Você está na página 1de 9

SPE 86622

Implicit Attitudes: A New Measure of Safety Culture


Calvin Burns, Kathryn Mearns & Peter McGeorge, Industrial Psychology Research Centre, University of Aberdeen

Copyright 2004, Society of Petroleum Engineers Inc.


culture in which safety is understood to be and accepted as,
This paper was prepared for presentation at The Seventh SPE International Conference on the number one priority. [1] There is considerable debate in
Health, Safety, and Environment in Oil and Gas Exploration and Production held in Calgary,
Alberta, Canada, 2931 March 2004. the literature on the definition, measurement and utility of the
This paper was selected for presentation by an SPE Program Committee following review of
concept of safety culture. [2] Many researchers have used
information contained in a proposal submitted by the author(s). Contents of the paper, as questionnaires to measure the workforces explicit attitudes
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any and perceptions about the state of safety on offshore
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at installations. [3-6] However, these questionnaires measure
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper safety climate which is the term used to describe the
for commercial purposes without the written consent of the Society of Petroleum Engineers is
prohibited. Permission to reproduce in print is restricted to a proposal of not more than 300 indicators of an organisations underlying safety culture. [5, 7]
words; illustrations may not be copied. The proposal must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Schein [8] argued that climate is only a surface manifestation
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435. of culture and that culture manifests itself in deeper levels of
unconscious assumptions. These unconscious aspects of
Abstract culture are less easy to measure simply because they are
unconscious and not directly accessible to individuals.
This paper describes two studies about the explicit and
implicit attitudes held by the workforce at UK gas plants about The concept of dual attitudes provides a way to
the trustworthiness of their workmates, supervisors and the measure the explicit attitudes that comprise the surface
plant leadership. Psychologists have shown that an individual features of safety culture and the unconscious assumptions
is able to hold different evaluations of the same object (a that characterise its core. Wilson, Lindsey & Schooler [9]
conscious, explicit attitude and an unconscious, implicit defined dual attitudes as different evaluations of the same
attitude) and that an individuals explicit and implicit attitudes object: a conscious, explicit attitude and an unconscious,
to something do not have to be the same. Organisational implicit attitude. Explicit attitudes are those attitudes that are
theorists have argued that climate, as measured by explicit constrained by conscious processing whereby individuals
attitudes, is only a surface manifestation of culture and that consider and state their attitude to attitude-objects.
culture manifests itself in deeper levels of unconscious Questionnaires like safety climate surveys measure explicit
assumptions. Thus, a model of dual attitudes about trust in attitudes. Implicit attitudes are traces of past experiences that
safety culture is propsed, which conceptualizes explicit can mediate favourable or unfavourable feeling, thought or
attitudes about trust as part of safety climate and implicit action toward social objects without conscious awareness.
attitudes about trust as part of the deeper levels of safety Compared to explicit attitudes, implicit can become active
culture. Attitudes about trust were investigated because relatively automatically and are more likely to guide
many researchers have proposed that trust is the foundation of behaviour under conditions of time pressure or heavy mental
an effective safety culture. Explicit attitudes about trust were workload. Wilson et al. [9] argued that an individuals
measured by a questionnaire while implicit attitudes were explicit and implicit attitudes to something do not necessarily
measured using a computerised task. The main finding of the have to be the same.
paper was that a different pattern of results emerged for
explicit and implicit attitudes about trust. The paper The current paper investigated attitudes about trust
concludes by explaining how the data support the proposed because many researchers have proposed that trust is the
model and poses questions for further research. foundation of an effective safety culture. [10-13] Before dual
attitudes about trust are considered, it is prudent to review the
Introduction theoretical importance of trust in safety culture.

