Latest / Safe As Podcast / E81: Is Safety Culture a Failed or Useful Concept for Organizations? A review of the Scientific Research

Transcript

E81: Is Safety Culture a Failed or Useful Concept for Organizations? A review of the Scientific Research · Safe As Podcast

0:00
When a serious accident happens, one of the recurring
0:03
explanations people reach for is culture.
0:05
There was a cultural issue. People weren't taking safety
0:08
seriously. Leadership tolerated the wrong
0:11
things. But when you look at the safety
0:13
culture research, the picture isn't so clear.
0:17
Safety culture is presented not as a settled idea, but as a
0:21
debated one, sometimes useful, sometimes measurable, sometimes
0:25
not. And sometimes a way of obscuring
0:29
real organisational issues underneath.
0:31
Hey everyone, welcome to Safe as I'm Doctor Ben Hutchinson.
0:35
Today I want to give you an overview of safety culture from
0:39
both a functionalist and an interpretive perspective.
0:42
Using several articles, I'll cover what safety culture is
0:46
usually trying to capture, how it differs from safety climate,
0:51
where their concept becomes potentially useful and where it
0:54
becomes unstable, and what the various authors think actually
0:58
matters in practise. Article 1 is from the OHS body
1:01
of knowledge called Organisational Culture, A Search
1:04
for Meeting written by Doctor David Boris.
1:07
This chapter begins with the historical trigger point that
1:10
appears again and again in the research the Chernobyl disaster
1:14
in 1986. According to the chapter, that
1:17
event helped trigger an explosion of academic and
1:20
organisational interest in safety culture, but the
1:23
chapter's main argument is that this growth and interest didn't
1:26
produce conceptual clarity. Instead, safety culture remains,
1:31
in the author's words, a confusing and ambiguous concept
1:35
with unresolved debates and definitional dilemmas.
1:39
A major issue in this chapter is the relationship between safety
1:42
culture, safety climate, and organisational culture.
1:45
The chapter argues that these terms have often been used
1:48
interchangeably but also contested.
1:52
One recurring distinction is that safety climate is treated
1:55
as more immediate, more perceptual, and easier to
1:58
change, while safety culture is treated as broader, deeper, and
2:02
more stable. So what is culture in this
2:05
context? It depends on who you ask to
2:08
shine. Organisational culture is a
2:11
pattern of shared basic assumptions learned by the group
2:14
as it's solved it's problems and it works well enough and it's
2:18
taught to new members as sort of the correct or appropriate way
2:22
to perceive, think and feel for Golden Mund saved your culture
2:26
then are those aspects of the organisational culture which
2:29
will impact on attitudes and behaviour related to increasing
2:33
or decreasing risk for Golden Mund.
2:35
It's a construct, meaning an abstract concept rather than
2:39
something concrete you can directly observe.
2:42
It's relatively stable and multi dimensional, shared by groups
2:47
and it constitutes practises with layers of rituals, heroes,
2:52
symbols, norms, values and it's said to be learnt.
2:55
But it's also functional where it supplies frames of reference.
2:58
For Dominic Cooper it's a subculture of the organisation
3:01
overall culture or cultures. It's a product of multiple goal
3:05
directed interactions between people, jobs and the
3:08
organisation, reflected in the dynamic reciprocal relationships
3:11
between people's perceptions and their attitudes towards
3:16
operationalizing safety goals and goal directed safety
3:19
behaviours. So some maintain that safety
3:21
culture is a specific subculture of the organisational culture of
3:24
cultures, which directs attentions to the patterns of
3:28
thoughts, values and actions related to safety.
3:31
One way to think about culture based more on the interpretive
3:34
frame comes from Shine's idea that culture operates in layers.
3:39
At the surface you have artefacts that the visible
3:42
things in the organisation, like meetings, posters, reports,
3:46
dress codes, PPE. But as the chapter argues, those
3:50
are easy to spot. But the point the chapter makes
3:53
is that they can still be hard to interpret on their own under
3:56
the superficial artefacts of the espouse, values and attitudes.
3:59
Things like policies, procedures, training, materials
4:03
and formal statements about what the organisation says matters.
