Sie sind auf Seite 1von 12

Journal of Occupational Health Psychology Copyright 2003 by the Educational Publishing Foundation

2003, Vol. 8, No. 4, 316 327 1076-8998/03/$12.00 DOI: 10.1037/1076-8998.8.4.316

Workplace Safety Performance: Conscientiousness,


Cognitive Failure, and Their Interaction
J. Craig Wallace Stephen J. Vodanovich
Georgia Institute of Technology University of West Florida

This research investigated the effects of cognitive failure on workplace safety and accidents over
2 studies. It was hypothesized that cognitive failure would directly predict safety behavior and
workplace accidents and predict these outcomes over and above conscientiousness. It was found
that cognitive failure uniquely accounted for workplace safety behavior and accidents. However,
it has been suggested by researchers that certain individual differences might interact to produce
differential effects. Thus, a moderated model was tested examining the interaction of cognitive
failure and conscientiousness. It was found that cognitive failure moderated the relationship
between conscientiousness and accidents and unsafe work behaviors. Overall, results suggest that
cognitive failure plays an important part in individual safety behavior, especially when consci-
entiousness is low.

Workplace accidents are reported to result in thou- sparse. Thus, the purpose of the present study is
sands of deaths and permanent disabilities each year twofold. First, it is to explore and confirm the rela-
in the United States, as well as increased accident- tionship among cognitive failure and conscientious-
related costs. For example, in 1982, accidents ac- ness, workplace safety behavior, and workplace ac-
counted for nearly 100,000 deaths and 9 million cidents. Second, it is to test the moderating effect of
disabling injuries (National Safety Council, 1983). In cognitive failure on the relationship between consci-
1998, there were almost 5,000 workplace deaths and entiousness and workplace safety and accidents. Two
almost 4 million disabling injuries in the United separate studies were conducted to examine these
States. This has been equated to over $108 billion issues.
being lost due to medical expenses, lost wages, and
administrative expenses (National Safety Council,
1999). Although the number of disabling injuries has
Cognitive Failure
somewhat decreased over the past few years, the Reason (1988) stated that failures consist of two
occurrence of job-related accidents is still high. unique classes: planning failures (e.g., mistakes) and
A multitude of studies have examined the cogni- execution failures (e.g., cognitive failures). Cognitive
tive aspects of accidents (e.g., Hale & Hale, 1972; failure was conceptualized as an enduring traitlike
Hansen, 1988), with most of this research being construct, and empirical work has supported this
focused on differentiating worker characteristics. view (Broadbent, Cooper, Fitzgerald, & Parkes,
Relatively recently, research has examined the rela- 1982; Vom Hofe, Mainemarre, & Vannier 1998).
tionship between the construct of cognitive failure Specifically, a cognitive failure has been described as
and accident rates (e.g., Larson, Alderton, Neideffer, a breakdown in cognitive functioning that results in a
& Underhill, 1997; Larson & Merritt, 1991). How- cognitively based mistake or error in task execution
ever, studies investigating this relationship are fairly that a person should normally be capable of complet-
ing (Martin, 1983), with some people being more
prone to experiencing cognitive failures than others
J. Craig Wallace, School of Psychology, Georgia Institute (Wallace, Kass, & Stanny, 2002). Reason (1988)
of Technology; Stephen J. Vodanovich, Department of Psy- further stated that people prone to cognitive failure
chology, University of West Florida. might have a rigid attentional focus, thereby creating
This article was presented in April 2003 at the 18th
Annual Society for Industrial/Organizational Psychology a cognitive management style that is inflexible. Such
Conference, Orlando, Florida. a style might allow for the occurrence of cognitive
We thank Gilad Chen for his helpful comments and breakdowns in dealing with intervening and concur-
feedback. rent stimuli. For instance, individuals with high lev-
Correspondence concerning this article should be addressed
to J. Craig Wallace, School of Psychology, Georgia Institute of
els of cognitive failure may be relatively unaware of
Technology, 654 Cherry Street, J. S. Coon Building, Atlanta, dynamic or novel conditions or that incoming infor-
Georgia 30332-0170. E-mail: CraigWallace@comcast.net mation exceeds their attentional capacity while per-
316
WORKPLACE SAFETY AND ACCIDENTS 317