In his report into the Piper Alpha disaster, Lord Cullen The generally accepted definition of trust and the one
stressed that It is essential to create a corporate atmosphere or which shall be used throughout this paper is a willingness to
accept vulnerability based upon the positive expectations of
2 SPE 86622

the intentions or behaviour of another [14-16]. The role of plant leadership shall be reported. These items were I trust
trust in developing and maintaining safe behaviour patterns in my workmates, I trust my supervisor, and I trust the plant
the workplace is made clear by considering the 3 sub- leadership. All questionnaire items used a 6-point scale with
components of a safety culture: a reporting culture, a just anchors of agree and disagree for each scale point.
culture and a learning culture [10]. With respect to a reporting
culture, it is important that the workforce believe that they will Computerised Task. The procedure for the computerised
not be penalised by the management or their fellow workers task was adapted from Fazio, Jackson, Dunton and Williams
for reporting accidents or near misses. Without this degree of [17] and consisted of 3 phases: a baseline phase, a priming
trust, a report will probably gloss over pivotal human factors phase and a recognition memory test.
information or no report may be made. An effective reporting
culture depends largely on how the organisation deals with Stimulus Materials. The computerised task was
blame and punishment and this is a component of a just conducted on an IBM compatible laptop computer using Super
culture. In a just culture, a clear line must exist between Lab version 1.03. All visual stimuli were presented on a
acceptable and unacceptable actions but in order to maintain single line in the centre of the screen in bold Times New
an atmosphere in which employees are willing to report their Roman 60 point font. Responses were made on a 4 button
accidents or near misses, they must trust the management to response box with keys labelled Trust, Distrust, Yes,
treat them fairly. Lastly, trust is an important part of a and No.
learning culture. Providing rapid and useful feedback when
incident reports are made should increase the workforce's trust The target words (words indicative of trust or
in the management through demonstrated commitment to distrust) used in this task were generated by 50 University of
safety. Further, it should reinforce the workforce's reporting Aberdeen first-year psychology students during a Group
behaviours and enhance their understanding of potential Participation Practical exercise. Collectively, these students
hazards. Clearly, trust is needed to engineer an effective generated 182 different words and phrases about trust and
safety culture. distrust. 50 different first-year students later rated these words
and phrases for trustworthiness (1 = total lack of trust, 2 =
Study 1 some lack of trust, 3 = not relevant, 4 = trust, 5 = high level of
trust). The 10 words with the highest trustworthiness ratings
This study consisted of two parts: a questionnaire and a were selected as targets indicative of trust and the 10 words
computerised task. The questionnaire measured explicit with the lowest trustworthiness ratings were selected as targets
attitudes about trust in participants' workmates, supervisors indicative of distrust. It should be noted that targets were
and the plant leadership while the computerised task measured selected so as not to include words with trust in them. As
their respective implicit attitudes about trust. The first author expected, the difference in trustworthiness between the groups
administered the questionnaire and computerised task onsite of targets was significant, t(49) = 62.048, p < 0.0001.
during one-to-one meetings with participants.

Participants. Sixty-two members (approximately 50%) of the Procedure. The purpose of the first phase was to
workforce at a UK gas plant hereafter referred to as Gas Plant obtain baseline data for the target adjectives. A single target
A, participated in this study. The first author randomly word was presented on the computer screen. Participants were
selected members of the workforce who were onsite during the instructed to press the key labelled Trust or the key labelled
three-week period in which the study was conducted for Distrust as quickly as possible to indicate their judgement of
participation in the study. Participants were selected from the the word. There were 20 targets; 10 of the words were
operations, maintenance, construction and gas plant services indicative of trust and the other 10 words were indicative of
job types. Twenty-eight of the participants were employed by distrust. Participants performed 2 blocks of trials. Each block
the operator company while the remaining 42 were employed consisted of the randomised presentation of the 20 words. The
by one of three different contractor companies. All average latency for any given word served as the participant's
participants were male as no women were employed in the job baseline latency for that word.
types sampled. Participants mean work experience in the oil
and gas industry was 17.91 years (S.D. = 8.22) and their The beginning of a trial was marked by the
average work experience at that particular gas plant was 9.12 presentation of a row of asterisks for 315 ms. Following this,
years (S.D. = 7.20). Participation in the study was completely the screen blanked for 135 ms before the onset of the target
voluntary and employees were given time away from their word. A given target remained on the screen until the
regular duties to take part. participant responded or for a maximum of 1.75 s. A 2.5 s
interval separated each trial. Participants responses and their
Questionnaire. Questionnaire items were based on the model response latencies (from target onset to key press to the
of organisational trust proposed by Mayer, Davis & nearest millisecond) were recorded. A block of practice trials
Schoorman [14]. For the purposes of this paper, only items preceded this phase. The purpose of the practice block was to
relating to generalised trust in workmates, supervisors and the
SPE 86622 3