4:07
And then at the deepest level there are basic assumptions that
4:12
taken for granted beliefs people may not even articulate directly
4:17
or even realise is influencing the way that they act.
4:20
Like assumptions about authority, behaviour,
4:24
relationships or how work is really meant to be done.
4:26
So the value in Shine's model, as discussed in this chapter, is
4:31
that it reminds US culture is not just what an organisation
4:34
displays on the surface, it's also the deeper assumptions
4:37
underneath that shape how safety is understood, how it's
4:40
prioritised and acted on in practise.
4:43
Another perspective frames it not as a or their culture, but
4:47
as cultures, plural. This says we have multiple
4:51
cultures, including micro cultures which emerge within
4:55
small groups like teams. Then we have mesocultures which
4:59
is at a higher level, like maybe a business unit organisation,
5:03
higher education institutes or regional ethnic and societal
5:07
groups and more, but it's still smaller than a national culture.
5:10
And then we have macro cultures which are large scale shared
5:14
beliefs across whole industries, society and more.
5:18
Alternatively, we can also talk of subcultures which are subset
5:22
cultures within larger cultural frames.
5:24
Interestingly, some research has shown that professional
5:28
subcultures can be more representative and influential
5:32
than organisational cultures. For instance, civil engineers
5:36
across different or even competing organisations may
5:40
share more cultural elements and norms than to those engineers
5:44
within their own organisational cultures.
5:46
Climate is often linked to what people perceive about policies
5:50
and practises and procedures, and more culture is often
5:52
treated as something more fundamental but also more
5:55
difficult to define and measure. Now this matters because the
5:59
chapter is not just reviewing the definitions, it's trying to
6:02
make sense of what the concept is actually doing in practise,
6:05
how it influences us in practise.
6:07
And one of the most useful ways it does that is by laying out
6:10
four central questions from the safety culture debates.
6:14
Can culture be managed? Can culture be changed?
6:17
Can culture be measured? And does a good culture improve
6:20
performance? These four questions are really
6:23
the backbone of the chapter. The first question can culture
6:27
be managed? The chapter presents 2 broad
6:30
positions. On the more functionalist side,
6:33
culture is treated as something like an organisation has
6:36
something that can be influenced through leaders, structures,
6:40
systems and collective practises.
6:42
On the more interpretive side, culture is treated more as
6:45
something like an organisation is a socially shared experience
6:50
that can't simply be controlled from the outside.
6:53
Importantly, the chapter doesn't resolve this debate, but it does
6:56
point to a more practical middle ground from Hofstede and Reason.
7:00
Well, values are really hard to engineer directly.
7:03
Collective practises can be influenced through structures
7:06
and systems. And for Andrew Hopkins at least,
7:09
he liked to reference this quote from Shine that ladies create
7:12
cultures by what they systematically pay attention to
7:16
or ignore. That flows straight into the
7:17
second question, Can culture be changed?
7:20
The chapter says a singular focus on changing values is
7:24
likely to be ineffective. Instead, the more promising path
7:28
is to change collective practises, with values
7:31
potentially shifting later through experience.
7:34
It argues that change efforts should focus on what people
7:37
actually do, how work is arranged and how risk is
7:41
controlled, and how decisions are made, rather than on vague
7:43
attempts to improve culture in the abstract sense.
7:47
The third question is measurement, and the chapter is
7:50
cautious here as well. It reviews tools like maturity
7:53
models and safety climate questionnaires, but it also
7:56
notes ongoing disagreement about what exactly should be measured
8:01
and whether surveys can really capture a concept as deep and
8:05
contested as safety culture. It specifically notes that
8:08
safety climate researchers have identified dimensions such as
8:12
management, safety systems, risks, work pressures, and
8:16
competence, but say there's no comparable clarity or
8:20
consistency around safety culture itself.
8:22
Then comes the 4th question. Does a good culture improved
8:25
performance? This is where the chapter draws
8:27
one of its, I think, clearest lines.
8:28
It says there's consensus along safety climate researchers that
8:32
a higher perceived safety climate, through its influence
8:36
on safety behaviour, is associated with fewer injuries.
