UNIT 2

Human Computer Interaction

Evaluation Techniques

1
Evaluation Techniques
 Evaluation
 tests usability and functionality of system
 occurs in laboratory, field and/or in collaboration with
users
 evaluates both design and implementation
 should be considered at all stages in the design life cycle

2
 evaluation occurs throughout the software development
lifecycle
• note that we focus here on user interface evaluation
 there are two main types of evaluation:
– evaluation by system designers or experts
typically conducted early in the development lifecycle,
though may give late feedback
on system performance
– evaluation by end users
typically conducted late in the development lifecycle,
though may give early feedback on system design

3
Goals of Evaluation

 assess extent of system functionality

 assess effect of interface on user

 identify specific problems

4
 Evaluating Designs
evaluation by experts
• there are many types of evaluations conducted by experts or
system designers:

Cognitive Walkthrough,
Heuristic Evaluation,
Review-based evaluation

5
Cognitive Walkthrough
 Proposed by Polson et al.
 “walkthrough” means a sequence of steps
 evaluates design on how well it supports user in learning
task
 usually performed by expert in cognitive psychology
 Expert 'walks through' design to identify potential
problems using psychological principles
 forms used to guide analysis

6
Cognitive Walkthrough (ctd)
 For each task walkthrough considers
 what impact will interaction have on user?
 what cognitive processes are required?
 what learning problems may occur?

 Analysis focuses on goals and knowledge: does the

design lead the user to generate the correct goals?

7
Heuristic Evaluation
 Proposed by Nielsen and Molich.

 usability criteria (heuristics) are identified

 design examined by experts to see if these are
violated

 Example heuristics
 system behaviour is predictable
 system behaviour is consistent
 feedback is provided

 Heuristic evaluation `debugs' design.

8
• the 10 heuristics are:
 1. visibility of system status (i.e., inform users with
feedback)
 2. match between system and real world (i.e., “speak”
user’s language; is system
 understandable by the user?)
 3. user control and freedom (i.e., can user “undo” and
“redo” actions?)
 4. consistency and standards (does system follow
standard conventions for symbols,
 menu items, etc?)
 5. error prevention (does system make it hard for the user
to make mistakes?)

9
Cont…
 6. recognition rather than recall (does system make it
easy for the user to remember
 7. flexibility and efficiency of use (does the system let
users tailor the interface to
 accommodate frequent actions or sequences of actions,
e.g., macros?)
 8. aesthetic and minimalist design (e.g., dialogues
shouldn’t contain extra information;
 interface shouldn’t be cluttered)
 9. help users recognize, diagnose and recover from errors
(are error messages in clear language?)
 10.help and documentation (does the system have any
documentation and/or a help facility?)

10
Review-based evaluation
 Results from the literature used to support or refute
parts of design.

 Care needed to ensure results are transferable to new

design.
– population of users (novice, experts)
– Assumptions made
– Analyses performed
 Model-based evaluation
 Design rationale can also provide useful evaluation
information

11
Evaluating through user Participation
 can be done at different stages of development
 system developers can simulate missing
(undeveloped) pieces of interface using techniques
like “Wizard of Oz” where a human takes the part of
the pieces that will later be automated
 elements of user evaluation:
– styles of evaluation
– experimental evaluation design
– observational techniques
– query techniques
– evaluation through physiological responses

12
users: styles of evaluation
 laboratory studies
– conducted in controlled settings
– advantages: no distractions
– disadvantages: but no context
 field studies
– conducted “in situ”, i.e., in situated settings
– advantages: context
– disadvantages: but can add distractions

13
Laboratory studies
 Advantages:
 specialist equipment available
 uninterrupted environment

 Disadvantages:
 lack of context
 difficult to observe several users cooperating

 Appropriate
 if system location is dangerous or impractical for
constrained single user systems to allow controlled
manipulation of use

14
Field Studies
 Advantages:
 natural environment
 context retained (though observation may alter it)
 longitudinal studies possible