familiarise the participants with the procedure by using between these mean scores were significant. There was a
different words than those presented in the first phase. strong trend suggesting that participants trusted their
workmates more than they trusted their supervisor, as the
The second phase involved a priming task. result, t(61) = 1.98, p = 0.052, approached significance at the
Participants were told that they would have to judge the 0.05 level. Participants trusted their workmates more than
meaning of the target words just like before, but that now they they trusted the plant leadership, t(59) = 4.50, p < 0.001 and
would also have to remember different job titles (primes) that they trusted their supervisor more than they trusted the plant
would flash before each target. Participants were told that it leadership, t(59) = 3.11, p = 0.003. Taken together, these
was important for them to pay attention to the primes because results suggest that group membership as part of the workforce
they had to complete a recognition memory test in the next influences trust for workmates, supervisors and the plant
phase. There were 10 primes in total. The 3 primes of interest leadership.
were workmates, supervisor and plant leadership. The
other 7 primes served as distractors. The procedure was Table 1: Means and Standard Deviations for Paired
similar to the first phase except that the row of asterisks was Samples of Questionnaire Items
now replaced by one of the 10 primes. On any given trial, a Pairs of Items n Mean S.D.
prime was presented for 315 ms, followed by a 135 ms I trust my workmates 62 5.032 0.128
I trust my supervisor 62 4.823 0.141
interval before the onset of the target; thus, the stimulus onset I trust my workmates 60 5.033 0.132
asynchrony was 450 ms. A 2.5 s interval separated each trial. I trust the plant leadership 60 4.267 0.154
I trust my supervisor 60 4.817 0.145
After a short practice block involving different I trust the plant leadership 60 4.267 0.154
primes and targets, 5 blocks of trials were presented. In each
of the 5 blocks, every prime was paired with one target Results: Computerised Task. Due to time constraints, 9 of
indicative of trust and one target indicative of distrust. Over the 62 participants were unable to take part in the
the course of this phase, the 3 critical primes were paired with computersied task. Therefore, the following data are based on
the same 5 targets indicative of trust and the same 5 targets a sample of 53 participants.
indicative of distrust. The distractor primes were paired with
targets such that each of the 20 targets only appeared once per Detection Data. During the priming phase,
block. Responses and response latencies (measured in participants were instructed to pay attention to the primes in
milliseconds from target adjective onset to response) were order to complete a recognition memory test in the next phase.
recorded. Accordingly, participants completed a detection task in which
they had to indicate whether a given job title had or had not
The third phase involved the recognition memory test been presented during the priming phase of the task. These
that the participants had been led to expect during the previous data were examined to check whether the participants paid
phase. The purpose of this phase was to determine whether attention to the primes, as doing so was required for implicit
the participants had followed the instructions to pay attention attitude activation. Detection scores were calculated for each
to the primes in the previous task. (Doing so was required for participant by subtracting the proportion of false alarms (filler
implicit attitude activation). Thus, this phase involved the not previously presented that the participant incorrectly
random presentation of the 10 primes used during the previous identified as having been presented) from the proportion of
task along with 10 new job titles (fillers) that were not correctly identified hits (previously presented job title
previously presented. Participants were instructed to press a correctly identified as such). Performance at chance levels
key labelled Yes if the job title was presented in the would be indicated by a score of 0 on this detection index.
previous task or press a key labelled No if the job title had Participants performed at better than chance levels; their mean
not been presented earlier. Each job title remained on the score was 0.76, which was statistically reliable; t(48) = 25.39,
screen for a maximum of 5 s or until the participant pressed a p < 0.001.
key. A 2.5 s interval separated each trial. Responses were
recorded. Facilitation Scores. For each participant, baseline
latency for each target was computed from the average of the
Results: Questionnaire. Table 1 displays the means and two presentations of the target during the first phase of the
standard deviations for pairs of items about trust for task. The latency for any given target when preceded by a
workmates, supervisors and the plant leadership. It should be given prime was subtracted from the baseline for that target to
noted that participants expressed a positive attitude about trust arrive at a facilitation score. Average trust facilitation scores
for their workmates, as the mean score was 5.03. Participants for every participant were then computed for the 3 critical
expressed a less positive attitude about trust for their primes. The average of all participants trust facilitation
supervisors, as the mean score was 4.82 and they expressed a scores for the 3 critical primes are displayed in Figure 1. The
slightly less positive attitude still about trust for the plant latencies for any trials on which participants made an error,
leadership, as the mean score was 4.27. Paired samples t-tests which averaged 5.53% were not included in these
were conducted in order to test whether the differences computations.
4 SPE 86622