forming tasks. This might be especially true in the tween automobile accidents and CFQ scores. For
presence of task automization (Heckhausen & Beck- instance, Larson and Merritt (1991), after excluding
mann, 1990). Additionally, accidents are commonly extreme number of automobile accidents (i.e., more
the result of faulty cognitive/attentional processes than eight accidents), found a positive relationship
such as poor selective attention, mental errors, and between cognitive failure and self-reports of accident
distractibility (Arthur, Barrett, & Alexander, 1991; rates. Similarly, Larson et al. (1997) found cognitive
Hansen, 1989; Mihal & Barrett, 1976). Therefore, it failure to be positively related to reports of automo-
appears plausible that the occurrence of cognitive bile accidents, occurrences of hospitalizations, and
failures and unsafe behaviors and accidents may be injuries from falling among a sample of United States
significantly related (Wallace & Vodanovich, 2003). Naval recruits. Finally, congruent with the results of
Early research by Reason (e.g., 1977, 1979, 1984) Larson et al. (1997), Wallace and Vodanovich (2003)
proposed that cognitive failures tend to occur under reported a positive relationship between cognitive
familiar conditions in which tasks are automated and failure and automobile accidents.
a persons attention is externally or internally dis- On the basis of this research, we believe that ones
tracted or preoccupied. For example, accidentally proneness for cognitive failure will positively relate
cheering for the opposing team, replacing the coffee to reports of on-the-job accidents as well as unsafe
lid with the lid for the sugar (Manly, Robertson, workplace behaviors. We operationalized accidents
Galloway, & Hawkins, 1999), or throwing away a and unsafe work behaviors as separate outcomes be-
book of unused matches and retaining the used match cause they have been shown to only moderately
(Broadbent et al., 1982) are all common, everyday relate to each other (Hofmann & Stetzer, 1996) and
examples of cognitive failure. However, other more because unsafe work behaviors do not necessarily
serious cognitive failures have been studied by vari- result in an accident. Thus, the following hypothesis
ous researchers (e.g., Broadbent et al., 1982; Reason, is presented to test and fill the gap in the literature
1977, 1979), such as an air traffic controller over- regarding accidents and unsafe work behaviors.
looking an inbound airplane, shutting down the lone
surviving engine when all other engines had failed, Hypothesis 1: Cognitive failure will positively
and pilots plotting the flight course by the wrong end relate to workplace accidents and unsafe work
of the compass needle. behaviors.
Over the past few years the predictive ability of
ones proneness for cognitive failure, captured by the Conscientiousness, Safety Behavior,
Cognitive Failures Questionnaire (CFQ; Broadbent et and Accidents
al., 1982), has been examined by correlating it with
several constructs. For instance, Reason and Lucas Facets of personality have long been shown to
(1984) found a positive relationship between cogni- possess a great deal of utility in predicting job per-
tive failure and absentminded shoplifting. Likewise, formance (Barrick & Mount, 1991; Salgado, 1997).
Robertson, Manly, Andrade, Baddeley, and Yiend Specifically, the five-factor model (FFM) of person-
(1997) and Wallace, Kass, and Stanny (2001) found ality, which consists of Conscientiousness, Agree-
a significant negative relationship between cognitive ableness, Extroversion, Openness to Experience, and
failure and performance on a no-go sustained at- Emotional Stability (Costa & McRae, 1992; Gold-
tention task (i.e., participants are required to withhold berg, 1992), has been used by many investigators in
a response). However, few studies have directly ex- this context. In particular, conscientiousness has been
amined the relationship between cognitive failure and shown to predict job performance in studies across a
workplace safety and accidents. variety of occupations and occupational levels (Stew-
One such study that has been conducted in this area art, 1999), including meta-analytic studies (e.g., Bar-
found a significant, positive relationship between CFQ rick & Mount, 1991; Salgado, 1997). A conscientious
scores and self-reports of work-related accidents (Wal- person is believed to possess qualities that reflect
lace & Vodanovich, 2003). The authors also found that dependability (e.g., thorough, careful, organized, re-
self-reports of accidents and safety behavior were sig- sponsible), competence, and dutifulness as well as
nificantly correlated with more objective reports of ac- volitional constructs such as hardworking and
cidents and safety (i.e., supervisor and organizational achievement oriented that a variety of jobs require
records). To date, this is the only study that has directly (Barrick & Mount, 1991). However, only a handful
examined the relationship between cognitive failure and of studies have examined the relationship between
occupational safety. the FFM (and specifically conscientiousness) and
Some studies have examined the relationship be- workplace accidents.
318 WALLACE AND VODANOVICH

In one study, Fallon, Avis, Kudisch, Gornet, and closely examine the relationship among cognitive
Frost (2000) did not find a significant relationship failure, unsafe behaviors, and accidents while con-
between conscientiousness and safety. Rather, con- trolling for conscientiousness. This was done to de-
scientiousness significantly predicted overall perfor- termine if ones proneness for cognitive failure
mance ratings, willingness to rehire, and employee would account for unique variance in safety out-
attendance. However, it should be noted that Fallon comes beyond conscientiousness. It is believed that
et al.s sample consisted of retail salespersons cognitive failure would account for incremental vari-
wherein the frequency of accidents is less common ance in safety behavior and accidents over and above
than other occupations. Also, the measure of safety conscientiousness scores. This is because the two
behavior used in this study was combined with rat- constructs can broadly be considered to reflect op-
ings of integrity, which could compromise the accu- posing regulatory styles. That is, highly conscien-
racy of measurement of workplace safety behavior. A tiousness individuals are typically hard-working, dis-
more recent study reported a significant relationship ciplined, and organized, which has been shown to
between accidents and conscientiousness (Cellar, lead to behaviors that are consistent with on-task
Nelson, Yorke, & Bauer, 2001). However, the sample processes and behaviors (e.g., Barrick, Stewart, &
consisted solely of undergraduate students. Finally, Piotrowski, 2002). Conversely, people with a higher
Arthur and Graziano (1996) found a significant, neg- propensity for cognitive failure tend to be absent-
ative relationship between conscientiousness and be- minded and forgetful, easily distracted, and com-
ing involved in driving accidents. That is, individuals monly involved in slips of action, all of which are
who rated themselves higher on the conscientious- consistent with off-task processes and behaviors,
ness scale of the FFM were less likely to be involved which result in a poor cognitive management style
in automobile accidents than those who rated them- (Larson et al., 1997; Reason, 1988; Wallace et al.,
selves lower on conscientiousness. Unlike Cellar et 2002). Thus the following hypothesis is presented:
al. (2001), Arthur and Graziano (1996) used both
undergraduates and employees from a temporary em- Hypothesis 3: Cognitive failure will account for
ployment service. unique variance in workplace accidents and unsafe
These disparate results do not provide conclusive work behaviors over and above conscientiousness.
evidence for the role of conscientiousness in predict-
ing safety behavior or accidents. Thus, one of the Cognitive Failure Conscientiousness
goals of the present study was to demonstrate the
relationship between conscientiousness and occupa- Hansen (1989) suggested that the impact of poten-
tional safety and accidents. It was anticipated that tial mediating and moderating variables within the
conscientiousness would be negatively related to un- individual differenceaccident relationship needs to
safe behaviors and accidents as highly conscientious be explored. He further stated that such techniques
individuals tend to exhibit behaviors that are more would allow researchers to elaborate on relationships
reflective of on-task processes (e.g., organized, duti- beyond typical accident research designs (see Han-
ful, careful), which we believe results in fewer acci- sen, 1988; Larson & Merritt, 1991). Recently, Witt,
dents and less engagement in unsafe work behaviors. Burke, Barrick, and Mount (2002) demonstrated the
Specifically, the following hypothesis is presented: moderating effects of conscientiousness and agree-
ableness in predicting job performance. It was found
Hypothesis 2: Conscientiousness will negatively that highly conscientiousness individuals who also
relate to unsafe work behaviors and workplace tended to be high on agreeableness received higher
accidents. performance scores as opposed to those who were
high on conscientiousness and low on agreeableness.
Cognitive Failure and Conscientiousness In a similar study, Tziner, Murphy, and Cleveland
(2002) found that conscientiousness actually moder-
Although previous work has found cognitive fail- ated the relationship between rater attitudes and rater
ure to be positively related to accidents (e.g., Larson behavior when assessing worker performance. These
et al., 1997), a similar relationship has not been findings highlight the possible interaction effects of
explored between cognitive failure and safety behav- stable individual differences and the different predic-
ior. Nor has cognitive failure been shown to relate to tive ability of the constructs at different levels.
unsafe behavior and accidents after controlling for Applying such theoretical reasoning to the present
more robust constructs such as conscientiousness. study, and heeding the request of Hansen (1989), it
Thus, another goal of the present study was to more seems likely that conscientiousness and cognitive
WORKPLACE SAFETY AND ACCIDENTS 319

