95. Quimby, A. R., & Watts, G. R. (1981). Human factors and driving
Lawton, R., & Parker, D. (1998). Individual differences in acci- performance (Department of Transport TRRL Report No. LR
dent liability: A review (Contract research report 175). 1004). Crowthorne, UK: Transport and Road Research
Sudbury, England: Health and Safety Executive Books. Laboratory.
Lee, T. R., Macdonald, S. M., & Coote, J. A. (1993, October). Ranney, T. A. (1994). Models of driving behavior: A review of their
Perceptions of risk and attitudes to safety at a nuclear repro- evolution. Accident Analysis and Prevention, 26, 733–750.
cessing plant. Presented at the Society for Risk Assessment Rasmussen, J. (1982). Human errors: A taxonomy for describing
(Europe) Fourth Conference, Rome. human malfunction in industrial installations. Journal of
Lester, J. (1991). Individual differences in accident liability: A Occupational Accidents, 4, 311–335.
review of the literature (Contractors’ Report No. 306). Reason, J. T. (1974). Man in motion: The psychology of travel.
Crowthorne, UK: Transport and Road Research Laboratory. London: Weidenfeld and Nicolson.
Loo, R. (1978). Individual differences in the perception of traffic Reason, J. T. (1987). The Chernobyl errors. Bulletin of the British
signs. Human Factors, 20, 65–74. Psychological Society, 106, 321–331.
Mayer, D. L., Jones, S. F., & Laughery, K. R. (1987). Accident Reason, J. T. (1990). Human error. Cambridge: Cambridge
proneness in the industrial setting. In Proceedings of the University Press.
Human Factors Society 31st Annual Meeting (pp. 196–199). Reason, J. T., Manstead, A. S. R., Stradling, S. G., Baxter, J. S., &
Santa Monica, CA: Human Factors and Ergonomics Society. Campbell, K. (1990). Errors and violations on the road: A real
Mayer, R. E., & Treat, J. R. (1977). Psychological, social and cog- distinction? Ergonomics, 33, 1315–1332.
nitive characteristics of high risk drivers: A pilot study. Reason, J. T., Parker, D., & Free, R. (1994). Bending the rules:
Accident Analysis and Prevention, 9, 1–8. The varieties, origins and management of safety violations.
McKenna, F. P., & Crick, J. (1991). Experience and expertise in haz- Leiden, Netherlands: Rijks Universiteit Leiden.
ard perception. In G. B. Grayson & J. F. Lester (Eds.), Reason, J. T., Parker, D., & Lawton, R. (in press). Organisational
Behavioural research in road safety: Proceedings of a seminar at controls and safety: The varieties of rule-related behaviour.
Nottingham University 26-27 September 1990 (pp. 39–46). Journal of Occupational and Organizational Psychology.
Crowthorne, England: Transport and Road Research Laboratory. Rosenthal, R. (1990). How are we doing in soft psychology.
McKenna, F. P., Duncan, J., & Brown, I. D. (1986). Cognitive abil- American Psychologist, 45, 775–777.
ities and safety on the road: A re-examination of individual Rotter, J. B. (1966). Generalized expectancies for internal versus
differences in dichotic listening and search for embedded fig- external control of reinforcement. Psychological Monographs,
ures. Ergonomics, 29, 649–663. 80, 609.
Downloaded from hfs.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016
ACCIDENT LIABILITY 671

Salminen, S., & Tallberg, T. (1996). Human errors in fatal and serious Wagenaar, W. A., Hudson, P. T. W., & Reason, J. T. (1990). Cog-
occupational accidents in Finland. Ergonomics, 39, 980–988. nitive failures and accidents. Applied Cognitive Psychology, 4,
Sanders, M. S., & McCormick, E. J. (1987). Human factors in 273–294.
engineering design. New York: McGraw-Hill. Weiner, E. L., Kanki, B. G., & Helmreich, R. L. (1993). Cockpit
Schenk, J., & Rausche, A. (1979). The personality of accident- resource management. London: Academic.
prone drivers. Psychologie und Praxis, 23, 179–186. West, R. (1993). Accident script analysis: Parts 1 and 2. Report to
Shaw, L., & Sichel, H. S. (1971a). Accident proneness. Oxford: Department of Transport (Contract No. N06100). London:
Pergamon. Department of Transport.
Shaw, L., & Sichel, H. S. (1971b). Accident research: International series Wong, W. A., & Hobbs, G. E. (1949). Personal factors in industri-
of monographs in experimental psychology. Oxford: Pergamon. al accidents. Industrial Medicine and Surgery, 18, 291–294.
Sherry, P. (1991). Person-environment fit and accident prediction. Wuebker, L. J., Jones, J. W., & Dubois, D. (1985). Safety locus of
Journal of Business and Psychology, 5, 411–416. control and employee accidents (Tech. Report). St. Paul, MN:
Sims, M. T., Graves, R. J., & Simpson, G. C. (1984). Mineworkers’ St. Paul Companies.
scores for the Rotter Internal-External Locus of Control Scale. Zuckerman, M. (1979). Sensation seeking: Beyond the optimal
Journal of Occupational Psychology, 57, 327–329. level of arousal. Hillsdale, NJ: Erlbaum.
Steffy, B. D., Jones, J. W., Murphy, L. R., & Kunz, L. (1986). A
demonstration of the impact of stress abatement programs on Rebecca Lawton is a research fellow in the
reducing employees’ accidents and their costs. American
Journal of Health Promotion, 1(2), 25–32.
Department of Psychology, University of Man-
Sutherland, V. J. (1993). The use of a stress audit. Leadership and chester, where she received her Ph.D. in psychology
Organisation Development Journal, 14, 22–28. in 1994.
Sutherland, V. J., & Cooper, C. L. (1991). Personality, stress and
accident involvement in the offshore oil and gas industry. Dianne Parker is a lecturer in the Department of
Personality and Individual Differences, 12, 195–204. Psychology, University of Manchester, where she
Taylor, D. H. (1981). The hermeneutics of accidents and safety.
Ergonomics, 24, 487–495. received her Ph.D. in psychology in 1992.
Tillman, W. A., & Hobbs, G. R. (1949). The accident-prone auto-
mobile driver: A study of the psychiatric and social back- Date received: February 28, 1997
ground. American Journal of Psychiatry, 106, 321–331. Date accepted: June 5, 1998

Downloaded from hfs.sagepub.com at PENNSYLVANIA STATE UNIV on May 17, 2016

You might also like