Human Factors in Design

1
Human Factors in Design

• The human factors engineering process
• Research
• Human capabilities
• Human activities
• Modeling
• Human-system interaction
• Define requirements
• Information flows
• Actions
• Design
• Evaluation
• This is an iterative process

2
Tools of HF Design

• Design data
• anthropometric data, design data compendiums,
  standards, principles guidelines, etc.
• e.g., Human Engineering Design Data Digest,
  Department of Defense Human Factors Engineering
  Technical Advisory Group (April 2000).
• Research methods
• Modeling
• Define Requirements
• Based on Models and engineering analysis methods
• Workload, safety, simulation, etc.
• Use to derive system requirements

3
Building Models

• Part 1 Models good for defining system and user
  requirements
• Affinity diagrams
• Flow model
• Cultural model
• Sequence model
• Physical model
• Artifact model
• Source
• H. Beyer and K. Holtzblatt (1999), Contextual
  Design A Customer-Centered Approach to Systems
  Designs. San Francisco, CA Morgan Kaufmann.
• Note for more details on the following example,
  you should visit the InContext website at
  www.incent.com and choose the Design Resources
  link at the top, then to CDTools, then
  CDTools Resources, and finally Shopping Data
  Browser on the left.

4
Affinity Diagram a good starting point

• Use the Post-It notes to record insights and
  quotes from your observations and interviews.
• One phrase or quote per note.
• Write big enough for all to see.
• Post the notes on the wall.
• Walk the wall and rearrange the notes into like
  categories.
• When everyone is agreed on the categories, give
  each category a meaningful name and summarize the
  findings.
• Use the affinity diagram to generate design
  ideas, identify requirements, and inform other
  models.

5
Affinity Diagrams Walk the Wall
The net makes the remote present to me
Being exposed to a big, unknown set of people is
scary
I want to control who is in touch with me
I dont want to stick my neck out
I like being able to look without presenting
myself
U2 229. Browsing web doesn't require you to
present yourself, like explaining yourself over
the phone
U3 325. Conscious decision about who to be in
contact with over emailsense of power
U3 331. Afraid of what shell get back if she
posts here email address to get specific info
D1 19. She checks all her sources that will
maintain her confidentiality before calling the
company directly
D3 201. The rules about relating on the web are
unclear- the address says you can write, but then
people dont write back.
U2 236. Has a sense of ethics on the net - wont
search for .edu email addresses, only uses to
look up particular names
U2 230. Sales contacts require gearing yourself
up exhausting if you dont get through email
doesnt require that
6
Flow Model

• Draw the primary user of the system in the center
  of the page.
• Use Post-It notes to add other users, people,
  roles, and physical objects as needed to define
  flows of work and information.
• Use annotated lines and arrows to indicate flows
  of information or work.
• Indicate opportunities for breakdowns in
  communications or work flow.
• Use the model to add to or refine requirements,
  define key interactions, and identify
  communication modes and methods.

7
Flow Model example
8
Cultural Model

• Draw a circle representing the primary user in
  center.
• Draw overlapping and concentric circles
  representing other entities that affect the
  primary user.
• Draw arrows indicating influences, constraints,
  and expectations.
• Identify individual and pervasive values that
  affect how the user will approach the task.
• Use the model to define subtleties that should
  affect system design.

9
An example of a Cultural Model
Corporation
User Department
User
Lose my data, lose your job
Your fixes dont work
System Manager Group
Equipment Vendor
System

Help Line
Manager
Coworker
Expert Support
Remote
Operator
System
Problem Handling
Management
Agencies
10
Sequence Model

• Define specific steps the user goes through to
  accomplish the task.
• Identify strategies and decision points.
• Identify breakdowns that make the task difficult
  to complete.
• Where appropriate, identify options and
  alternative strategies.
• Use the model to further define requirements,
  identify design opportunities, and begin to
  define potential interaction methods.
• (Well come back to this later )

11
Physical Model

• Diagram in detail the physical space in which the
  task is performed.
• Identify both official and unofficial
  designation of locations.
• Identify paths taken through the space during
  task performance.
• Define how users use the space to accomplish the
  task.
• Identify breakdowns where the physical space
  inhibits task performance.
• Use the model to develop system design
  requirements and opportunities.

12
Physical Model
13
Artifact Model

• Draw or diagram the artifacts used to accomplish
  the task.
• Specify the users intent in using the artifact.
• If necessary, identify variants of the artifacts.
• Identify potential breakdowns where the artifact
  inhibits task performance.
• Use the model to define user requirements and
  identify potential design directions.

