HTA As a Framework for Task Analysis

1
HTA As a Framework for Task Analysis

• A. Shepherd 1998
• Brenda Crook

2
2 Themes

• Justification of HTA as a framework for analyzing
  tasks rather than method for modeling behavior
• Treatment of cognitive task elements within this
  framework

3
Hierarchical Task Analysis (HTA)

• Process of exploring tasks through a hierarchy of
  goals and plans
• Task problem facing an operator
• Goal what a person is expected to do
• Plans conditions when subordinate goals should
  be carried out
• Distinguishes between the task and the behavior
  recruited to carry out the task

4
HTA

• Task Analysis identifies strategies by which the
  operator can be facilitated to solve the problem
• Influenced by Systems Thinking
• System any complex of inter-related parts
• Characteristics
• Functions purpose the system serves in broad
  context
• System components/sub-components and their
  interaction
• Hierarchical description of structure of
  components

5
HTA Systems Thinking Influence

• Hierarchical System Description Can Be
• Just a description of system at some level of
  detail
• To imply degree of control in which higher levels
  in the hierarchy influence behavior of lower
  levels
• HTA supports Systems-Influenced model of human
  behavior through
• Concepts of Feedback and Control
• Pursuit of Goal through adaptive behavior
• Control of Behavior through Feedback

6
HTA Feedback and Control

• Miller et al (1960) proposed TOTE
• Describes relationship between action and
  feedback in controlling performance
• T Operator Tests environment against specified
  criterion
• O if test is mismatch, Operates to affect
  change
• T Tests again to check if discrepancy still
  exists
• E Exits if match accomplished
• O Operates again if match not accomplished

7
HTA Feedback and Control

• Annett and Duncan (1967) proposed IAF
• Feedback and control in approach to task analysis
• I Input
• A Action
• F Feedback
• Failure to carry out an operation successfully
  will be due to failure of handling I A or F

8
HTA Production Systems

• TOTE and IAF can be compared to production
  systems used in cognitive modeling
• Production condition-action statement
• Condition T (in TOTE) or I or F (in IAF)
• Action O (in TOTE) or A (in IAF)
• Production is triggered when circumstances match
  condition
• When a condition consistently matches a
  specified pattern, the action should follow
• Feedback different set of conditions warrants a
  different action
• Connection between task analysis and cognitive
  modeling

9
HTA Analytical Strategy vs Model

• Systematic examination of tasks to establish
  hypotheses to enable goals to be attained
• HTA as model of behavior suggests
• Everyone organizes task data the same way
• Utilization information in pursuit of goals is
  consistent across all circumstances
• Doesnt address changes in mental organization of
  task information as more or less skill is achieved

10
HTA Strategic Framework

• 9 Elements
• Relies upon other methods and perspectives to
  enable completion
• Systematic strategy important particularly in
  large multiple-activity tasks
• Hierarchical representation of task
• Ensures fit of task and behavior to system goals
• Identifies adjustments to task or task affordance
  to support behavior capabilities
• Identifies modifications in behavior to meet task
  demands

11
HTA Strategic Framework

• Setting Goals
• Identify main work goal associated with problem
• Observing Constraints
• Constraints associated with goal attainment
• Can be environmental, management, design
• Calculating Criticality
• Importance of goal and reliability of execution
• Can be subjective or data-based
• Only examine critical task elements

12
HTA Strategic Framework

• Ceasing Redescription
• Analysis stops for non-critical tasks identified
  in 3
• Generating Hypotheses if performance is not
  satisfactory
• Examine operator-system interaction
• Human error analysis, model behavior to determine
  weaknesses
• Make design suggestions
• Observe constraints throughout
• Cost-Benefit Analysis
• Hypotheses costs
• Changes as project/analysis progresses

13
HTA Strategic Framework

• Recording the Analysis
• Provides documentation of progress, how task was
  represented, proposed design solutions, rejected
  hypotheses
• Redescription
• May not have design hypotheses
• Re-examine task in terms of sub-systems/subordinat
  e operations
• Challenging Constraints
• When no hypotheses can be established

14
HTA as Strategy vs Behavior Model

• Framework that relates various stages of decision
  making encountered in a task
• Generalized task analysis strategy that employs a
  variety of methods as part of task analysis vs.
  as a precursor
• Hierarchy of operations and plans and thus
  distinguishes task from behavior

15
HTA and Cognition Analysis

• Preece et al (1994) distinction between HTA and
  Cognitive Task Analysis (CTA)
• HTA establishes accurate description of steps
  required to complete a task
• CTA represents knowledge that needed to
  complete the task
• Argument successful performance in all tasks
  depends upon interaction between physical and
  cognitive elements

16
CTA in HTA Framework

• CTA methods can be utilized within HTA frame
• GOMS Goals, Operations, Methods, Selection rules
• TAG Task, Action, Grammar
• TAKD Task Analysis for Knowledge Description
• Use within HTA frame
• economy of effort
• Ensures examination of task elements not done in
  vacuum