The data in Figure 1 should be interpreted as follows. score that is negative indicates the lack of an implicit
A mean trust facilitation score that is positive indicates an association of the prime with trust, as the response latencies
implicit association of the prime with trust, as the response when preceded by that prime were longer than the no-prime
latencies when preceded by that prime were quicker than the baselines.
no-prime baselines. Conversely, a mean trust facilitation
Figure 1: Mean Facilitation Scores from the Computerised Task Gas Plant A
(Error Bars represent the Standard Error of the Mean)

100

80 51.1759
Facilitation Score (ms)

60

40
9.7189
Facilitation Score; Trust
20

0
Workmates Supervisor Plant Leadership
-20

-40 -7.4415

Prime

In order to determine the strength of participants Table 2: Means and Standard Deviations for Paired
implicit attitudes about trust, one-sample t-tests were Samples of Facilitation Scores
conducted on each of the facilitation scores for each prime to Pairs of Facilitation Scores Mean t df p
determine whether there was a significant difference from the Trust Workmates 51.18 -2.073 52 0.043
Trust Supervisor 9.72
baseline (i.e. a facilitation score of 0). The trust facilitation Trust Workmates 51.18 -2.388 52 0.021
score for workmates was the only facilitation score that was Trust Plant Leadership -7.44
significantly different from the baseline, t(52) = 2.05, p = Trust Supervisor 9.72 1.031 52 0.307
0.045. These results suggest that participants had a strong Trust Plant Leadership -7.44
implicit attitude about trust for their workmates but not for
their supervisors or the plant leadership. As seen in Table 2, the trust facilitation score for
workmates was stronger than that for supervisor and the plant
Paired samples t tests were also conducted in order to leadership. The difference between the trust facilitation score
test for differences in facilitation scores for trust between for supervisor and the plant leadership was not significant.
specific levels of the organisational hierarchy. Table 2
displays the means, t statistic and p value for each pair of In order to establish whether implicit and explicit
facilitation scores. attitudes about trust are separate constructs, correlation
analyses were conducted between the trust facilitation scores
and questionnaire items about trust for workmates, supervisor
and the plant leadership, respectively. Table 3 presents the
correlation coefficients and p values from these analyses. No
relationship was observed between the explicit and implicit
attitude measures of trust, respectively, for workmates,
supervisor and the plant leadership. These results are
SPE 86622 5