failure might interact in a similar way. Reason (1988) years, with 83% claiming safety was a major concern in
stated that people prone to cognitive failure are not their job and only 8% claiming safety was of no concern.
All of the participants completed the measures described
able to handle interfering stimuli during task engage-
below (i.e., CFQ, safety questionnaire, accident data sheet,
ment. Consequently, these individuals fail to com- and conscientiousness measure) in a single session. Each
plete the task. Highly conscientious individuals, in employee was sent a message by e-mail that contained a
contrast, tend to be more task-oriented, organized, link to a secure Web page on which the measures and their
diligent, methodical, and purposeful (Witt et al., corresponding directions were located. Participants had
been informed prior to receiving the e-mail that a study was
2002). They repeatedly check their progress, which is being conducted that could help create a safer working
likely to lead to less errors and ultimately to less environment and were encouraged to participate. Participa-
accidents. This suggests that conscientiousness tion was voluntary, and all data were collected anony-
would be negatively related to unsafe work behaviors mously. Each employee was also informed that no other
employee of the organization would be allowed access to
and accidents at higher levels of cognitive failure. In
individual reports. Participants completed the survey at their
other words, if individuals are low on conscientious- leisure. Reminders were sent out 1 week after initial recruit-
ness, then an additional scattiness, through high ment e-mails. Two-hundred fifty workers were contacted,
propensity for cognitive failure, might lead to higher which resulted in a return rate of 75.6%. After completing
rates of unsafe work behaviors and accidents. Indi- the survey, participants were directed to another Web page
that provided debriefing information and contact informa-
viduals who are high on conscientiousness might be tion if they desired further information.
able to overcome a propensity for committing a cog- Again, it should also be noted that both unsafe work
nitive failure by tending to engage in behaviors that behaviors and workplace accidents were treated as separate
would reduce the occurrence of accidents. Addition- outcomes given previous research that has found only a
moderate relationship between the two, which suggests that
ally, because conscientiousness is considered to be an
these behaviors are relatively distinct (e.g., Hechanova-
integral part of the most basic personality taxonomy Alampay & Beehr, 2001; Hoffman & Stetzer, 1996). Thus,
(Hough & Schneider, 1996), we believe its relation- we examined these behaviors separately to more closely
ship to safety will be moderated by cognitive failure. examine the relationship conscientiousness and cognitive
Thus the following hypothesis is presented: failure might share with both unsafe behaviors and
accidents.
Hypothesis 4: Cognitive failure will moderate
the relationship between conscientiousness and Materials
unsafe work behaviors and accidents such that
the negative relationship between conscientious- CFQ. The CFQ (Broadbent et al., 1982) comprises 25
items that inquire about a persons minor mistakes (e.g.,
ness and unsafe work behaviors and accidents physical blunders, forgetfulness, and distractibility) in ev-
will be stronger at higher levels of cognitive eryday life over the past 6 months to assess ones proneness
failure than at lower levels. for cognitive failure. Although there was originally no time
period included with the instructions on the CFQ, the au-
To examine and clarify these issues and test the thors of the CFQ found that participants were unclear re-
garding how to reply to the items on the scale. In addition,
proposed hypotheses, we conducted two separate it was believed that this time period might be sensitive to
studies in an effort to increase the generalizability of temporary state rather than underlying trait . . . and, thus
the findings. Specifically, we examined the hypothe- useful as an indicator of occupational stresses (Broadbent
sized relationships using a sample of full-time pro- et al., 1982, p. 3). However, Broadbent et al. (1982) con-
duction workers and a sample of military machinists. cluded that the CFQ appears to measure a general liability
to failure (p. 13) and, after further administrations of the
CFQ, they decided that a 6-month time frame was optimal
Study 1 to use when asking participants to complete the scale
(Broadbent et al., 1982). Subsequent research has continued
to use this 6-month time frame (e.g., Martin, 1983; Reason
Method & Lucas, 1984; Wallace & Vodanovich, 2003), and test
retest reliability results tend to support the stability of the
Participants and Procedure construct (e.g., r .82 over a 2-month time period; Vom
Hofe et al., 1998). The internal consistency of the CFQ in
A sample of 219 production workers located in the south- the present study was .90. Additionally, the CFQ has been
eastern United States participated in the present study. correlated with a number of other constructs such as Type A
Within the sample, 91 participants were female, and the behavior pattern, attention deficit/hyperactivity disorder,
mean age was 33.2 years (SD 6.5). Employees were boredom proneness (Wallace et al., 2002), and vigilance
responsible for various activities in the facility that ranged performance (Robertson et al., 1997; Wallace et al., 2001).
from maintenance to assembly-line work. However, all of Workplace Safety Questionnaire. Hoffman and Stetzers
the participants were involved in the production of the (1996) Workplace Safety Questionnaire was used as a mea-
organizations product (i.e., chemical production). The av- sure of safety behavior and performance. It contains 29
erage experience level of the sample was found to be 4.4 items that relate to improper tool use, improper work strat-
320 WALLACE AND VODANOVICH