8:40
But the outcomes of safety culture research are much less
8:43
clear. The chapter says there's little
8:45
evidence of a direct relationship between safety
8:47
culture and elements of performance and injuries and
8:50
incident performance. That's not the same as saying
8:52
the whole contract has no value. In fact, the chapter spends time
8:56
reviewing different efforts to describe a more positive or
8:59
negative safety culture, if that's even possible, which is
9:02
itself a heavily challenged idea.
9:04
It summarises recurring theme from major sources like
9:07
leadership, communication, employee involvement, trust,
9:11
learning, flexibility, openness about failure and integration of
9:16
safety into ordinary work. It discusses Nick Pigeons, work
9:19
on management commitment, shared concern about hazards, realistic
9:23
norms and reflection through feedback systems.
9:26
It also discusses Andrew Hales emphasis on safety being
9:29
prioritised alongside other organisational goals.
9:32
And it's a belief that their causes and contributing factors
9:36
and solutions lie across multiple interacting factors and
9:39
not just individual behaviour. But after all of that, the
9:42
chapter still comes back to the same practical conclusion.
9:45
Trying to change safety culture directly is not the most useful
9:49
way forward. Instead, it argues that
9:51
workplace safety is better served by changing
9:54
organisational and management practises that have an immediate
9:58
and direct impact on risk control.
10:01
That includes things like reporting systems and decision
10:04
making, resources, procedures, learning processes, worker
10:08
engagement, accountability and leadership visibility.
10:11
And I think this is where the chapter becomes particularly
10:13
valuable for practise. It doesn't just say focus on
10:16
practises, it also asks what an organisation that focuses on
10:20
safety actually looks like. It lists characteristics such as
10:24
rewarding bad news, challenging good news, maintaining a clear
10:29
and shed collective picture of risk, supporting risk control
10:33
through structures and standards and providing suitable tools and
10:36
materials, seeing on learning, making invisible risks visible,
10:41
encouraging whistle blowing, integrating safety into all
10:44
aspects of the business like design, and having leaders who
10:47
actively and visibly promote safety.
10:50
The article provides A framework for a cultural understanding
10:53
within organisations. Being 1.
10:55
Deciphering and unpacking work go deeper than vague value
10:59
statements and grasp the meaning of the acts, the objects, the
11:03
words, the rituals, the things that we tell ourselves. 2 It's a
11:06
metaphor for the organisation and needs more specific views to
11:10
construct meaning. So safety culture by itself can
11:14
be quite abstract and vague. Three.
11:16
Try to see culture where you don't expect it. 4 Culture both
11:20
guides and constrains and blinds us to issues and risks.
11:24
It's some sort of regulatory framework with both helpful and
11:27
obstructive elements. 5 Cultural meanings aren't spontaneous but
11:31
reflect ideologies and actions of powerful agents.
11:34
Power and power gradients are ever present. 6 Culture isn't
11:39
static in uniform. All right, the second article
11:41
again from the OHS body of knowledge titled Organisational
11:44
Culture Reviewed and Repositioned again from David
11:48
Borris. The second chapter is written as
11:50
a follow up and asks A straightforward question.
11:53
What changed between 2014 and 2019 since that first chapter?
11:57
The answer is partly that the confusion around safety culture
12:00
remains, but the evidence for safety climate became much
12:04
stronger. The chapter says that there is
12:06
strong evidence for a relationship between safety
12:08
climate and a whole range of OHS outcomes, and weak evidence for
12:13
a relationship between safety culture and similar safety
12:16
outcomes. On that basis, the chapter says
12:19
health and safety professionals should emphasise safety climate
12:23
over safety culture. I think what makes this chapter
12:25
useful in one sense is that it sets that conclusion against the
12:29
reality of industry practise. The chapter argues that there's
12:32
a clear research versus practise gap.
12:35
In industry, people talk almost exclusively about safety
12:39
culture. Executives talk about it, So do
12:42
job ads. But it's said that the evidence
12:44
base is much more supportive of a safety climate, and that many
12:49
safety culture surveys are really measuring safety climate
12:52
anyway because they focus on perceptions of systems,
12:56
procedures and expected behaviour and those norms and
12:59
values, etcetera. It also revisits the definitions
13:01
of safety culture, which is everyone's favourite part.