 Disadvantages:
 distractions
 noise

 Appropriate
 where context is crucial for longitudinal studies

15
Evaluating Implementations
 Requires an artefact:
• simulation
• prototype
• full implementation
 Types of evaluation
• Empirical or experimental methods
• Observational methods
• Query techniques

16
Experimental evaluation
 controlled evaluation of specific aspects of
interactive behaviour
 evaluator chooses hypothesis to be tested
 a number of experimental conditions are considered
which differ only in the value of some controlled
variable.
 changes in behavioural measure are attributed to
different conditions

17
Experimental factors
 Subjects
 who – representative, sufficient sample
 Variables
 things to modify and measure
 Hypothesis
 what you'd like to show
 Experimental design
 how you are going to do it

18
Variables

 independent variable (IV)

 characteristic changed to produce different conditions
 e.g. interface style, number of menu items

 dependent variable (DV)

 characteristics measured in the experiment
 e.g. time taken, number of errors.

19
Hypothesis
 prediction of outcome
 framed in terms of IV and DV

 e.g. “error rate will increase as font size decreases”

 null hypothesis:
 states no difference between conditions
 aim is to disprove this

 e.g. null hyp. = “no change with font size”

20
 Observational Methods

Think Aloud, Cooperative evaluation, Protocol analysis,

Automated analysis, Post-task walkthroughs

21
Think Aloud
 user observed performing task
 user asked to describe what he is doing and why,
what he thinks is happening etc.

 Advantages
 simplicity - requires little expertise
 can provide useful insight
 can show how system is actually use
 Disadvantages
 subjective
 selective
 act of describing may alter task performance

22
Cooperative evaluation
 variation on think aloud
 user collaborates in evaluation
 both user and evaluator can ask each other questions
throughout

 Additional advantages
 less constrained and easier to use
 user is encouraged to criticize system
 clarification possible

23
Protocol analysis
 paper and pencil – cheap, limited to writing speed
 audio – good for think aloud, difficult to match with
other protocols
 video – accurate and realistic, needs special equipment,
obtrusive
 computer logging – automatic and unobtrusive, large
amounts of data difficult to analyze
 user notebooks – coarse and subjective, useful insights,
good for longitudinal studies

 Mixed use in practice.

 audio/video transcription difficult and requires skill.
 Some automatic support tools available

24
Automated analysis – EVA
 Workplace project
 Post task walkthrough
 user reacts on action after the event
 used to fill in intention
 Advantages
 analyst has time to focus on relevant incidents
 avoid excessive interruption of task
 Disadvantages
 lack of freshness
 may be post-hoc interpretation of eventS
 Examples
 Noldus Pocket Observer XT (http://www.noldus.com)

25
Post-task Walkthroughs
 Transcript played back to participant for comment
 immediately  fresh in mind
 delayed  evaluator has time to identify questions
 useful to identify reasons for actions and alternatives
considered
 necessary in cases where think aloud is not possible

26
 Query Techniques

Interviews
Questionnaires

27
Interviews
 analyst questions user on one-to-one basis
usually based on prepared questions
 informal, subjective and relatively cheap

 Advantages
 can be varied to suit context
 issues can be explored more fully
 can elicit user views and identify unanticipated problems
 Disadvantages
 very subjective
 time consuming

28
Questionnaires
 Set of fixed questions given to users

 Advantages
 quick and reaches large user group
 can be analyzed more rigorously
 Disadvantages
 less flexible
 less probing

29
Questionnaires (ctd)
 Need careful design
 what information is required?
 how are answers to be analyzed?

 Styles of question
 general
 open-ended
 scalar
 multi-choice
 ranked

30
Choosing an Evaluation Method
 when in process: design vs. implementation
 style of evaluation: laboratory vs. field
 how objective: subjective vs. objective
 type of measures: qualitative vs. quantitative
 level of information: high level vs. low level
 level of interference: obtrusive vs. unobtrusive
 resources available: time, subjects,
equipment, expertise

31
32