SPE 86624

Understanding Safe and Unsafe Behaviours: A Scientific Approach

Rhona Flin, Industrial Psychology Research Centre, University of Aberdeen, Scotland

Copyright 2004, Society of Petroleum Engineers Inc.

This paper was prepared for presentation at The Seventh SPE International Conference on
Health, Safety, and Environment in Oil and Gas Exploration and Production held in Calgary,
Alberta, Canada, 2931 March 2004.
This paper was selected for presentation by an SPE Program Committee following review of
information contained in a proposal submitted by the author(s). Contents of the paper, as
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
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
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
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
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.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract
The proximal causes of a significant percentage of oil and gas
industry accidents relate to human error and unsafe behaviour
at the worksite. In order to maximize safe behaviours and
reduce unsafe behaviours, supervisors and managers need to
understand the factors influencing them before introducing
behavioural safety interventions. This paper argues that the
same level of scientific expertise should be applied to the
diagnosis, design and evaluation of behavioural safety
interventions as is devoted to engineering interventions.
Introduction
The oil and gas industry is at the forefront of modern
engineering technology and practice. This success is founded
on well established, scientific processes which are used to
establish the causes of technical failures by gathering
diagnostic data, to design and test interventions, and then to
evaluate whether they were fit for purpose. It would be
inconceivable for an oil company to ask a psychologist to
design a pipe connector or to diagnose a generator failure.
Qualified specialists with relevant knowledge and skills are
used for these tasks. This paper argues that the same level of
scientific expertise should be applied to the diagnosis, design
and evaluation of behavioural safety interventions.
In recent years, the oil and gas industry has realised that
they need to address the behavioural basis of safety, as well as
technological and procedural solutions (1),. They tend to refer
to this as Human Factors or Behavioural-Based Safety. The
term 'safety behaviour intervention' is used here to mean any
intervention designed to enhance safety by changing human
behaviour. This would cover safety observation and
reinforcement programmes, as well as various kinds of safety
training, coaching, role modelling etc. If we look at the
introduction of behaviour-based safety interventions, are they
based on the same level of attention to scientific knowledge as

we see in engineering or industrial chemistry? When

investigating accidents or designing safety behaviour
programmes do oil companies use qualified behavioural
science specialists? Behavioural science specialists are
psychologists (who study the science of mind and
behaviour), educational specialists or other types of human
factors specialists (human factors is the label for a range of
disciplines which study human performance e.g. ergonomists,
psychologists, engineers, physiologists). My experience of
working with the oil and gas industry for the last fifteen years
is that this is not always the case.
If the objective is to implement behavioural based safety
interventions (e.g. team or leadership safety training, safe
driver training or STOP type observation programmes), then
these should be designed with the same level of scientific
rigour that would be applied to designing and implementing an
engineering intervention. The required steps are very similar
see Figure 1.
The steps are as follows:
1.
2.
3.
4.

Diagnose the safety problem and identify critical

behaviours and their context.
Design the intervention to encourage desired
behaviours and to reduce undesired behaviours.
Implement the intervention.
Evaluate during following year.

Diagnosing Safety Behaviour Problems

First, the safety problem needs to be properly diagnosed
which behaviours are present or absent in this situation, who is
involved, when do they tend to happen (context), what
encourages or discourages them, etc? This can be carried out
by various forms of task analysis as well as incident and
accident analysis. Task analysis refers to a body of techniques
used to examine the core tasks that make up a given job. There
are well established methods available for this purpose (2).
Data gathering methods include the study of documentation
(eg procedures manuals); observations; interviews;
questionnaires. It is also important to gather data on protective
safety behaviours and recovery from errors as well as unsafe
behaviours. In many modern workplaces there is now much
less reliance on manual actions and therefore if the task
involves monitoring, thinking, problem solving then
cognitive task analysis techniques (3) can be used. A training
needs analysis identifies the gap between current levels of
knowledge, skills and attitudes and the desired level.
Techniques like TTRAM (Task and Training Requirements