13:04
So while the 1st chapter suggested that we should stop
13:07
trying to define safety culture, this chapter carries that
13:10
argument forward it it keeps organisational culture as the
13:13
broader concept of interest and uses the way we do things around
13:17
here as it's working definition. But it also notes that this
13:20
definition is controversial and can be seen as too broad or even
13:24
too mechanistic, too behavioural.
13:26
Still, it argues that the emphasis on way way we do things
13:30
around here highlights that the culture is a group phenomenon,
13:33
while the emphasis on do points towards practical action and the
13:38
collective work required for good health and safety outcomes.
13:41
Another definition, and one that I've always liked, comes from
13:44
Pigeon and O'Leary. He said that safety culture, or
13:47
by extension a culture of safety, is the set of
13:49
assumptions and their associate practises which permit beliefs
13:53
about danger and safety to be constructed.
13:56
In turn then, such a culture is itself created and recreated as
14:02
members repeatedly behave and communicate in ways which seem
14:06
to them to be natural, obvious and unquestionable, and as such
14:11
will serve to construct a particular version of risk,
14:14
danger and safety. Peasant and O'Leary draw on Carl
14:16
Wachs work, noting that an organisation is of necessity
14:21
defined not so much by what its members attend to or pay
14:24
attention to, but what they choose to ignore, and further
14:27
quoting their article. As Vaughan observed about NASA's
14:30
cultures prior to the challenge of space shuttle disaster, their
14:33
culture provided a way of seeing that was simultaneously a way of
14:37
not seeing. Hence these institutionalised
14:39
assumptions and norms that make up our cultures, quoting Pigeon
14:43
and O'Leary, have the capacity to simultaneously illuminate
14:46
some hazards or shifting attention away from others.
14:49
So cultures can also blind us to risks.
14:53
The chapter reviews the post 2014 evidence and it identifies
14:56
6 systematic reviews, a growing focus on interventions,
15:01
increased attention to maturity models, and recurring work in
15:05
industries like transport and construction.
15:07
But it remains cautious about what we conclude from the
15:09
evidence. It says the relationship between
15:11
safety culture interventions and safety performance is still
15:14
inconclusive. It says maturity models have
15:16
become more popular, but whether they improve health and safety
15:19
outcomes is still unknown. And that's really important.
15:22
We're largely in the dark about whether these glossy popular
15:26
safety culture maturity approaches like the Bradley
15:28
Curve and others are effective for their goals or result in any
15:32
measurable improvements in anything.
15:34
The chapter then summarises the evidence on what we know, and
15:38
what we know includes how to define safety climate.
15:41
That group and supervisor roles are important, of course.
15:45
That safety climate surveys potentially can predict unsafe
15:48
behaviours and maybe safety outcomes.
15:50
But that might be a reverse causality effect.
15:53
And evidence for safety culture interventions is inconclusive.
15:56
What we don't know includes how to define safety culture, how to
16:00
evaluate safety culture interventions, how to actually
16:03
change safety culture, the exact difference between safety
16:07
culture and safety climates, how to consistently measure both,
16:09
and whether safety culture maturity models improve
16:12
outcomes. It argues that given the
16:14
evidence as of this publication's date, perhaps we
16:17
should emphasise safety climate over safety culture.
16:20
And instead of asking whether the organisation has a good or a
16:23
bad safety culture, maybe we should ask what intervention are
16:27
we trying and how will we know if it works.
16:30
The chapter outlines a three stage intervention evaluation
16:34
process as it before, during and after.
16:37
During the intervention, ask how is it going, who the feedback
16:40
should come from and what should be adjusted afterward.
16:44
Ask what should be now measured, when should it be measured and
16:48
how often. This is also where the chapter
16:49
introduces the phrase culture of safety as potentially a better
16:53
way forward than safety culture. It argues that after decades of
16:57
searching for a strong, proactive, positive or mature
17:00
safety culture, that search has been largely unsuccessful.
17:04
So rather than clinging to safety culture, which has a
17:07
whole lot of baggage attached to it, the chapter suggests about a
17:10
culture of safety defined as the way the organisation works
17:14
through leadership practises, organisational structures,
17:18
organisational practises and management systems.