consistent with the dual attitudes literature and suggest that [10] proposed that trust is the foundation of an effective safety
explicit and implicit attitudes about trust are separate culture but did not differentiate between explicit and implicit
psychological constructs. attitudes about trust. Schein [8] argued that climate, as
determined by espoused values and beliefs, is only a surface
Table 3: Correlation Coefficients and p Values for manifestation of culture and that deeper levels of culture are
Explicit and Implicit Attitude Measures of Trust characterised by underlying and usually unconscious
Trust Facilitation Score / r p assumptions that determine perceptions, thought processes,
Questionnaire Item feelings and behaviour (p. 112). Accordingly, many
Trust workmates -0.015 0.916
I trust my workmates researchers regard explicit attitudes and perceptions about
Trust supervisor -0.044 0.755 safety as part of safety climate, which serves as an indicator of
I trust my supervisor the organisations underlying safety culture [5, 7]. A more
Trust plant leadership 0.081 0.571 complete model of safety culture is thus proposed which
I trust the plant leadership
conceptualises explicit attitudes about trust as part of safety
climate and implicit attitudes about trust as part of the deeper
Discussion. The results of the questionnaire suggest that levels of safety culture.
group membership within the organisational hierarchy had a
significant influence on explicit attitudes about trust for other It is interesting to note that the implicit and explicit
individuals and groups within the hierarchy. In particular, attitudes about trust reported in this study produced very
members of the workforce indicated that they trust their different patterns of results. The trust facilitation score for
workmates more than their supervisors and the plant workmates was significantly greater than the baseline but the
leadership. As evidenced in Table 1, employees also indicated trust facilitation scores for supervisor and the plant leadership
that they trust their supervisors more than the plant leadership. were not. These results suggest that participants had strong
These results are consistent with the predictions made by implicit attitudes about trust for their workmates but not for
social identity theory. [18] Social identity theory states that their supervisors or the plant leadership. In sharp contrast,
individuals tend to associate positive beliefs and feelings with participants reported positive explicit attitudes about trust for
the groups to which they belong (i.e. an ingroup) but the their workmates, supervisors and the plant leadership. It
beliefs and feelings that individuals associate with other should be stressed that Wilson et al.s [9] model of dual
groups (i.e. outgroups) may be positive, negative, or neutral. attitudes does not preclude one from holding a positive
Members of the workforce trusted their workmates more than explicit attitude and a positive implicit attitude toward the
their supervisors and the plant leadership because the latter are same thing. Dual attitudes are proposed to develop through
members of an outgroup (i.e. the management). However, explicit, conscious and implicit, unconscious cognitive
employees trusted their supervisors more than the plant systems that give rise to associative evaluations
leadership because their supervisors are part of the same work independently. Thus, in terms of the newly proposed model,
team (i.e. an ingroup) whereas the plant leadership are the results are not necessarily indicative of a good safety
members of the senior management (i.e. an outgroup). Thus, culture. In a good safety culture, strong positive explicit and
the effect of group membership on trust may be an artefact of implicit attitudes about trust for workmates, supervisors and
any hierarchical social system. the plant leadership should be expressed.
In order to assess safety culture more completely, the Study 2
current study measured dual attitudes about trust for
workmates, supervisors and the plant leadership. Correlation The purpose of this study was to gather wider support for the
analyses (See Table 3) revealed no relationship between data presented in Study 1. It should also be noted that the
implicit and explicit attitude measures about trust for computerised task was adapted from the method described in
workmates, supervisor and the plant leadership, respectively. Study 1 in an attempt to reduce the variability in the
These results suggest that implicit and explicit attitudes about facilitation score data and produce a more valid measure of
trust are separate constructs. If implicit and explicit attitudes implicit attitudes about trust.
about trust were the same psychological construct, then a
strong positive relationship should have emerged between The current study consisted of two parts: a
each of the trust facilitation scores and the questionnaire items questionnaire and a computerised task. The questionnaire
about trust because these measures would have been measured participants explicit attitudes about trust in their
measuring the same construct, just in different ways. workmates, supervisors and the plant leadership while the
However, this was not the case and the results are consistent computerised task measured their respective implicit attitudes
with the predictions made by Wilson et al.s [9] model of dual about trust. The study was conducted on site by the first
attitudes. author. Each participant completed his questionnaire in groups
of no more than 10 people. The computerized task was then
The concept of dual attitudes about trust has the conducted on a one-to-one basis with the first author.
potential to extend current models of safety culture. Reason
6 SPE 86622

Participants. Twenty members (approximately 20%) of the After a short practice block involving different
workforce at a UK gas plant hereafter referred to as Gas Plant primes and targets, 10 blocks of trials were presented. Each
B, participated in both parts of this study. block consisted of 10 randomised trials, in which each of the
primes appeared twice, followed by one of the 20 target
The 20 participants were drawn from all job types words. In order for each prime to be followed by all 20 target
except operations. Three of the participants were employed words in a systematic manner, trust targets were paired
by the operator company while the remaining 17 were randomly with distrust targets to yield 10 pairs of target
employed by one of three different contractor companies. words. The 10 pairs of target words were then paired with
Eighteen of the participants were male and two were female. each of the primes such that over the course of the 10 blocks,
Participants mean work experience in the oil and gas industry each pair of target words followed each prime but did not
was 13.15 years (S.D. = 10.70) and their average work appear more than once per block. It was presumed that
experience at the current plant was 4.97 years (S.D. = 4.07). pairing each prime with 10 trust targets and 10 distrust targets
Participation in the study was completely voluntary and would produce more valid measures of implicit attitudes about
employees were given time away from their regular duties to trust than was offered by the previous study. However, due to
take part. time constraints, the current study only used five primes in
order to keep the number of trials the same as in the previous
Questionnaire. Questionnaire items were based on the model study.
of organisational trust proposed by Mayer, Davis &
Schoorman [14]. For the purposes of this paper, only items The third phase involved the recognition memory test
relating to generalised trust in workmates, supervisors and the that participants had been led to expect during the previous
plant leadership shall be reported. These items were I trust phase. The purpose of this phase was to determine whether
my workmates, I trust my supervisor, and I trust the plant participants had followed the instructions to pay attention to
leadership. All questionnaire items used a 6-point scale with the primes in the previous task. Thus, this phase involved the
anchors of agree and disagree for each scale point. random presentation of the five primes used during the
previous task along with five new job titles (fillers) that were
Computerised Task. The procedure for the computerised not presented previously. Participants were instructed to press
task was adapted from the previous study and consisted of 3 the key labelled Yes if they thought the job title was
phases: a baseline phase, a priming phase and a recognition presented in the previous task or press the key labelled No if
memory test. they thought the job title had not been presented earlier. Each
job title remained on the screen for a maximum of 5.00 s or
Stimulus Materials. The stimulus materials were until the participant pressed a key. A 2.50 s interval separated
exactly as per Study 1. each trial. Responses were recorded.