Analysis Methodology (4), are useful for identifying team

rather than individual training needs.
Organisational level information on safety is usually
obtained from climate surveys to determine how people feel
about the way safety is managed, how people behave in
relation to risks and safety rules, how error is treated, will
people speak up about safety issues. These questionnaires are
usually completed anonymously and the data aggregated into
work units or pooled to reflect the whole asset or company (5);
(6). Such data can be very valuable in building up a picture of
prevailing attitudes and behavioural norms. They can also be
used to determine whether an organisation is 'ready' for a
particular intervention and can pinpoint areas of concern that
may need to be explicitly addressed (e.g. lack of trust) when
the behavioural programme is designed.
Accident analysis can provide important information,
provided that the analytic tool captures non-technical
(cognitive and social behaviours) as well as technical
behaviours and organizational (context) factors (7). The HFIT
(Human Factors Investigation Tool) method was designed for
this purpose for the oil and gas industry (8) and others such as
SYNERGY or TRIPOD are available. Industries such as
aviation and rail also have confidential incident reporting
systems, which allow workers to report near misses, incidents,
safety concerns not anonymously but confidentially to
allow follow up interviews, then de-identification usually to
an outside body, although some of the airlines also have their
own in-house systems. These can provide a wealth of
behavioural data for designing safety interventions, such as
targeted training.
Diagnostic analyses of safe or unsafe behaviours should
be based on an understanding of human behaviour. There is a
wealth of scientific information available on human and
organizational factors, how they can affect behaviour and how
the behaviours can be changed. Much of this (eg human error,
decision making, stress, team safety, motivation, safety
climate) is in the aviation psychology, military psychology or
industrial psychology literatures, but the broader human
factors domain is also a rich source of information on failures
in human /technology interaction, fatigue, accident analysis
etc). (See Appendix 1 for a list of sources). There may be
empirical evidence on the desired and undesired behaviours
and what influences them in the oil and gas industry (see
papers by Sneddon on situation awareness and Burns on trust,
this conference volume). Traffic accidents are a major causes
of injuries and fatalities in the onshore oil and gas industry see for instance (9) for a recent study on drivers' thinking just
before a simulated traffic collision.
It is usually advisable to base the behavioural diagnosis
and intervention design on an accepted scientific theory or
model of human behaviour. Behavioural observation and
reinforcement programmes, such as STOP tend to focus on the
discouragement of unsafe acts at the worksite. Some of these
programmes (and there are many different versions available)
do have a proper scientific basis in learning theory, using
operant conditioning of behaviour ie manipulating the
consequences of behaviour in order to shape it. However more
comprehensive psychological models of accident causation are
available. The majority of accidents with a 'human factors'
causation are due to an individual or team ultimately making

SPE 86624

an error. Errors take a number of different forms - mistakes,

slips, rule violations, omissions, commissions etc - (see (10)
for a fuller explanation of human error). The likelihood of
error is influenced by a number of other factors, such as the
organizational safety climate or the specific working
conditions. Reason's Swiss Cheese Model (11) is a popular
safety framework for understanding both sharp end errors and
latent organizational conditions. Widely used in the aviation
industry, is Helmreich's Threat and Error Model - it focuses on
the threats present for a given work situation (e.g. weather
conditions, faulty equipment) - error enforcing conditions and the errors that occur. This model now constitutes the
basis of a safety audit system LOSA (Line Operations Safety
Audit) adopted by a number of large airlines. For information
see website for Human Factors Group, University of Texas at
Austin, Department of Psychology.
Purchasing or Designing a Behavioural
Safety Intervention
Having identified the desired or undesired behaviours and
their context, then the objectives of the intervention should be
established. (These will be used in the evaluation phase to
determine whether the intervention has been successful or
not). At that point an appropriate intervention can be
purchased or designed. If purchasing a method, it would be
wise to check the scientific literature (see Appendix 1) to
whether this technique has been formally tested by
independent researchers (ie not only by its developers) and
what results are available. These can take the form of literature
reviews examining the empirical evidence for the
effectiveness of a particular safety technique, such as crew
resource management (12); (13). Safety interventions often
involve some kind of training to raise awareness of the critical
behaviours and may also offer opportunities for practice, such
as role play. Further opportunities for practice can be provided
in the workplace or in some cases, in a simulator.
For behaviour change to occur in the workplace, the
context has to be understood, for instance support for these
behaviours has to be provided by peers, supervisors and
managers. If workers are being trained to stop an unsafe job,
or to voice safety concerns or admit error, then this requires a
very supportive environment. Good safety programmes
sometimes fail because only the workforce were trained, and
the managers actually do not encourage or reinforce the
trained behaviours when they are enacted at the worksite.
Similarly, if supervisors are being trained in leadership
techniques that their teams to do not understand or appreciate,
then it will be very difficult to effect change. Lack of
assertiveness in co-pilots was shown to be a component cause
of hull loss accidents in the late 1970s. The airlines discovered
quickly that it was not enough to train co-pilots in
assertiveness techniques, unless they also trained captains to
listen and to encourage co-pilots to monitor and challenge the
captains actions (14)
Measurement instruments of individual differences (e.g.
personality questionnaires) which may be being considered as
part of safety training, can be checked to ascertain their
psychometric properties (15). If introducing training to
enhance safety behaviour, then this should be designed on
accepted educational principles to maximise learning (16).