17:20
Practically, it's said that if interventions improve the
17:22
conditions under which workers work, and if workers perceive
17:26
those interventions as meaningful, which can be checked
17:29
using simple climate measures, and more, than that's what's
17:33
actually matters. In that sense, successful
17:36
interventions reduce the need to talk about culture in the
17:39
abstract at all. In other words, get the
17:41
intervention and structures right and let culture take care
17:45
of itself. Article 3 is Taming Prometheus
17:48
talk about safety and culture from Susan Silby.
17:50
This article takes a more critical and sociological
17:53
approach. She starts by observing that
17:56
safety culture has become a common explanation for accidents
17:59
and a common recipe for improvement in complex socially
18:02
technical systems. But her core argument seems to
18:04
be that the dominant ways of talking about safety culture are
18:08
theoretically weak because they rely on individualist and
18:12
reductionist assumptions. Silbi identifies 3 concepts of
18:16
culture and safety discourse. Culture is a causal attitude.
18:20
Culture is engineered organisation.
18:22
The culture is emerged and indeterminate.
18:24
The 1st 2 are the main focus for critique.
18:27
In the first, culture is treated as a measurable source of
18:30
product or individual attitudes and behaviours.
18:33
In the second, culture is treated as something that can be
18:35
designed or engineered into safer organisations through
18:38
reliability learning. Structural redundancy argues
18:42
that both approaches fail to adequately deal with how culture
18:45
works in complex organisations. Her preference is the third
18:48
concept. Culture is emerging and
18:50
indeterminate. In that view, culture isn't just
18:53
a management variable or a survey score.
18:55
It's produced through ongoing practise, interpretation,
18:59
conflict and structure. So it's both a product of action
19:03
and a context of action, meaning it can't be simply engineered
19:07
nor understood without attention to power, authority and
19:10
competing interests. Silby argues that the rise of
19:13
safety culture discourse aligns with the weakening of older
19:16
systems of collective responsibility.
19:18
Regulation, insurance, labour litigation, and other
19:21
institutional forms at once helped distribute the cost of
19:24
technological harm. From that logic, safety culture
19:27
has become part of the broader shift towards responsibilizing,
19:30
where individuals are encouraged to take responsibility for
19:33
outcomes shaped by much larger systems.
19:35
BP Texas City is using an example for her.
19:39
BP uses shallow notion of safety culture focus on individual
19:43
attitudes and actions, even though the deeper issues
19:45
involved underfunded management, overloaded workers, under
19:49
resourced safety programmes and failures in process safety.
19:52
Silby argues that safety culture discourse often ignores power,
19:56
hierarchy, inequality and conflict.
19:59
It can treat organisations as if everyone shares the same
20:02
interest and sees the same risk, when really organisations are
20:06
structured by buried information, competing interests
20:10
and priorities, and uneven ability to influence decisions.
20:14
That's why she argues that future research should pay more
20:16
attention to these matters and power gradients in all.
20:19
So he pushes the safety culture discussion away from slogans and
20:23
scores towards deeper questions like who gets heard, what counts
20:28
as credible knowledge, what gets normalised, what gets ignored,
20:33
and how system structure shapes responsibility.
20:35
Another perspective comes from Dominic Cooper's 2016 article
20:39
Navigating the Safety Culture Construct.
20:41
A review of the evidence. Cooper presents a functionalist
20:45
view of safety culture. His main argument seems to be
20:48
that the field has become bogged down by too many definitions and
20:51
too much conceptual confusion, while industry still needs
20:54
practical guidance on what actually prevents process safety
20:58
incidents and serious injuries and fatalities.
21:00
He reviews three major models of safety culture and reaches a
21:03
clear conclusion. The Goldenman's Shine layered
21:07
model isn't supported by the evidence that is published.
21:11
Links to incidents and injuries, while Cooper's own reciprocal
21:14
model and reasons model are said to be linked.
21:16
Cooper's own model treats safety culture as the product of
21:19
interactions between psychological factors, behaviour
21:22
and situational factors such as systems and structures.
