Culture Representation in HRA

1
Culture Representation in HRA

• David Gertman
• Steven Novack
• Julie Marble
• December, 2006

2
Why quantify cultural aspects for systems?

• All systems are vulnerable to compromise
• The human is a significant system element
• Human behavior is more difficult to predict than
  other system elements
• Culture is a universally significant influence on
  human behavior
• To more accurately assess and estimate system
  performance requires that you understand the
  cultural context

3
Culture

• shaping of human expectations and behavior for
  a group of people based upon the collection of
  attitudes, values, beliefs, and norms for
  performance.
• Culture influences the probability that a
  particular course of action is followed

4
Cultural Aspects of IO
Effects
Scope
Level
Criteria
Duration
Degrade
Disrupt
Influence
Deny
Destroy
Deceive
Exploit
Adversary influence
Offensive Ops
Adversary influence
Military Leadership
Military Leadership
Industry
Adversary influence
Adversary influence
CMO
General Population
Targets (High Level)
Industry
Religious Groups
5
Anticipate System Behavior

• To predict you need to understand each system
  element
• Comprehend inter-relationships between system
  elements on a functional level
• Define temporal and spatial demands on each
  element
• The human response is not only a function of
  training, but reflects the extent to which a
  culture or subculture trusts that technology
• Importance of the human response increases as
  duration elongates

6
Research Objective

• Develop a method to quantitatively estimate
  cultural influence on system performance
• Address known deficiencies
• Lack of culture-specific data
• Based on US operator information and procedures
• Screening level approach to uncertainty
• General methodology applied to a wide range of
  datasets

7
History of accounting for human actions and
decisions

• Human factors was established by the military as
  a part of systems life cycle development in the
  1950s
• Human reliability analysis (HRA) has evolved from
  the US nuclear communitys need to support
  probabilistic safety analysis including risk
  significant, safety critical operator actions
• Since the 1970s, HRA data sets have been
  developed from operating experience, behavioral
  science studies, and consensus expert judgment
  methods
• Traditionally, human factors deals with
  successful performance, HRA is different
• HRA an accepted approach to quantifying human
  reliability

8
Human Reliability Analysis (HRA)

• Estimates performance in complex, critical
  systems
• Base rate of error modified by effects of
  context/performance shaping factors (PSFs)
• CAM uses a set of 8 pre-defined PSFs from HRA
• Fatigue, stress, workload, fitness for duty,
  complexity, time, ergonomics/HMI, experience and
  training and culture
• PSFs significantly affect reaction times and
  error probabilities within the system
• HRA does not assess culture other than safety
  culture
• All data derived from US operators and applied to
  US contexts
• Significant differences in communications and
  work practices due to culture, and
• Technology acceptance reflects availability and
  value

9
Populating system logic models with HRA data

• Human system performance can be characterized in
  the event trees fault trees, influence diagrams
  or event sequence diagrams
• INL prefers event sequence diagrams
• Allows visualization of multiple sequences and
  response pathways in a single diagram
• Allows for calculation of three important
  parameters time, probability, and human
  decisions and actions
• Used successfully for a number of DoD-sponsored
  efforts

10
ESD example
11
Culture

• Behavior and expectations that distinguish the
  members of one group from another
• Performance with the same equipment can vary
  widely due to culture (both safety records and
  availability)
• Influences how the actions of others are
  perceived, and response to events
• Combines with situational aspects to affect human
  decision-making and actions
• Changes the probability associated with pathways
  and values in ESDs

12
Cultural Adjustment Method (CAM)

• Enhanced Human Reliability Analysis (HRA) Method
• Produces qualified Human Error Probabilities
  (HEPs) for use in ESDs
• Provides inputs for determining ESD pathways and
  response probabilities
• Addresses uncertainty

13
CAM Overview

• The CAM approach updates the human error
  probabilites
• CAM has analysts evaluate the influence of
  culture parameters
• Power distance
• Uncertainty avoidance
• Technology acceptance
• Ratings are used to determine a modifier
  (multiplier) for the base failure rate

14
CAM Factor 1 Technology Acceptance

• Technology Acceptance Model (Davis 1989)
  Factors
• 1. Perceived ease of use of technology
• Transparency and predictability of system
• 2. Perceived usefulness of technology
• Degree users think the technology helps them work
• 3. Innovation
• Degree to which a technology affords the user
  additional or enhanced capabilities
• Does not predict use or acceptance
• 4. Degrees of acceptance
• Compliance Use because it is required
• Identification Use to fit in to a group
• Internalization Use because tool is an aid

15
CAM Factors 2 3 Power Distance and Uncertainty
Avoidance

• Hofstede
• Power Distance - how a culture distributes
  authority, superior subordinate relationships,
  equality, social distance, hierarchical aspects
• Uncertainty avoidance - extent to which a culture
  tolerate situations for which rules are ill
  defined, reliance on procedures, extent to which
  individuals are reinforced to take initiate in
  unsure situations,acceptance of uncertainty in
  everyday life
• These factors are determined to be orthogonal

16
CAM Process
17
CAM Uncertainty Approach

• Uncertainty approach consistent using small or
  large datasets (Bayesian Updating)
• Weight of evidence applied to PSFs
• Uncertainty about the value of the PSF
• The confidence in the data reflects the size of
  the underlying data set.
• Minor modification to the error rate
• Consistent with the weight of evidence concept

18
Example Application

• Facility response to loss of power involving
  SCADA
• Three different cultures considered
• US Culture, Culture A, Culture B
• Time to respond as well as the decision made on
  which action to take first is computed for the 3
  cases
• Will the linemen check the substation first or go
  directly to the facility?
• Resulting pathways could determine mission
  success
• Culture influences trust in SCADA technology
  deference to authority, and reliance on work
  practices
• HEP portion of the analysis focuses on the
  relative success or failure associated with the
  actions selected.

19
Culture effects calculated using CAM
20
Conclusion Areas for CAM application

• All pillars of IO (OPSEC,PSYOP,CNO,MILDEC, EW)
• Planning, training and execution
• CIP
• HRA for chemical and nuclear facilities
• Emergency response (DHS)