14
Artifact Model
15
Building Models

• Part 2 Models good for defining interaction
• HTA
• GOMS
• OFM
• Process
• Step 1 develop an understanding of the user and
  task (interviews, questionnaires, observation,
  etc.)
• Step 2 decide on a modeling framework
• Step 3 build the model
• Step 4 test/refine/modify
• (Step 5 Use the model to drive design, testing,
  etc.)

16
Making a PBJ sandwich Hierarchical Task
Analysis (HTA)

• 1. Describe top-level goal
• 0. Make a peanut butter and jelly sandwich.
• 2. Develop a plan for achieving that goal
  (including error handling)
• Plan 0 Do 1-5, in order. If some ingredient is
  missing, do 5.
• 1. Get plate and knife.
• 2. Collect ingredients.
• 3. Assemble sandwich.
• 4. Eat and enjoy.
• 5. Put ingredients away.

17
Making a PBJ sandwich Hierarchical Task
Analysis (HTA)

• 3. For each step in the plan, decide if more
  detail is required. Continue until sufficient
  detail is defined.
• e.g., for step 1
• Plan 1 Do 1-3 if no clean implements, do 4.
• 1.1 Go to cabinet and retrieve 1 plate.
• 1.2 Go to drawer and retrieve 1 knife.
• 1.3 Take knife and plate to table.
• 1.4 Wash knife/plate as necessary.

18
Your turn

• Continue the HTA for
• Plan 2
• Plan 3
• Plan 5
• Discuss which parts of your plan need more
  specification?

19
Making a PBJ sandwich GOMS (Goals, Operators,
Methods, Selection rules)

• 1. Describe top-level goal
• GOAL Make a peanut butter and jelly sandwich.
• 2. Describe a methods for achieving that goal
  (including selection rules and alternatives)
• GOAL Get plate and knife.
• GOAL Collect ingredients.
• GOAL Assemble sandwich.
• Eat and enjoy.
• GOAL Put ingredients away.

20
Making a PBJ sandwich GOMS

• 3. For each GOAL in the method, describe a more
  detailed method.
• e.g., GOAL Collect ingredients.
• GOAL Get bread.
• GOAL Get peanut butter.
• GOAL Get jelly.
• Selection Rule
• Goto refrigerator
• Goto pantry

21
Making a PBJ sandwich GOMS

• 4. Continue until desired/necessary level of
  detail.
• 5. Use the (HTA or GOMS) model to
• Design documentation.
• Predict performance.
• Evaluate device/task designs.
• Design.

22
Modeling more complex tasks the Operator
Function Model (OFM)

• Hierarchical/Heterarchical task decomposition
• Activities are decomposed hierarchically (as in
  HTA and GOMS)
• Functions - highest-level activities (e.g.,
  navigate, communicate, and aviate are pilot
  functions)
• Subfunctions - activities required to accomplish
  functions
• Task - lower level (more specific) activities to
  accomplish functions or subfunctions
• Actions - manual, cognitive, or perceptual
• Heterarchical structure accounts for concurrent
  activities, usually defined at the same level.
• Operators may choose from among these activities
  or the activities may result from system state(s).

23
OFM example Chinese dinner party

• Steps (from Mitchell, 1998)
• 1. Prepare a high-level written description of
  the system of interest

24
OFM example Chinese dinner party

• Steps (from Mitchell, 1998)
• 2. Identify the high-level activities the
  operator performs.
• 3. Define the heterarchy, specifying conditions
  for transitioning, initiating, or terminating
  activities.

25
OFM example Chinese dinner party

• 4. Define the hierarchy, including conditions
  that start or end activities.
• 5. Validate the model. (Emphasis on direct
  observation, mapping actions into the model,
  resolving discrepancies.)
• 6. Refine the model as the system evolves.

26
Summary User/Task Modeling

• What is a user/task model?
• Model - a mathematical/physical system, obeying
  specific rules and conditions, whose behavior is
  used to understand a real (physical, biological,
  human-technical, etc.) system to which it is
  analogous in certain respects.(Bailey, 1989)
• A good model is one that adequately represents
  and can be used to generate testable hypotheses
  about the underlying system.
• User/task models - specifically focus on modeling
  the users goals and objectives, as well as
  his/her understanding of the task.

27
Summary User/Task Modeling

• Dimensions of models
• Conceptual . Computational
• Descriptive ..Normative
• Levels of specificity
• Device .... task ... meta-cognitive

28
Using Models in System Design
Define Reqts.
Research
Model
Design
Evaluation

• DESIGN IS AN ITERATIVE PROCESS!!
• A basis for defining requirements
• Identify information and action requirements, as
  well as potential sources of difficulty for the
  operator/user (high workload, ambiguities, etc.),
  task importance, who else is involved, etc.
• The more detailed the model, the more useful for
  design
• A basis for test evaluation
• Heuristic evaluation does the system as
  designed support the activity you modeled?
• Defining testing procedures, metrics, etc.