21:25
From there, Cooper zooms to six recurring safety culture
21:28
characteristics, management and supervision, safety systems,
21:32
risk, work, pressure, competence, procedures and
21:36
rules. He argues that safety climate
21:38
and safety attitude measures only show weak and inconsistent
21:42
links to actual outcomes, while values and norms have little
21:47
direct evidence behind them. In contrast, situational factors
21:50
like safety management systems and behavioural factors show
21:53
much stronger and more consistent relationships with
21:56
real safety outcomes. This finding contradicts to
21:59
reviews from Boris in the OHS body of Knowledge, who said that
22:03
largely the extent research seems to support safety climate,
22:07
not safety culture. Anyway, Cooper's arguments lead
22:10
to his main conclusion. Using psychological surveys
22:13
alone as a proxy for safety culture is fatally flawed.
22:17
Instead, organisations should focus more on their effort on
22:20
improving systems managerial behaviour and the concrete
22:23
safety issues that shape everyday work.
22:25
That is actually consistent with the other earlier works.
22:28
Cooper argues that 80% or more of safety culture improvement
22:31
efforts should be directed towards situational behavioural
22:34
factors, with management behaviour being especially
22:37
critical. So overall, Cooper treats safety
22:39
culture less as a deep interpretive mystery and more as
22:42
a practical, observable and manageable organisational
22:46
problem. It seems to be stop relying on
22:48
abstract attitudes is the main explanation, and focus instead
22:51
on the structures, leadership practise and behaviours that
22:54
actually reduce risk. All right, so that's some pretty
22:57
dense theoretical or philosophical discussion.
22:59
What about studies exploring the links between safety culture and
23:03
various elements of safety, incident performance and et
23:05
cetera? So one is from Calte ET al 2018.
23:09
The review says safety climate and culture and safety
23:11
performance are related, but the relationship is unclear because
23:15
of the definitions and tools. And performance measures vary
23:18
across studies and reports that reactive performance measures
23:21
tend to show a negative relationship with safety climate
23:24
and culture, while proactive measures tend to show a positive
23:28
relationship with safety compliance, appearing especially
23:31
consistent. It also states there was no
23:33
reviewed studies evaluating safety interventions against
23:36
changes in safety performance, making causal interpretation
23:39
really challenging and limited to from Vican ET al.
23:43
They reviewed 34 studies and found that most studies reported
23:46
reduce adverse events in hospitals when patient safety
23:50
culture scores increased. Teamwork climate and safety
23:53
climate tended to be the most frequent dimensions associated
23:57
with that reduced adverse events.
23:59
This finding supports a review demonstrating that teamwork and
24:03
communication training improves patient safety culture scores
24:07
and suggest that improving these dimensions may actually reduce
24:10
adverse event rates. Additionally, these findings
24:13
raise the question about the concepts of patient safety
24:15
culture and patient safety climate.
24:17
Previous reviews suggests that patient safety culture
24:20
questionnaires capture patient safety climate and the tangible
24:24
themes are not the really intangible themes under the
24:27
surface of culture. 3 is from Finn ET al 2024.
24:30
This review finds that safety culture interventions and
24:32
hospitals most often affect teamwork, leadership, reporting,
24:36
just culture and perception of safety.
24:38
It reports that staff related benefits including higher job
24:42
satisfaction, higher stress recognition, lower emotional
24:46
exhaustion, lower burnout, lower turnover, and better working.
24:50
It says the clearest effects appeared in interventions with
24:53
longer time frames, stronger leadership support,
24:57
comprehensive theory, informed designs and a unit level focus,
25:01
while overall effects across the literature were still mixed and
25:06
heterogeneous. So it's really difficult to
25:09
understand if there is a direct link between them at scale.
25:12
So despite the many nuances and limitations of research, as in
25:16
varying definitions of what safety culture is, weak study
25:19
designs, problems of dependent measures, etcetera.
25:22
Where the safety culture has positive effects on a whole
25:25
range of performance measures isn't really certain.
25:27
But there does seem to be a trend with positive things like
25:31
safety compliance behaviour and the like.
25:32
But it also appears that these improvements seem to be more
25:36
related to safety climates, which is just easier to capture
25:40
and research using questionnaires, et cetera.
25:42
Indeed, Diane Vaughn has argued that you need more than
25:45
questionnaires to capture culture.
25:48
It permeates across and within all elements of the
25:51
organisation. Now I want to quickly cover some
25:53
final points before we wrap up. So from Hopkins in a 2018
25:58
article. Culture is a misunderstood and
26:00
misused idea and has led to endless confusion.