Procedure. The baseline phase in the current study Results: Questionnaire. Participants expressed a positive
was conducted exactly like the baseline phase described in attitude about trust for their workmates, as the mean score was
Study 1. The purpose of this phase was to obtain baseline data 4.70. Participants expressed a less positive attitude about trust
for the target words. for their supervisors, as the mean score was 4.42 and they
expressed a slightly less positive attitude still about trust for
The second phase involved a priming task. the plant leadership, as the mean score was 4.37. A repeated
Participants were told that they would have to judge the measures analysis of variance, with 3 within subject factors,
meaning of the target words just like before, but now they revealed that there was no significant difference between
would also have to remember different job titles (primes) that participants trust for their workmates, supervisors and the
would flash before each target word. Participants were told plant leadership. This finding is contrary to the pattern of
that it was important for them to pay attention to the primes results reported in Study 1, which supported social identity
because they had to complete a recognition memory test in the theory. The fact that participants in this study did not trust
next phase. There were five primes in total. The three critical their workmates more than their supervisors and did not trust
primes were workmates, supervisor and plant their supervisors more than the plant leadership may be due to
leadership. The other two primes served as distractors. The the small, unrepresentative sample of the workforce.
procedure was similar to the first phase except that the row of
asterisks was now replaced by one of the five primes. On any Results: Computerised Task. The results of the
given trial, a prime was presented for 315 ms, followed by a computerized task are as follows.
135 ms interval before the onset of the target; thus, the
stimulus onset asynchrony was 450 ms. Responses and Detection Data. During the priming phase,
response latencies (measured in milliseconds from target word participants were instructed to pay attention to the primes in
onset to response) were recorded. A 2.50 s interval separated order to complete a recognition memory test in the next phase.
each trial. Accordingly, participants completed a detection task in which
they had to indicate whether a given job title had or had not
SPE 86622 7