egies, inappropriate personal protection equipment, and im- 1992). Similarly, the Mini-Marker (Saucier, 1994) also
proper storage of tools. A high score on the measure con- demonstrated strong correlations with the NEOFive-Factor
stitutes unsafe work behavior. Hoffman and Stetzer reported Inventory with correlations ranging from .54 to .70 (Moo-
the coefficient alpha of the scale to be .89 for the entire radian & Nezlek, 1996).
scale. In the present study the internal consistency was .83.
Accident data. Participants were asked to complete a
short questionnaire devised for the present study pertaining
to their work history. Specifically, the following question
was asked: How many work-related accidents (i.e., minor Results
and major accidents) have you been responsible for causing
over the last 3 years? Examples of work-related accidents It was proposed that cognitive failure scores would
were provided to participants in an effort to provide them be positively related to workplace accidents and un-
with the range of accidents that were considered minor safe behaviors (Hypothesis 1). Correlation analyses
and major. For example, a minor accident was described
as something that delayed the completion of a project but
fully supported this hypothesis (see Table 1). Signif-
did not result in bodily harm requiring medical attention icant, positive relationships emerged between cogni-
beyond basic first aid. An example of a major accident was tive failure and accidents (raw r .29, p .05;
a bodily injury in which medical attention was required or in transformed r .35, p .05) as well as between
which equipment was damaged. Major accidents are similar cognitive failure and unsafe work behaviors (r .31,
to the notion of microaccidents as described by Zohar
(2000). We inquired about accidents over the 3 years as p .05).
accidents are not overly common in short durations (e.g., 3 Hypothesis 2 proposed that a negative relationship
months) and aggregated minor and major accidents into a would exist between conscientiousness and work-
single outcome (i.e., accidents). This was also done in an place accidents and unsafe work behaviors, respec-
effort to overcome the problem associated with skewed
distributions commonly associated with accident data (Zo-
tively. Correlation analysis indicated full support for
har, 2000, 2002). However, an examination of the accident these hypotheses as well. Specifically, significant
distribution found a positively skewed distribution. There- negative relationships were found between conscien-
fore, we performed a square-root transformation on the tiousness and accidents (raw r .17, p .05;
accident data. Similar to Hofmann and Stetzer (1996), both transformed r .17, p .05) and between consci-
the raw and transformed accident data are analyzed and
presented. entiousness and unsafe work behaviors (r .33,
The use of self-report accident data is consistent with the p .05).
approach used by Larson and Merritt (1991) and Larson et To examine the impact that cognitive failure has
al. (1997) in their examination of the association between on safety performance, we conducted two hierarchi-
cognitive failures and accident/injury rate. Additionally,
Hoffman and Stetzer (1996) and Wallace and Vodanovich
cal regression analyses. Initially, the predictive utility
(2003) have demonstrated that self-report and objective of cognitive failure was tested for its unique impact
reports of accidents and safety are significantly related (e.g., on workplace accidents over and above conscien-
r .65). tiousness (see Table 2). Conscientiousness was en-
Conscientiousness. Sauciers (1994) Mini-Markers, tered in the first step of the equation and was found to
which is a brief version of Goldbergs (1992) Big-Five
Adjective Checklist, was used as a measure of the FFM; be a significant predictor, F(1, 191) 5.4, p .05.
specifically, the Conscientiousness scale was used ( At the second step, cognitive failure was entered into
.85). Goldbergs (1992) measure of the FFM has been the equation and was also found to be significant,
correlated with other measures of the FFM such as the F(2, 190) 10.6, p .05. The final model accounted
NEOPersonality Inventory and has been shown to have
validity coefficients ranging from .50 to .70 for like factors.
for 10% of the variance in workplace accidents.
Additionally, Goldbergs five factors have been validated Thus, cognitive failure uniquely accounted for 7% of
against the Hogan Personality Inventory (see Goldberg, the variance in workplace accidents over and above

Table 1
Means, Standard Deviations, and Zero-Order Correlations for Study 1
Variable M SD 1 2 3 4 5
1. Cognitive failure 38.04 13.65 (.90)
2. Unsafe behavior 37.03 13.34 .31* (.83)
3. Accidents 2.73 5.74 .29* .20*
4. Conscientiousness 30.41 5.23 .32* .33* .17* (.85)
5. Transformed accidents 0.97 1.02 .35* .28* .87* .17*
Note. Reliabilities are presented in parentheses on the diagonal.
* p .05.
WORKPLACE SAFETY AND ACCIDENTS 321

Table 2 the interaction term, we found that the interaction of


Regression of Safety Behavior and Accidents on cognitive failure and conscientiousness was signifi-
Conscientiousness and Cognitive Failure cant for unsafe behaviors, t(189) 2.94, p .05, but
Variable t R2 R2 F
was only marginally significant for accidents, raw
t(189) 1.89, p .061; transformed t(189)
Accidents (raw) 1.72, p .09. These results partially supported
Step 1
C .17 2.32* .028 .028 5.40* Hypothesis 4 (see Table 3). Also see Figures 1 and 2.
Step 2 These results generally supported all proposed hy-
C .08 1.03 potheses. However, the analyses did not control for
CF .29 3.93* .101 .073 10.63* certain descriptive variables that might influence the
Unsafe behaviors
Step 1
results. For example, experience has been shown to
C .34 4.97* .116 .116 24.72* influence ones safety behavior (Frone, 1998). This
Step 2 could happen for several reasons such as prior in-
C .26 3.74* volvement in accidents and subsequent learning from
CF .24 3.35* .166 .05 18.65* such experiences (Rundmo, 1995) or the develop-
Accidents (transformed)
Step 1 ment of informal safety rules and procedures that
C .17 2.34* .03 .03 5.46* employees only develop with experience (Greenberg,
Step 2 1971). Similarly, age and gender have also been
C .06 0.77 suggested to differentially relate to occupational
CF .35 4.95* .14 .11 15.36*
safety. For example, Kingma (1994) pointed out that
Note. Degrees of freedom for Step 1 F(1, 191), t(191); most occupational accidents occur to those who are
for Step 2 F(2, 190), t(190). C conscientiousness; younger (i.e., under 19), and some recent empirical
CF cognitive failure.
* p .05. work has supported this suggestion. Specifically,
Barling, Loughlin, and Kelloway (2002) reported a
weak negative correlation between age and work
injuries and safety-related events. With regard to
gender, several significant relationships with acci-
conscientiousness using the raw accident data (see
Table 2). When examining the transformed data, the
same pattern of results were obtained, F(2, 190)
15.4, R2 .11, p .05.
A hierarchical regression was also conducted to Table 3
examine the utility of cognitive failure in the predic- Regression of Interaction on Safety Outcomes:
tion of unsafe work behaviors. The first step in the Study 1
analysis regressed unsafe work behaviors on consci- Variable t R2 R2 F
entiousness. This model was found to be significant, Accidents (raw)
F(1, 191) 24.7, p .05. On the second step, Step 1
cognitive failure was added to the equation. The C .08 1.03
model was found to be significant, F(2, 190) 18.6, CF .29 3.93* .10 .10 10.63*
p .05, accounting for 16% of the variance in unsafe Step 2
C CF .63 1.89 .12 .02 8.64*
work behaviors with cognitive failure uniquely ac- Unsafe behaviors
counting for 5% of the variance. These results sup- Step 1
ported Hypothesis 3. C .26 3.74*
Lastly, Hypothesis 4 was tested. However, we CF .24 3.35* .15 .15 18.65*
Step 2
centered the variables before creating the interaction C CF .93 2.94* .19 .04 15.82*
terms to control for any spurious effects (Bell & Accidents (transformed)
Kozlowski, 2002; Cohen & Cohen, 1983). At the first Step 1
step of the moderated regression analysis, cognitive C .06 4.95*
failure and conscientiousness were entered into the CF .35 0.78 .13 .13 15.36*
Step 2
equation, which proved to be significant predictors of C CF .52 1.72 .15 .02 11.17*
raw accidents (p .05), transformed accidents (p
Note. Degrees of freedom for Step 1 F(2, 190), t(190);
.05), and unsafe behaviors (p .05). At the second for Step 2 F(3, 189), t(189). C conscientiousness;
step, the interaction term was entered into the equa- CF cognitive failure.
tion. However, when examining the significance of p .10. * p .05.
322 WALLACE AND VODANOVICH