17
CTA in HTA Framework

• HTA supports recent approaches to cognitive
  modeling unitary theories of cognitive
  architecture (Hjaer and Hansen 1995)
• Role of knowledge within processes of learning,
  inference, and decision making in producing
  behaviors
• Manner in which information is acquired, stored,
  organized and retrieved to account for decision
  making and performance across range of task
  elements
• Demonstrates need to account for behaviors that
  interact across task hierarchy
• Mapping cognitive task elements onto cognitive
  skills difficult
• Physical tasks are visible

18
CTA in HTA Framework

• HTA supports CTA through inferring cognitive task
  elements onto cognitive skills by stating plans
• Decision process can be inferred
• Different actions consistently carried out in
  different circumstances
• Monitoring can be inferred
• Operator jumps to action following period of
  inactivity
• Diagnosis, Rectification, Compensation, Recovery
• Carrying out tasks to maintain system status

19
HTA Advantages in CTA

• Can identify what is supposed to be achieved
• Task must first be understood to anticipate
  demands on cognition
• Cognitive task variation based on demands placed
  on individual
• Can establish interaction between cognition and
  action
• Can identify inter-relationships between
  cognitive skills underlying and interacting with
  different cognitive task elements

20
HTA Advantages in CTA

• Can Identify task elements in context
• Factors that influence performance of a cognitive
  task element
• Goal context situational factors such as crisis
• Frequency, Predictability, and Coincidence
• Predictable events can be automated
• Determine unpredictable events and when they may
  occur together increased cognitive demands may
  limit performance
• Priming prompts to operator for next step
• Decision Outcome and Criticality

21
Survey Results in Quality and Practicability in
Task Analysis

• Ainsworth and Marshall 1998

22
Quality and Practicality in Task Analysis

• Surveyed Defense and Nuclear Industries
• Purpose
• Examine ways in which TAs were conducted
• Examine extent of variability between TAs
• Examine suitability of different TA methods

23
Defense Industry Results

• 27 reports obtained
• Use of TA method
• Depended on purpose of TA
• Multiple methods often listed
• Ranged across eight different TA methods
• HTA, cognitive modeling, scenario-specific
  modeling, workload analysis, qualitative
  assessments, link analysis, functional flow
  analysis, taxonomies and error analysis

24
Defense Industry Results

• Data Collection
• Data sources not clearly defined affects
  confidence of findings and recommendations
• Mostly used sequential task lists little HTA for
  complex tasks order
• Good use in general of decomposition tables
• More detailed decomposition detailed
  recommendations
• High reliance on subjective workload analysis to
  determine functional allocations
• of limited use by itself

25
Defense Industry Results

• Graphical Techniques
• Not always used effectively
• Confusion about techniques
• Failures to exploit representations
• Inappropriate scales
• Too complex diagrams trivial/missed
  relationships
• Data Presentation
• Analysts not identified/ SME background unknown
• ? Basis of recommendations opinion vs.
  understanding

26
Defense Industry Results

• Data Presentation (cont.)
• Level of detail of analysis inconsistent
• Leads to difficulty in prioritizing
  recommendations/making alternate resolutions
  difficult
• Methodology for task timings not reported

27
Nuclear Industry Results

• 70 reports
• Each TA categorized by respondents into one of
  six phases of system life cycle
• Concept, feasibility, project definition,
  detailed design, acceptance/commissioning,
  in-service
• TA not used much during concept phase
• Defense industry tended to use TA earlier in
  design process
• Depth of analysis high when HF expertise used

28
Nuclear Industry Results

• TA Method
• HTA widely used
• Task plans with text vs flow diagrams
• Complex plans resulted in unclear diagrams
• Task decomposition in over half
• More complex decomposition more detailed recs
• 12 TAs were task descriptions from written
  procedures
• Task steps missed/unchecked
• Recs tended to be generic and without
  psychological insights

29
Nuclear Industry Results

• Data Collection
• Most popular were walk/talk-throughs, informal
  discussions
• Effective use and format
• Scenario specific focus
• Timelines used widely
• Use was good
• Basis for timing unclear in some
• Error analysis depth/scope varied
• Data Presentation
• Variable presentation of reports, some with no
  HTA diagrams
• No structured overview of task
• Model used by analyst not illustrated

30
Conclusions

• Over 100 TAs examined in 2 surveys
• Provided useful insights into ways that TAs are
  utilized by organizations
• Identified areas of improvement

31
Conclusions

• Differences between defense and nuclear industry
  use of TAs
• Presentation of TA objectives not clear to
  defense analyst
• Depth of analysis/level of HF expertise was
  greater in the nuclear industry

32
Recommendations

• Define objectives of TA clearly
• Avoids superficial/costly study
• Report methods and data sources
• Enables user to assess weight of recommendations
• Ensure task analysis for complex tasks is
  accomplished by someone with psychological
  expertise
• Analysis more insightful
• Develop guidelines to address various stages of
  conducting a TA
• Advice should be systematic vs prescriptive/inflex
  ible rules

33
Questions/Discussion?