26:03
One problem is that safety culture doesn't really say
26:05
anything about the content of the culture or the cultures.
26:09
It has so much existing baggage that Hopkins suggests that, in
26:12
his words, it should be abandoned.
26:14
Instead, he suggests that if we're highlighting an
26:16
environment that prioritises safety, then perhaps a culture
26:20
of safety would be preferable, and maybe less confusing too.
26:24
Research on performance improvements resulting from
26:26
safety culture interventions are intermixed with safety climate
26:30
research. Some argue that climate is
26:32
actually a part of safety culture, but it does make it
26:35
difficult to laser in on the effects of safety culture
26:37
directly when it's relying so heavily on safety climate
26:41
interventions. 3 is a study called Does the concept of
26:43
safety culture help or hinder systems thinking?
26:46
So these authors like Roland, Hagen and Riemann suggest that
26:50
the vague notions of safety culture might actually distract
26:54
attention away from addressing the underlying organisational
26:58
issues like design or system syncing, and therefore we focus
27:03
more on trying to address this nebulous construct of safety
27:06
culture over the actual issues we're really concerned about.
27:09
They argue that statements such as we have problems in safety
27:13
culture, or the incident was caused by poor safety culture
27:16
should be treated with equal scepticism as the old
27:19
statements. We have a problem with humans
27:21
making errors, and the incident was caused by human error.
27:25
Indeed, for them, safety culture, if used at all, should
27:30
be the start of the investigation rather than its
27:32
conclusion. 4 is from and Thompson.
27:35
Like Sylvie, there are conflations between culture and
27:38
power which are treated as similar, but they're not.
27:41
As stated by Charles Perot, we miss a great deal when we
27:45
substitute power with culture. And then Thompson explored where
27:49
the safety culture was predictive of a major accident
27:52
by comparing the results of cultural surveys before and
27:56
after the Snore Alpha blowout. While pre disaster cultural
27:59
items scored pretty positively, with people scoring adherence to
28:03
procedures, prioritisation of safety and organisational
28:07
learning pretty well and all of them were said to be key aspects
28:10
of a safety culture. In contrast, post accident it
28:13
was said to have had a safety priority subservient to
28:16
production goals, lack of risk assessments, severe procedural
28:20
departures and more. In that case it wasn't
28:22
predictive and in many areas quite conflicted and
28:26
contradictory. So for that author, the way
28:29
cultural values were studied resulted in part due to the
28:32
constrained nature of the surveys.
28:35
It tapped into perceived attitudes, values and
28:37
perceptions, but regarding deeper cultural levels require
28:41
that respondents should tell us more than they actually know.
28:44
Therefore, gaining info on cultures requires a more
28:47
interactive assessment where insiders and outsiders engage in
28:52
a process of joint inquiry to uncover cultural assumptions.
28:56
This emphasises taking the actors point of view rather than
28:59
relying on the researchers point of view based on a priority
29:03
category, things that we've already determined we want to
29:05
study. 5 There is next to no robust evidence underpinning the
29:09
concept of safety culture maturity models is we're pretty
29:14
much in the dark about whether they achieve their goals or
29:16
improve anything relevant. 6 Finally, Halliwell and
29:20
colleagues studied safety culture in US construction.
29:23
They actually argued that the elimination of safety culture
29:26
from the safety science lexicon should be a goal.
29:28
As per their words. It's deceptively simple and
29:32
neatly abstract, but it's enigma has plagued the research since
29:37
its inception and, you know, efforts to precisely define and
29:40
operationalise safety culture. In their words, it may simply be
29:44
too tainted and problematic for any future utility.
29:48
Problematically, since safety culture has captured nearly
29:50
everything, values, beliefs, norms, attitudes, perceptions,
29:54
assumptions, behaviours and everything, if it means
29:57
everything then it might as well be nothing.
29:59
If all things safety are also part of safety culture, are we
30:03
not just talking about safety? And if everything is important
30:06
for safety culture, then nothing is.