dents have been found (Barling, Kelloway, & Iverson,


2003; Iverson & Erwin, 1997), with men tending to
be involved in more accidents than women (Frone,
1998). Although these relationships are important to
consider, they are not a primary focus of the present
study. However, we felt that these variables needed to
be controlled for in follow-up analyses to more fully
ascertain if the relationships found in this study still
exist after the inclusion of certain control variables.
Therefore, we conducted follow-up analyses con-
trolling for age, gender, and job experience in assess- Figure 2. Effects of Consciousness Cognitive Failure
(CF) on work accidents in Study 1.
ing Hypotheses 3 and 4. Hypothesis 3 was still sup-
ported when controlling for age, gender, and job
experience in predicting raw accidents (p .05),
transformed accidents (p .05), and unsafe behav-
iors (p .05). Cognitive failure significantly ac-
and Graziano (1996) in their examination of consci-
counted for 8% of the variance in raw accidents, 7%
entiousness and accident involvement. However,
in transformed accidents, and 5% of the variance in
only partial support was found for Hypothesis 4. That
unsafe behaviors over and above all other variables.
is, the moderating effect of cognitive failure and
The same pattern of results was also obtained when
conscientiousness was only moderately significant
reexamining Hypothesis 4. Specifically, the interac-
for accidents. However, tests for moderation are no-
tion term was found to be significant for unsafe
toriously low on power, and the sample used was not
behaviors (p .05). As before, the interaction term
very large. This was part of the impetus for conduct-
for raw accidents (p .09) and transformed accident
ing Study 2 that used a larger sample.
data (p .10) was only marginally significant.
As with previous studies investigating accidents,
one possible shortcoming of Study 1 is the use of
Discussion self-reports of accidents. However, as stated previ-
ously, several researchers have found empirical evi-
The results provide support that cognitive failure is
dence that self-reports of accidents are significantly
related to safety behavior and workplace accidents.
related to more objective reports of accidents. Addi-
This was still the case when controlling for a number
tionally, Grunberg, Moore, and Greenberg (1996)
of variables (i.e., age, gender, experience, and con-
and Barling et al. (2003) have argued adamantly that
scientiousness). In reaching this hypothesis, we
workers have no basis to falsify injury reports and
found that cognitive failure is positively related to
that it is possible that objective organizational reports
workplace accidents and unsafe behaviors, whereas
may systematically underestimate the occurrence of
conscientiousness is negatively related to unsafe
such accidents. However, research is needed to in-
work behaviors and accidents. This lends support for
vestigate this possible limitation. For instance, it is
the results found by Cellar et al. (2001) and Arthur
possible that people who observe themselves better
and describe themselves more realistically (e.g.,
higher propensity for cognitive failure) may also re-
port more areas in which they are not conscientious.
Thus, they may perceive a higher rate of accidents
and unsafe behaviors as much as they perceive their
own mistakes.1 Thus, Study 2 used more objective
reports of these outcomes in an effort to overcome
such limitations. Additionally, Study 2 was also con-
ducted to overcome the possibility of contamination.
That is, our treatment and assessment of minor acci-
dents (e.g., delayed completion of work, first aid-
based medical attention) might be contaminated by

Figure 1. Effects of Consciousness Cognitive Failure 1


We thank an anonymous reviewer for suggesting this
(CF) on unsafe work behaviors in Study 1. alternative explanation.
WORKPLACE SAFETY AND ACCIDENTS 323

