HCI Research Project. Research Paradigms Theoretical (in the style of mathematics) –Mathematical...
25
HCI Research HCI Research Project Project
HCI Research Project. Research Paradigms Theoretical (in the style of mathematics) –Mathematical deduction –Simulation –Analysis of algorithms The researcher:
Research Paradigms Theoretical (in the style of mathematics)
Mathematical deduction Simulation Analysis of algorithms The
researcher: Checks many examples Conjectures a result in the form
of a theorem Constructs the proof of the theorem from known
results
Slide 3
Research Paradigms (2) Experimental (in the style of
psychology) Hypotheses checked with controlled experimentation The
researcher: Poses hypotheses Extensive background reading
Appropriate theoretical model Conducts controlled experiments
Performs statistical analysis of data Draws conclusions
Slide 4
Sample Research Questions Why dont typical users invoke
advanced search strategies? What is the preferred method of making
a choice from a lengthy list? Note behavior of scroll bar, key
index What is the best semantics for the Tab key operating on a Web
form? What is the best shopping cart checkout design?
Slide 5
Sample Research Questions (2) What features should pie menus
have in a gesture-based interface? How should information scent be
shown on search reports? What intelligent agents are most helpful
in a tiny interface? What ZUI widgets are most helpful for
information retrieval?
Slide 6
Questions to Hypotheses Key is measurability Identify
Independent variable(s): those things the researcher can control
Dependent variables: those things the researcher can measure Aside:
What is measurable?
Slide 7
Measuring Things Objectively Time: to do, to correct, to learn
Counts: tasks performed, errors made Distance, or other geometrical
attributes Text: density, complexity, reading level Images: size,
color Subjectively Preference Rating scale
Slide 8
Good Experimental Design Limit number of independent variables
Limit number of values for each independent variable Make dependent
variables easy to measure Develop good working definitions Control
for external bias Guard against learning and practice
Slide 9
Good Experimental Design (2) Base hypotheses on theoretical
model Phrase hypotheses in measurable terms Choose subjects
representative of larger group
Slide 10
Example 1 From Goren-Bar, What Do Users Prefer Hypothesis 1: .
users in general will find the user-oriented search method as more
suitable for searching the web, compared to search methods that do
not consider the user perspective.
Slide 11
Example 1 (continued) Hypothesis 2: Novice and knowledgeable
users will not have the same response to both search
approaches.
Slide 12
Example 2 From St. Amant, Navigation and Intelligent Assistance
Assume that the precision and recall of an intelligent assistant
can be controlled. Hypothesis 1: Is any assistance better than no
assistance?
Slide 13
Example 2 (continued) Hypothesis 2: Does task performance
change with the precision of the assistance? (Precision =
proportion of assistance that is relevant) Hypothesis 3: Does task
performance change with the recall of the assistance? (Recall =
proportion of relevant information that is provided)
Slide 14
Example 3 From Basgodan, Touch in Shared Virtual Environments
Hypothesis: Touch is an essential element in performing a task in a
shared virtual environment.
Slide 15
Human Subjects Informed consent Minimized risks Institutional
Review Board (IRB) Expedited review Application form Application
form Application form
Slide 16
Organization Teams of two or three Membership approved by
instructor Members choose leader
Slide 17
Items to be Submitted Initial phase Team membership preferences
Hypotheses, independent and dependent variables Bibliography IRB
approval formincludes experimental design and materials Oral
presentation of project
Slide 18
Items to be Submitted (2) Investigative phase Progress reports
as required Final phase Draft of Introduction, Background and
Theoretical Foundation Draft of Results, Analysis, Conclusion Final
report: oral and written Poster abstract and design
Slide 19
Research Ideas
Slide 20
Turning Issues into Research 1. Need earlier prompts with some
car navigation systems 2. Help systems need to be helpful 3. Should
order of input matter? 4. Should the system respond after it
receives input from the user? 5. Speech recognition systems are
poor, not very accurate
Slide 21
Issues into Research (2) 6. Time delays in responding to input
are annoying, may have high error cost. 7. Controls should match
mental models 8. Systems should display current data and use a
continuous update strategy 9. Error messages should make sense to
the user.
Slide 22
Research Questions What is the task analysis of purchasing an
airplane ticket? How well is it supported by an airline website as
displayed on a smart phone? How should a PDA support a shopping
list? How can an email interface support task management?
Slide 23
Questions (2) How do people manage their digital photographs?
What is the best interface for a GPS- based navigation system? What
are the features of a good proximity-aware recommender system?
Slide 24
Questions (3) What design features promote the success of
reprogrammable keys? Or of soft keyboards? How fast should
automatic scrolling of information on limited size screens
proceed?
Slide 25
Sources for More Ideas CHI Conference Proceedings ACM
Transactions on HCI Course web site