SPE 86624

Evaluation
The final step in the process is to determine whether or not the
safety intervention has met the stated objectives. (17)
evaluation hierarchy which assesses effects in terms of
knowledge, attitudes, behaviours and organisational indicators,
is often used for this purpose (see for example, (18)). If a
behavioural safety intervention is to be properly evaluated,
then baseline data should be collected prior to its introduction.
This pre-testing can show levels of knowledge, attitude,
behaviour, and organisational outcomes (eg safety records)
before the safety intervention is launched. These data should
be collected again after the intervention to determine whether
any change has occurred. Another evaluation design involves
setting up a matched control group who do not receive the
intervention and their subsequent performance can be
compared with the test group. Evaluation data should be
collected not just immediately after the intervention but also 612 months later. What is important is whether behaviour
change is sustained and is being supported at the worksite.
Safety climate surveys can also be used to track whether
cultural change through the organisation is being achieved.
Continued analyses of accident and near miss data can also
indicate whether the desired behaviour change is occurring or
the problem behaviours are still in evidence.
Postscript
Human behaviour is influenced by a wide range of external
and internal factors. Silver bullet interventions for changing
safety behaviour do not exist in psychology any more than
they do in engineering. Simple behavioural safety solutions
which are claimed to fix a company's safety problems will
have been designed by someone who does not understand the
complexity of human behaviour or of organizational life.

avoidance science. Understanding response in collision

incipient conditions. Ergonomics, 46, 1111-1135.
(10) Reason, J. (1990) Human Error. Cambridge: Cambridge
University Press.
(11) Reason, J. (1997) Managing the Risks of Organizational
Accidents. Aldershot: Ashgate. (14) Jentsch, F. & SmithJentsch, K. (2001) Assertiveness and team performance. In E.
Salas, C. Bowers, & E. Edens (Eds) Improving Teamwork in
Organisations. Mahwah,NJ: LEA.
(12) OConnor, P., Flin, R. & Fletcher, G. (2002) Techniques used to
evaluate Crew Resource Management training. A literature
review. Human Factors and Aerospace Safety, 2, 217- 233.
(13) Salas, E. et al (2001) Team training in the skies. Does Crew
Resource Management (CRM) training work? Human
Factors,43,641-674.
(14) Jentsch, R. & Smith-Jentsch, K. (2001) Assertiveness and team
performance. In E. Salas, C. Bowers, & E. Edens (Eds)
Improving Teamwork in Organisations. Mahwah, NJ: LEA.
(15) Lindley, P. (2001) (Ed). Review of Personality Assessment
Instruments (Level B) for use in Occupational Settings.
Leicester: British Psychological Society.
(16) Goldstein, I. & Ford, K. (2002) Training in Organizations. (4th
ed). Belmont, CA : Wadsworth.
(17) Kirkpatrick, D. (1998) Evaluating Training Programs.The Four
Levels. (2nd ed.) San Francisco: Berrett-Koehler.
(18) O'Connor, P., Flin, R., Fletcher, G. & Hemsley, P. (2002)
Methods used to evaluate the effectiveness of flightcrew CRM
training in the UK aviation industry. Human Factors and
Aerospace Safety,2, 235-255.

References
(1) OGP (2001) Human Factors. London: International Association
of Oil and Gas Producers.
(2) Kirwan, B. & Ainsworth. P. (1992) Introduction to Task
Analysis. London: Taylor and Francis.
(3) Seamster, T., Redding, R. & Kaempf, G. (1997) Applied
Cognitive Task Analysis in Aviation. Aldershot: Ashgate.
(4) Swezey, R., Owens, J. Bergendy, M. & Salas, E. (1998) Task and
training requirements methodology. Ergonomics, 41, 16781697.
(5) Mearns, K., Whitaker, S. & Flin, R. (2001) Benchmarking safety
climate in hazardous environmenst. Risk Analysis, 21, 771-786.
(6) Flin, R., Mearns, K., O'Connor, P. & Bryden, R. (2000) Safety
climte: Identifying the common features. Safety Science, 34,
177-192.
(7) Strauch, B. (2001) Investigating Human Error. Incidents,
Accidents and Complex Systems. Aldershot: Ashgate.
(8) Gordon, R., Flin, R. & Mearns, K. (2001) Collecting human
factors data from accidents and incidents. SPE Production and
Facilities, May, 73-83.
(9) Hancock, P. & De Ridden, S. (2002) Behavioural accident

Appendix 1
Relevant scientific journals human behaviour and safety.
Accident Analysis and Prevention
Ergonomics
Human Factors
Human Factors and Aerospace Safety
International Journal of Aviation Psychology
Journal of Applied Psychology
Journal of Safety Research
Quality and Safety in Healthcare
Safety Science
Work and Stress

SPE 86624

Figure 1 Diagnosis, Design and Evaluation Process

Organisational climate

Diagnostic
Research

Design Safety
Intervention

Accident
analysis

Organisational climate

Analysis of
safe perfmance

Implement and
Evaluate

Positive loop

Refine

Near-miss or
accident

Negative loop