certain CFQ items that inquire about physical mis- Results


takes and blunders (e.g., do you drop things?).
To assess the hypotheses presented in the present
article, we used the same analytic procedures as
Study 2 described in Study 1. Again, the data were centered
before creating the interaction term to control for any
This study was designed to replicate the findings in spurious effects (Bell & Kozlowski, 2002; Chen,
Study 1 by retesting the four hypotheses presented Gully, Whiteman, & Kilcullen, 2000; Cohen & Co-
using a different sample. Additionally, this study hen, 1983). Using the centered data, we obtained the
used supervisor and organizational reports of unsafe same pattern of results in the present study as in
behaviors and accidents to overcome the limitation of Study 1 (see Table 4) except that full support was
self-reported outcomes used in Study 1. obtained for all hypotheses. Specifically, a positive
relationship emerged between cognitive failure and
accidents (r .23, p .05) and unsafe work behav-
Method iors (r .32, p .05), whereas a negative relation-
ship emerged between conscientiousness and acci-
Participants dents (r .14, p .05) and unsafe work behaviors
A sample of 263 enlisted military personnel participated (r .29, p .05).
in this study. Participants were machinists/mechanics who Additionally, we replicated the results found
worked on aviation equipment (e.g., airplanes, helicopters). regarding the uniqueness of cognitive failure in
Within the sample, 208 participants were male with an
accounting for accidents, F(2, 233) 10.47, p
average age of 24.26 years (SD 5.91). The sample con-
sisted of 89% Caucasian, 4% African American, 3% His- .001, with cognitive failure uniquely accounting
panic, 2% Asian, and 2% other or nondisclosed. Addition- for 6.2% of the variance over and above conscien-
ally, 78% of the sample believed that safety was a major tiousness (see Table 5). We also replicated the
concern in the workplace, 11% felt it was a moderate
concern, and 10% felt it was of no concern. Also, during results regarding unsafe work behaviors in that
data collection we were informed that safety had been a cognitive failure uniquely accounted for 6.7% of
serious concern in this area of work for quite some time, the variance in unsafe work behaviors, F(2, 233)
which is congruent with the 78% believing that safety was 20.23, p .001.
a major concern in their work.
To test the interaction between cognitive failure
and conscientiousness, we again used hierarchical
regression. We found that cognitive failure moder-
Materials and Procedure
ated the relationship between conscientiousness and
The same materials used in Study 1 were used in the unsafe work behaviors, t(232) 1.95, p .05; see
present study. However, the general measure of unsafe work Table 6). Additionally, the interaction was found to
behaviors (Hoffman & Stetzer, 1996) was used to facilitate be a significant predictor of accidents, t(232)
data collection. That is, supervisors were requested to com-
plete the survey to overcome the problem of common 2.68, p .05. These results fully support Hypoth-
source variance associated with self-reports. The measure of esis 4 in that the negative relationships between con-
unsafe work behaviors ( .82) contains six items that are scientiousness and accidents and unsafe work behav-
found on the complete 29-item version but rather pertain to
iors were stronger at a higher propensity for cognitive
unsafe behaviors in general (e.g., Generally rushing
through a job due to time pressure thereby increasing risk to failure than at lower levels (see Figures 3 and 4).
yourself). Additionally, immediate supervisors (noncom- Similar to Study 1, we conducted follow-up anal-
missioned officers) were requested to report the number of yses to determine if the results would hold after
recordable injuries each worker had on his or her record for
the past 2 years (i.e., 2000 2001). Eleven supervisors com-
considering additional variables. Again, we con-
pleted the safety ratings and provided reports of accidents. trolled for age, gender, and job experience. Hypoth-
Accidents in Study 2 were more consistent with Zohars esis 3 was still supported for accidents (p .05)
(2000, 2002) conceptualization of microaccidents (i.e., ac- and unsafe behaviors (p .05), with cognitive
cidents requiring at least minor medical attention). Such
accidents were categorized as major accidents in Study 1. failure accounting for 7% of the variance in accidents
Unlike Study 1, a test of the distribution of microaccidents and unsafe behaviors over and above all other vari-
revealed that it was uniformly distributed. Each worker ables. Hypothesis 4 was also supported in the fol-
separately completed the Conscientiousness scale of Sauc-
low-up analyses. The interaction term was still sig-
iers (1994) FFM measure and the CFQ. Complete results
were available from 236 workersupervisor pairs, which nificant for unsafe behaviors (p .05) and accidents
equates to a response rate of 89%. (p .05).
324 WALLACE AND VODANOVICH

Table 4
Means, Standard Deviations, and Zero-Order Correlations for Study 2
Variable M SD 1 2 3 4
1. Cognitive failure 38.26 13.4 (.90)
2. Unsafe behavior 10.81 4.1 .32* (.82)
3. Accidents 2.23 0.5 .23* .21*
4. Conscientiousness 30.30 5.3 .28** .29* .14* (.81)
Note. Reliabilities are presented in parentheses on the diagonal.
* p .05. ** p .01.

Discussion mitting cognitive failures and might have a rigid


attentional focus that creates a cognitive management
Empirical support was found for the moderating style that is inflexible. Such an inflexible cognitive
role of cognitive failure in the relationship between management style might allow for greater occur-
conscientiousness and unsafe work behaviors and rences of cognitive failure in dealing with intervening
accidents. Specifically, the impact of cognitive fail- and concurrent stimuli. In the present studies, it is
ure on accidents and unsafe behaviors is much stron- possible that mechanical and production workers
ger for those who score low on conscientious. In with high levels of cognitive failure might be prone
other words, cognitive failure scores are more pre- to commit a mistake in their work when faced with
dictive of workplace safety problems when individ-
new stimuli as these stimuli exceed their ability to
uals are relatively deficient in conscientiousness. This
attend to a given operation, especially when the task
finding may suggest that personality tendencies such
has been automated (Heckhausen & Beckmann, 1990).
as cognitive failure should be considered when exam-
It has been noted that interfering stimuli could be
ining models of accident occurrence and prevention.
the result of self-regulatory strategies that consume
These results are supportive of the findings reported by
Cellar et al. (2001) in their examination of the rela- cognitive resources in such a fashion that it does not
tionship between accidents and conscientiousness. permit the successful completion of the behavior
Additionally, Study 2 confirmed the results of (Kanfer, Ackerman, Murtha, Dugdale, & Nelson,
Study 1 using different criteria measurements. How- 1994). This suggests that people prone to committing
ever, it would be useful for future studies to identify cognitive failure may engage in more off-task pro-
additional mediating and moderating variables (e.g., cesses when executing tasks, which could be a me-
negative affect, boredom) that may facilitate or diating link between ones propensity for cognitive
hinder other such relationships. The relationships failures and safety-related outcomes. Specifically, it
identified in the present study suggest a moderate link is believed that more distal traits (e.g., conscientious-
among conscientiousness, cognitive failure, safety ness and cognitive failure) relate to outcomes such as
behavior, and accidents. To reiterate, Reason (1988) accidents and unsafe behaviors through more proxi-
stated that certain individuals are more prone to com- mal states (e.g., off-task behaviors). It could also be

Table 5
Regression of Safety Behavior and Accidents on Conscientiousness and Cognitive Failure
Accidents Unsafe behaviors
Variable t R 2
R 2
F t R2 R2 F
Step 1
C .16 2.39* .026 .026 5.73* .30 4.67* .092 .092 21.86*
Step 2
C .08 1.12 .21 3.26*
CF .26 3.85* .088 .062 10.47* .27 4.12* .159 .067 20.23*
Note. Degrees of freedom for Step 1 F(1, 234), t(234); for Step 2 F(2, 233), t(233). C conscientiousness; CF
cognitive failure.
* p .05.
WORKPLACE SAFETY AND ACCIDENTS 325

Table 6
Regression of Interaction on Safety Behavior and Accidents for Study 2
Accidents Unsafe behaviors
Variable t R2 R2 F t R2 R2 F
Step 1
C .27 4.12* .08 3.85*
CF .21 3.27* .159 .159 20.23* .26 1.12 .09 10.47*
Step 2
C CF .17 2.68* .186 .027 16.26* .13 1.95* .103 .015 8.3*
Note. Degrees of freedom for Step 1 F(2, 233), t(233); for Step 2 F(3, 232), t(232). C conscientiousness; CF
cognitive failure.
* p .05.