been presented during the priming phase of the task. These Table 5: Means, t Statistic and p Value for Pairs of
data were examined to check whether the participants paid Facilitation Scores
attention to the primes, as doing so was required for implicit Pairs of Facilitation Scores Mean t df p
attitude activation. Detection scores were calculated for each Trust Workmates 94.679 1.177 19 0.254
participant by subtracting the proportion of false alarms (filler Trust Supervisor 79.600
not previously presented that the participant incorrectly Trust Workmates 94.679 2.832 19 0.011
identified as having been presented) from the proportion of Trust Plant Leadership 48.984
Trust Supervisor 79.600 3.111 19 0.006
correctly identified hits (previously presented job title Trust Plant Leadership 48.984
correctly identified as such). Performance at chance levels
would be indicated by a score of 0 on this detection index. As seen in Table 5, the trust facilitation score for
Participants performed at better than chance levels; their mean workmates was greater than that for the plant leadership, p =
score of 0.89, was significantly greater than the chance score 0.011 but not greater than that for supervisor. These findings
of 0; t(19) = 32.91, p < 0.001. indicate that participants had stronger implicit attitudes about
trust for their workmates than they did for the plant leadership
Facilitation Scores. For each participant, baseline but not for their supervisors. Also, the trust facilitation score
latency for each target was computed from the average of the for supervisor was greater than that for the plant leadership, p
two presentations of the target during the first phase of the = 0.006. This finding indicates that participants had stronger
task. The latency for any given target when preceded by a implicit attitudes about trust for their supervisors than they did
given prime was subtracted from the baseline for that target to for the plant leadership.
arrive at a facilitation score. Average trust facilitation scores
for every participant were then computed for the 3 critical In order to provide further support that implicit and
primes. The average of all participants trust facilitation explicit attitudes about trust are separate constructs,
scores for the 3 critical primes are displayed in Figure 2. The correlation analyses were conducted between the trust
latencies for any trials on which participants made an error, facilitation scores and questionnaire items about trust for
which averaged 1.95% were not included in these workmates, supervisor and the plant leadership, respectively.
computations. Table 6 presents the correlation coefficients and p values from
these analyses. No strong positive relationship was observed
The data in Figure 2 should be interpreted as per that between the explicit and implicit attitude measures of trust,
in Figure 1. However, it is likely that the current study offers respectively, for workmates, supervisor and the plant
a more valid measure of implicit attitudes about trust because leadership. These results are consistent with the findings of
each prime was paired with 10 trust targets, whereas in the the previous study and provide further evidence that explicit
previous study each prime was only paired with five trust and implicit attitudes about trust are separate psychological
targets. constructs.
In order to determine the strength of participants
implicit attitudes about trust, one-sample t-tests were Table 6: Correlation Coefficients and p Values for
conducted on each of the facilitation scores to determine Trust Facilitation Scores and Questionnaire Items
whether there was a significant difference from the baseline about Trust
(i.e. a facilitation score of 0). As seen in Table 4, all Trust Facilitation Score / n r p
facilitation scores were significantly greater than the baseline. Questionnaire Item
These results suggest that participants had strong implicit Trust workmates 20 -0.003 0.991
attitudes about trust and for their workmates, supervisors and I trust my workmates
Trust supervisor 19 0.326 0.173
the plant leadership. I trust my supervisor
Trust plant leadership 20 0.186 0.433
I trust the plant leadership
Table 4: Means, t Statistic and p Value for One
Sample Test (Test Value = 0)
Discussion. In order to assess safety culture more completely,
Facilitation Score Mean t df p the current study measured dual attitudes about trust for
Trust Workmates 94.680 4.261 19 < 0.001 workmates, supervisors and the plant leadership. Correlation
Trust Supervisor 79.600 3.912 19 0.001
analyses (See Table 6) revealed no relationship between
Trust Plant Leadership 48.984 2.441 19 0.025
implicit and explicit attitude measures about trust for
workmates, supervisor and the plant leadership, respectively.
Paired samples t tests were also conducted in order to
These results are consistent with the findings of the previous
test for differences in facilitation scores for trust between
study and suggest that implicit and explicit attitudes about
specific levels of the organisational hierarchy. Table 5
trust are separate constructs. If implicit and explicit attitudes
displays the means, t statistic and p value for each pair of
about trust were the same psychological construct, then a
facilitation scores.
strong positive relationship should have emerged between
8 SPE 86622