30:08
They cover different conceptual streams of safety culture stream
30:12
1A fuel of engineering or normative safety culture being
30:16
grounded in organisational management and processes around
30:19
safety. This is quite a positivistic
30:21
lens and it's revealed by analysing documents and data
30:25
etcetera, stream. 2 An analytical or pragmatic view
30:29
reflecting via measurable attitudes and behaviours and
30:32
also an interpretivist lens revealed by proxies like surveys
30:35
and interviews. 3 and emerge in an anthropological view being
30:40
socially constructed and revealed by shared
30:43
understandings and beliefs etcetera.
30:45
A constructionist constructivist lens and revealed via immersive
30:49
longitudinal studies and ethnographic work.
30:52
You don't tend to see that a lot in this sort of research world.
30:54
But in all, safety culture is even more problematic for the
30:57
construction industry since it's driven by project driven work
31:01
and transient workforces which move between projects.
31:04
Safety culture is seen as something generally shaped
31:07
gradually over time by multiple levels, resulting in a
31:10
relatively stable phenomena, factors not typically associated
31:14
with construction. Therefore, this begs the rather
31:17
fundamental question of whether the highly variable and
31:20
inconsistent construction industry even has the capacity
31:25
to develop safety culture within its normal scope of operations.
31:28
Instead of focusing on the vacuous concept of safety
31:31
culture, they suggest focusing on the constituent elements.
31:35
Instead of using safety culture as a proxy for worker engagement
31:38
or whatever else, actually focus on understanding worker
31:41
engagement. So what can we make of all of
31:44
those findings? I can't possibly resolve the
31:47
confusion around culture, nor was that my intention.
31:51
I don't want to try and give you a 5 point action plan, but
31:54
here's some takeaways I took from the articles.
31:57
One, focus on practises and shared understandings and
31:59
routines, not abstract culture change.
32:02
I think the strongest practical message across the Viruses
32:05
chapters is to improve concrete organisational and management
32:08
practises, reporting, leadership, learning, resources,
32:13
risk control, rather than trying to change some vague elements of
32:18
safety culture directly. True, consider using safety
32:21
climate to measure parts of your change.
32:24
As argued in the second Bars chapter, Safety climate seems to
32:27
have the stronger evidence base, and it's undoubtedly a more
32:31
practical way to assess whether interventions are working within
32:35
a very narrow window 3. Evaluate your interventions over
32:39
assuming they just have an impact.
32:40
The articles repeatedly show that evidence for safety culture
32:43
interventions is mixed or inconclusive, so the practical
32:46
question should be whether a specific intervention works
32:49
before, during, and after implementation.
32:52
Leadership, reporting and teamwork seem to be recurring
32:55
leverage points. Probably no surprises. 4 Be
32:59
cautious about claiming performance gains from culture
33:02
alone. The papers show mixed links with
33:04
performance. Some studies report better
33:06
reporting, teamwork and job satisfaction, or even lower
33:09
incidence, but others show no measurable improvement in
33:12
quality scores or error management or whatever else. 5
33:15
Use culture as a lens, not a catch all explanation.
33:18
The interpretive critiques argue that culture is most useful for
33:23
understanding context and meaning, power and how risks
33:27
become normalised, not as a simple label for blame or a
33:31
standalone explanation for performance.
33:33
So in all, safety culture is conceptually unstable and
33:38
shouldn't be treated as a simple causal performance lever.
33:41
What seems more useful in practise is to improve concrete
33:45
organisational conditions, the leadership structures, incentive
33:49
reporting, communication, learning, staffing, resources,
33:53
risk control, all this stuff, and then use safety, climate or
33:56
other defined indicators and measures to see whether those
33:59
changes been experienced in meaningful ways.
34:02
At the same time, the interpretive critiques warn
34:05
against assuming organisations are unified, assuming everyone
34:09
shares the same interests or the same culture, or treating poor
34:13
outcomes as simply a matter of worker attitude rather than the
34:17
structures and power balances and system design.
34:19
So I think a defensible practical take away from the
34:23
articles is, in my view, improve the work system, evaluate the
34:27
intervention and measure the experience of that change
34:30
carefully, and be cautious about claiming that a single culture
34:33
score explains performance, if at all.
34:36
That's it. On Safe As.
34:37
Please subscribe, like and comment and links in the show
34:41
notes to support the channel.