that constructs such as boredom or fatigue (Wallace that those with a tendency to be more absentminded
& Vodanovich, 2003) could greatly impact ones or scattered (i.e., higher rate of cognitive failures)
safety behavior regardless of personality, motiva- were those likely to be involved in more accidents.
tional, or cognitive tendencies, especially in highly An additional limitation is that the sample contained
monotonous work environments or jobs that require a large portion of men. Future research should exam-
sustained attention. Future studies of this type may ine the relationships found in the present studies
help to reduce the frequency of on-the-job accidents using more diverse samples that are more represen-
and prevent injuries, especially in highly safety-sen- tative of the general population as opposed to the
sitive jobs such as air traffic controllers, truck drivers, military, which is typically male dominated.
and airline pilots. Similarly, people scoring high on
measures of this nature may benefit from possible Conclusion
interventions aimed at reducing a rigid cognitive
management style, thus possibly allowing these indi- The results of the present investigation have
viduals to be better suited to handle novel stimuli in greatly expanded the work of other researchers in-
familiar tasks and environments. However, before vestigating workplace safety. However, further in-
using such an approach, future research should be vestigations are needed that identify additional me-
conducted to provide researchers and practitioners diating and boundary conditions that may affect
alike a better understanding of cognitively based mis- accident rates as the effect sizes in the present study
takes and their relationship to accidents. are moderate and allow room for additional con-
Finally, it is important to note limitations that exist structs to be included when predicting occupational
in Study 2. One possible limitation are the ratings safety. For instance, similar to the research on auto-
provided by supervisors. It is possible that ratings mobile accidents, it is possible that other personality
were subject to halo errors by supervisors assuming traits and individual differences, such as Type A
behavior pattern (Magnavita et al., 1997), may mod-

Figure 3. Effects of Consciousness Cognitive Failure Figure 4. Effects of Consciousness Cognitive Failure
(CF) on unsafe work behaviors in Study 2. (CF) on work accidents in Study 2.
326 WALLACE AND VODANOVICH

erate this relationship. Additionally, there is the issue Cohen, J., & Cohen, P. (1983). Applied multiple regression/
of temporal order of the data in the present research. correlation analysis for the behavioral sciences (2nd
ed.). Hillsdale, NJ: Erlbaum.
That is, we used currently collected data to predict
Costa, P. T., & McCrae, R. R. (1992). NEO PI-R profes-
archival data, which does not allow for strong causal sional manual. Odessa, FL: Psychological Assessment
inferences. It would be beneficial for future research Resources.
to collect temporally ordered data to overcome this Fallon, J. D., Avis, J. M., Kudisch, J. D., Gornet, T. P., &
limitation. Regardless, the present study has appre- Frost, A. (2000). Conscientiousness as a predictor of
productive and counterproductive behaviors. Journal of
ciably expanded earlier studies in safety research
Business and Psychology, 15, 339 349.
(e.g., Larson & Merritt 1991; Reason, 1984) and Frone, M. R. (1998). Predictors of work injuries among
confirmed the findings of others (e.g., Wallace & employed adolescents. Journal of Applied Psychology,
Vodanovich, 2003) by demonstrating the ability of 83, 565576.
cognitive failure and conscientiousness to predict un- Goldberg, L. R. (1992). The development of markers for the
big-five factor structure. Psychological Assessment, 4,
safe work behaviors and workplace accidents using 26 42.
both self-report and objective criteria. Additionally, Greenberg, L. (1971). Learning as a factor in accident
this research used two distinct samples, which adds experience. Personnel Psychology, 24, 7176.
support and stability for the external validity of the Grunberg, L., Moore, S., & Greenberg, E. (1996). The
results. We hope other researchers will be able to use relationship of employee ownership and participation to
workplace safety. Economic and Industrial Democracy,
this study as a stepping stone to gain further insight 17, 221241.
into work safety and help continue to reduce the Hale, A. R., & Hale, M. (1972). A review of the industrial
number of workplace accidents. accident literature. London: Her Majestys Stationery
Office.
References Hansen, C. P. (1988). Personality characteristics of the
accident involved employee. Journal of Business and
Psychology, 2, 346 365.
Arthur, W., Barrett, G. V., & Alexander, R. A. (1991).
Hansen, C. P. (1989). A causal model of the relationship
Prediction of vehicular accident involvement: A meta-
analysis. Human Performance, 4, 89 105. among accidents, biodata, personality, and cognitive fac-
Arthur, W., & Graziano, W. G. (1996). The five-factor tors. Journal of Applied Psychology, 74, 8190.
model, conscientiousness, and driving accident involve- Hechanova-Alampay, R., & Beehr, T. A. (2001). Empow-
ment. Journal of Personality, 63, 593 618. erment, span of control, and safety performance in work
Barling, J., Kelloway, E. K., & Iverson, R. D. (2003). High teams after workforce reduction. Journal of Occupa-
quality work, job satisfaction, and occupational injuries. tional Health Psychology, 6, 275 -282.
Journal of Applied Psychology, 88, 276 283. Heckhausen, H., & Beckman, J. (1990). Intentional action
Barling, J., Loughlin, C., & Kelloway, E. K. (2002). De- and action slips. Psychological Review, 97, 36 48.
velopment and test of a model linking safety-specific Hoffman, D. A., & Stetzer, A. (1996). A cross-level inves-
transformational leadership and occupational safety. tigation of factors influencing unsafe behaviors and ac-
Journal of Applied Psychology, 87, 488 496. cidents. Personnel Psychology, 49, 306 339.
Barrick, M. R., & Mount, M. K. (1991). The big five Hough, L. M., & Schneider, R. J. (1996). Personality traits,
personality dimensions and job performance: A meta- taxonomies, and applications in organizations. In K. R.
analysis. Personnel Psychology, 44, 126. Murphy (Ed.), Individual differences and behavior in
Barrick, M. R., Stewart, G. L., & Piotrowski, M. (2002). organizations (pp. 31 88). San Francisco: Jossey-Bass.
Personality and job performance: Test of the mediating Iverson, R. D., & Erwin, P. J. (1997). Predicting occupa-
effects of motivation among sales representatives. Jour- tional injury: The role of affectivity. Journal of Occupa-
nal of Applied Psychology, 87, 4351. tional and Organizational Psychology, 70, 113128.
Bell, S. B., & Kozlowski, S. W. (2002). Goal orientation Kanfer, R., Ackerman, P. L., Murtha, T. C., Dugdale, B., &
and ability: Interactive effects on self-efficacy, perfor- Nelson, L. (1994). Goal setting, conditions of practice,
mance, and knowledge. Journal of Applied Psychology, and task performance: A resource allocation perspective.
87, 497505. Journal of Applied Psychology, 79, 826 835.
Broadbent, D. E., Cooper, P. F., Fitzgerald, P., & Parkes, Kingma, J. (1994). Causes of occupational injuries. Percep-
K. R. (1982). The Cognitive Failures Questionnaire tual and Motor Skills, 79, 10251026.
(CFQ) and its correlates. British Journal of Clinical Larson, G. E., Alderton, D. L., Neideffer, M., & Underhill,
Psychology, 21, 116. E. (1997). Further evidence on dimensionality and cor-
Cellar, D. F., Nelson, Z. C., Yorke, C. M., & Bauer, C. relates of the cognitive failures questionnaire. British
(2001). The five-factor model and safety in the work- Journal of Psychology, 88, 29 38.
place: Investigating the relationships between personality Larson, G. E., & Merritt, C. R. (1991). Can accidents be
and accident involvement. Journal of Prevention and predicted? An empirical test of the cognitive failures
Intervention in the Community, 22, 4352. questionnaire. Applied Psychology: An International Re-
Chen, G., Gully, S. M., Whiteman, J., & Kilcullen, R. N. view, 40, 37 45.
(2000). Examination of relationships among trait-like Maganavita, N., Narda, R., Sani, L., Carbone, A., Lorenzo,
individual differences, state-like individual differences, G. D., & Sacco, A. (1997). Type-A behavior pattern and
and learning performance. Journal of Applied Psychol- traffic accidents. British Journal of Medical Psychology,
ogy, 85, 835 847. 70, 103107.
WORKPLACE SAFETY AND ACCIDENTS 327