each of the trust facilitation scores and the questionnaire items also indebted to Dr. Kathryn Mearns and Dr. Peter McGeorge
about trust because these measures would have been of the University of Aberdeen for their substantial
measuring the same construct, just in different ways. contributions to all stages of the research.
However, this was not the case and the results are consistent
with the predictions made by Wilson et al.s [9] model of dual References
attitudes.
1. Cullen, W.D., The public inquiry into the Piper Alpha
The previous study proposed a model of dual disaster. 1990, London: HMSO.
attitudes about trust which conceptualizes explicit attitudes 2. Cox, S. and R. Flin, Safety culture: Philosopher's stone or
about trust as part of the superficial levels of safety culture man of straw? Work & Stress, 1998. 12: p. 189-201.
3. Cox, S. and A. Cheyne, Measuring safety culture in the
(i.e. safety climate) and implicit attitudes about trust as part of offshore environment. 1998, HMSO: Norwich.
the deeper levels of safety culture. The current findings 4. Flin, R., et al., Risk perception and safety in offshore oil
provide further support for this model as implicit and explicit workers. Safety Science, 1996. 22: p. 131-145.
attitudes about trust were found to be separate constructs. 5. Mearns, K., et al., Human and organisational factors in
offshore safety. 1997, HMSO: Norwich.
It is interesting to note that the implicit and explicit 6. Mearns, K., et al., Measuring safety climate on offshore
attitudes about trust reported in this study produced a more installations. Work & Stress, 1998. 12: p. 238-254.
similar pattern of results than in the previous study. In the 7. Reichers, A. and B. Schneider, Climate and culture: An
current study, the trust facilitation scores for workmates, evolution of constructs, in Organisational climate and
culture, B. Schneider, Editor. 1990, Jossey Bass: San
supervisor and the plant leadership were significantly greater Francisco.
than the baseline. These results suggest that participants had 8. Schein, E., Organizational Culture. American
strong implicit attitudes about trust for their workmates, Psychologist, 1990. 45: p. 109-119.
supervisors and the plant leadership. Also, participants 9. Wilson, T.D., S. Lindsey, and T.Y. Schooler, A model of
expressed positive explicit attitudes about trust for their dual attitudes. Psychological Review, 2000. 107: p. 101-
workmates, supervisors and the plant leadership. In sharp 126.
contrast, the previous study found that the trust facilitation 10. Reason, J., Managing the risks of organizational accidents.
score for workmates was significantly greater than the 1997, Aldershot: Ashgate.
baseline but the trust facilitation scores for supervisor and the 11. Westrum, R., Organisational dynamics and safety, in
Applications of psychology to the aviation system, N.
plant leadership were not. These results suggest that McDonald, N. Johnston, and R. Fuller, Editors. 1995,
participants in the previous study had strong implicit attitudes Avebury: Aldershot.
about trust for their workmates but not for their supervisors or 12. Hudson, P., Developing an aviation safety culture. SAA
the plant leadership. However, these participants still reported Journal, 2001.
positive explicit attitudes about trust for their workmates, 13. Helmreich, R. and A. Merritt, Culture at work in aviation
supervisors and the plant leadership. In terms of the newly and medicine. 1998, Aldershot: Ashgate.
proposed model, strong positive explicit and implicit attitudes 14. Mayer, R.C., J.H. Davis, and F.D. Schoorman, An
about trust for workmates, supervisors and the plant leadership integrative model of organizational trust. Academy of
should be expected from a good safety culture. Thus, the Management Review, 1995. 20: p. 709-734.
15. Rousseau, D.M., et al., Not so different after all: A cross-
results from the current study are more indicative of a good discipline view of trust. Academy of Management Review,
safety culture than are the results of the previous study. It 1998. 23: p. 393-404.
would be inappropriate to compare the facilitation scores from 16. Williams, M., In whom we trust: Group membership as an
the two studies more formally due to the fact that the current affective context for trust development. Academy of
study used a slightly different method to measure implicit Management Review, 2001. 26: p. 377-396.
attitudes in an attempt to develop a more reliable and valid 17. Fazio, R.H., et al., Variability in automatic activation as an
measure. Also, due to time constraints, the current study did unobtrusive measure of racial attitudes: A bona fide
not have as large and representative a sample of the workforce pipeline? Journal of Personality and Social Psychology,
as did the previous study. Future studies should use the 1995. 69: p. 1013-1027.
current method in order to validate the model of dual attitudes 18. Tajfel, H. and J.C. Turner, The social identity theory
about trust in safety culture by comparing implicit and explicit of intergroup behaviour, in Psychology of intergroup
attitudes about trust and safety performance between gas relations, S. Worchel and W.G. Austin, Editors.
plants or offshore installations. 1986, Nelson-Hall: Chicago. p. 7-24.

Acknowledgements

The first author gratefully acknowledges Shell Exploration &


Production (UK) and the University of Aberdeen for funding
this research as part of a PhD studentship. The first author is
SPE 86622 9

Figure 2: Mean Facilitation Scores from the Computerised Task Gas Plant B
(Error Bars represent the Standard Error of the Mean)

140

120
Facilitation Score (ms)

100

80
Facilitation Score; Trust
60
94.68
40 79.6

48.98
20

0
Workmates Supervisor Plant Leadership
Prime

Você também pode gostar