Using engineering psychology in design

1
Using engineering psychology in design

• Presentation to Engineering Psychology class
• University of Toronto
• January 2004
• Paul McInerney
• IBM DB2 UCD

2
Outline

• What is Usability and UCD (user-centered design)?
• Examples
• Progress indication
• Automated advisors

3
What is usability?

• Usability
• The effectiveness, efficiency, and satisfaction
  with which a specified set of users can complete
  a specified set of tasks in a particular
  environment (International Standards
  Organization)
• example A certified database administrator in a
  large bank can install IBM DB2 and make a
  network connection between the client and the
  server.
• User centered design
• An approach to achieving usability which
  involves (1) design of the total user experience
  and (2) a multidisciplinary team.
• Similar concepts
• user-centered design, human-computer interaction,
  user experience design, human factors,
  engineering psychology, ease of use

4
Discipline and industry presence

• Associations, e.g.,
• Usability Professionals Association (UPA) -
  www.upassoc.org
• ACM SIG CHI (Computer-Human Interaction) -
  www.sigchi.org
• U of T HFIG
• Standards - ISO, etc.
• Education - courses, degrees, textbooks
• Research topic
• e.g., International J of HCI

5
UCD at IBM (1) - www.ibm.com/easy
6
UCD at IBM (2)

• Corporate vs. project teams
• Corporate team tools, method development
• Project-specific UCD departments, e.g., DB2 UCD
• Research teams
• Leadership
• VP for Ease of Use
• UCD Corporate Lead
• Management process
• Ease-of-use objectives
• Monthly executive reviews of metrics

7
UCD on a product development team

• Development (coding)
• Project Management
• Requirements Management
• Testing
• User Manual Writing
• User Centered Design
• ....

8
What does a UCD team do?

• Activities to... Influence product direction
• define ease-of-use objectives for the release
• periodically assess achievement of objectives
• Activities to... Understand users and the
  competition
• user and task analyses - customer visits,
  interviews
• scenario generation
• competitor product assessment
• Activities to... Design product externals
• lead design for selected GUIs
• contributor, consultant, reviewer of selected
  product designs
• Activities to... Evaluate design
• user sessions design explorations, evaluations,
  validations, beta

9
Skills and knowledge for UCD

• concepts, facts and methods from human
  engineering, human factors
• product development process
• application domain
• user experience technology, e.g., GUI vs. web
  pages
• Questions and Comments?

10
Progress Indication

• Further reading
• www.ibm.com/developerworks/web/library/us-progind/

11
Examples

12
Progress indication in everyday life

• Q How much longer until we arrive at our
  destination?
• A It normally takes two hours but there is a
  traffic jam ahead.
• A I have no idea Ive never driven there
  before.
• A We've been traveling for 1 hour.
• A We are 70 of the way there.
• A There are 4 roads on our route and we're on
  the 2nd road.
• A I estimate we'll get their in 39 minutes and
  14 seconds. ....lttwo minutes latergt we'll be
  there is 15 minutes and 34 seconds.
• A ltno answergt
• A We should get there with 30 minutes to spare.
• Lessons
• insight from other domains
• some answers are better than others
• providing an answer requires modeling things we
  dont control

13
Engineering psychology of waiting

• Impacts of poor waiting experiences
• Drop-off
• Dissatisfaction
• Dysfunctional behaviour, e.g., repeatedly press
  Submit button
• Wasted time
• Attributes of good waiting experiences
• Known end point
• Linear progress for predictable end-point
• Consistent waiting period
• Periodic reassurance that things are proceeding
  normally
• Indication when things are NOT proceeding
  normally
• Ability to end

14
Sample guidelines

• Time remaining is better than time elapsed
• e.g., 3 minutes remaining vs. 4 minutes elapsed
• For time estimates, use human-scale precision
• e.g., less than 1 minute remaining vs. 34.5
  seconds remaining
• e.g., about 4 hours vs. 3 hours and 54 minutes
• Total progress is better than progress on current
  step
• Progress bar design
• start at 1 complete, not 0
• don't display 100 until ready to exit
• show smooth, linear progress

15
Challenges in getting progress information

• Designing the display is the easy part getting
  the information to drive the display is the hard
  part
• Getting worker components to report progress
• impact on performance
• knowing remaining work
• converting work remaining into time remaining,
  e.g., 500 records left to load
• etc.
• Aggregating progress of individual worker
  components
• individual workers active in sequence
• lack of common infrastructure
• lack of common work metrics, e.g., 5 records
  created vs. 5 objects created
• etc.

16
Commentary

• Where design insight/guidelines come from
• familiarity with research on related domain
  (waiting in line-ups)
• reasoning based on general knowledge of
  engineering psychology to cover gaps
• observations and testing of existing good and
  poor designs
• design team discussions of design options
• Design bloopers
• Questions and Comments?

17
Advisor tools
18
Advisors Examples and Overview

• Examples
• An IT specialist wants to adjust settings to make
  a system run faster.
• A researcher needs to select the appropriate
  statistical test to analyze some data.
• A grandmother wants advice on buying the
  appropriate computer model.
• Overview
• Tool that provides advice on selecting a course
  of action among a set of alternatives
• Most advisor attributes/design questions are
  invariant across domains

19
Exercise

• Application domain
• Selecting a cruise ship itinerary
• Sample user problem (scenario)
• Laura wants to plan a world tour cruise to stop
  at particular ports of call. She wants to travel
  about 100 days starting in March spend less than
  40K, and take no more than 3 different cruises.
• Cruise advisor preliminary design
• Input/select desired ports of call, etc.
• Advisor displays the cruise with the best fit
  with a button to Book tickets now
• Instructions
• Suggest improvements to the preliminary design

20
Design issues

• Design concept decisions
• scope - advise only or implement recommendation
  too
• idiom - wizard vs. other tool
• expertise high vs. low broad vs. narrow range
• heavyweight or lightweight functionality
• business goals e.g., reduce costs of customer
  rep. staff
• User experience success factors
• support the full task flow
• design for trustworthiness and credibility
• design for range of expertise

21
Essential task flow of advisor tools

• Describe problem/choice/material to analyze to
  advisor
• Describe priorities and constraints
• tune tradeoffs - define "best"
• tune tool behaviour (e.g., algorithm used,
  resource limits)
• Wait for advisor to generate recommendations
  (online vs. not)
• Review recommendations
• review recommendations for acceptability
• assess validity/trust consider rationale, work
  done
• Subsequent tasks
• iterate/what-if (repeat steps 1 to 4)
• save recommendations for later review or action
• act on recommendation now

22
Designing recommendations that will be accepted

• Provide some options leave final decision to
  user
• Provide pros and cons of recommendation
• Provide ancillary information
• Explanation for recommendation
• Amount of work completed
• Access to rejected options
• Avoid recommendations that look funny

23
Commentary

• Advisor task flow pattern as reusable knowledge
  asset
• an entire fill in the blanks design solution
  template but..
• requires appropriate skill/training to apply
• limited by the cases envisioned by the pattern
  designer
• tool for evaluating designs and predicting
  usability problems
• Questions and Comments?
• Exercise discussion
• Did you find the guidance helpful to improving
  your original list of design suggestions?
• Other comments or questions?

24
Summary

• Common design assignments benefit from
  psychological insight
• waiting gt progress indication
• trust and reliance on automation gt advisors
• Product design requires
• going beyond the research
• addressing technical constraints
• understanding the application domain