Manly, T., Robertson, I. H., Galloway, M., & Hawkins, K. Saucier, G. (1994). Mini-Markers: A brief version of Gold-
(1999). The absent mind: Further investigations of sus- bergs unipolar big-five markers. Journal of Personality
tained attention to response. Neuropsychologia, 37, Assessment, 63, 506 516.
661 670. Stewart, G. L. (1999). Trait bandwidth and stages of job
Martin, M. (1983). Cognitive failure: Everyday and labora- performance: Assessing differential effects for conscien-
tory performance. Bulletin of Psychonomic Society, 21, tiousness and its subtraits. Journal of Applied Psychol-
97100. ogy, 84, 959 968.
Mihal, W., & Barrett, G. V. (1976). Individual differences Tziner, A., Murphy, K. R., & Cleveland, J. N. (2002). Does
in perceptual information processing and their relation to conscientiousness moderate the relationship between at-
automobile accident involvement. Journal of Applied titudes and beliefs regarding performance appraisal and
Psychology, 61, 229 233. rating behavior? International Journal of Selection and
Mooradian, T. A., & Nezlek, J. B. (1996). Comparing the Assessment, 10, 218 224.
NEO-FFI and Sauciers Mini-Markers as measures of the Vom Hofe, A., Mainemarre, G., & Vannier, L. (1998).
big five. Personality and Individual Differences, 21, Sensitivity to everyday failures and cognitive inhibition:
213215.
Are they related? European Review of Applied Psychol-
National Safety Council. (1983). Accident facts 1983 edi-
ogy, 48, 49 55.
tion. Chicago: Author.
Wallace, J. C., Kass, S. J., & Stanny, C. (2001). Predicting
National Safety Council. (1999). Injury facts 1999 edition.
performance in go situations: A new use for the Cog-
Itasca, IL: Author.
Reason, J. T. (1977). Skill and error in everyday life. In M. nitive Failures Questionnaire? North American Journal
Howe (Ed.), Adult learning. London: Wiley. of Psychology, 3, 481 490.
Reason, J. T. (1979). Actions not as planned: The price of Wallace, J. C., Kass, S. J., & Stanny, C. (2002). Cognitive
automation. In G. Underwood & R. Stevens (Eds.), As- failures questionnaire revisited: Correlates and dimen-
pects of consciousness (pp. 67 89). London: Penguin. sions. Journal of General Psychology 129, 238 256.
Reason, J. (1984). Lapses of attention in everyday life. In R. Wallace, J. C., & Vodanovich, S. J. (2003). Can accidents
Parasuraman & D. R. Davies (Eds.), Varieties of atten- and industrial mishaps be predicted? Investigating work-
tion (pp. 127). London: Academic Press. place performance. Journal of Business and Psychology,
Reason, J. (1988). Stress and cognitive failure. In S. Fisher 17, 503514.
& J. Reason (Eds.), Handbook of life stress, cognition, Witt, L. A., Burke, L. A., Barrick, M. R., & Mount, M. K.
and health (pp. 405 421). New York: Wiley. (2002). The interactive effect of conscientiousness and
Reason, J. T., & Lucas, D. (1984). Absent-mindedness in agreeableness on job performance. Journal of Applied
shops: Its incidence, correlates and consequences. British Psychology, 87, 164 169.
Journal of Clinical Psychology, 23, 121131. Zohar, D. (2000). A group-level model of safety climate:
Robertson, I. H., Manly, T., Andrade, J., Baddeley, B. T., & Testing the effect of group climate on microaccidents in
Yiend, J. (1997). Oops!: Performance correlates of manufacturing jobs. Journal of Applied Psychology, 85,
everyday attentional failures in traumatic brain injured 587596.
and normal subjects. Neuropsychologia, 35, 747758. Zohar, D. (2002). Modifying supervisory practices to im-
Rundmo, T. (1995). Perceived risk, safety status, and job prove subunit safety: A leadership-based intervention
stress among injured and noninjured employees on off- model. Journal of Applied Psychology, 87, 156 163.
shore petroleum installations. Journal of Safety Re-
search, 26, 8797.
Received October 7, 2002
Salgado, J. F. (1997). The five factor model of personality
and job performance in the European community. Jour- Revision received June 6, 2003
nal of Applied Psychology, 82, 30 43. Accepted June 11, 2003 y

Das könnte Ihnen auch gefallen