1. Usability Specifications
Howell Istance
School of Computing
(chapter 3.6 : McCracken and Wolfe )

2. Acknowledgements The majority of slides have been produced by
Daniel D. McCracken (City College of New York) and
Rosalee J. Wolfe (DePaul University) to accompany their book User-Centered Website Development (Pearson Prentice Hall, 2003)
These have been edited to provide a European context for the examples used.

3. Usability Specifications
A system specification covering usability of the form “the system shall be easy to use” is of little use
What ‘easy to use’ means is open to many different interpretations (the client may have a very idea of this from the software or site developer)
A specification which can not be tested is also of little use
We need a measurable definition of ‘usability’ based on
Performance measures
Preference (or subjective) measures

4. Performance measures: a sampling
Time to locate a book at the Barnes & Noble website
Time to fill in customer information and place order
Number of times the Back Button is used, indicating that user cannot find desired information
Number of clicks to find the time of a TV show
Percentage of tasks completed correctly
Number of calls to support line
“Number of times the search engine is used” is a useful measure, according to Jared Spool, but I didn’t list it because we don’t talk about it in the book. In watching thousands of users, Spool finds that use of the internal search engine is almost always a sign that the user is making no progress toward her goal. And internal search engines currently range from poor to terrible. (Not talking about Google, obviously.) Users have preconceptions about search, however. After a testing session a group of users was asked how they liked the search engine. Almost all said, “Oh, it was great”—even though the videotape showed they never used it, or that the site didn’t have a search engine. The number of times the search engine is used is a negative, of course.

5. Preference measures Often obtained using a Likert Scale
Strongly disagree
Disagree
Neutral
Agree
Strongly agree
It was easy to find what I wanted
It was simple to choose size and color
I could pay for my purchase quickly

6. Definition - ISO9241 / 11
Usability - degree to which specified users can achieve specified goals in a particular environment with effectiveness, efficiency and satisfaction
effectiveness - measures of accuracy and completeness of the goals achieved
efficiency - measures of resources (e.g time, money, effort) used to achieve goals
satisfaction - measures of the physical comfort and subjective acceptability of the product to its users

7. Components of a usability specification
statement of the usability attribute
statement of how it will be measured
statement of criteria that will represent attainment of the specification
statement of the subset of users to which the specification applies
statement of pre-conditions of measurement (e.g period of training
statement of the usability attribute: learnability of reference finding system
statement of how it will be measured: time to retrieve class number of a book, given its title and author
statement of criteria that will represent attainment of the specification
statement of the subset of users to which the specification applies
statement of pre-conditions of measurement (e.g period of training

8. Example of a usability specification
In the context of a library information system e.g.
Efficiency of first time use measured by retrieval time of specified information
’90 % of student users without any prior training should be able to retrieve the shelf number of a book, given the author and the title within 10 times the length of time an expert user would take to complete the same task’

9. Example of a usability specification
In the context of a library information system e.g.
Efficiency of first time use measured by retrieval time of specified information
’90 % of student users without any prior training should be able to retrieve the shelf number of a book, given the author and the title within 10 times the length of time an expert user would take to complete the same task’
Statement of usability attribute

10. Example of a usability specification
In the context of a library information system e.g.
Efficiency of first time use measured by retrieval time of specified information
’90 % of student users without any prior training should be able to retrieve the shelf number of a book, given the author and the title within 10 times the length of time an expert user would take to complete the same task’
Statement of how it will be measured

11. Example of a usability specification
In the context of a library information system e.g.
Efficiency of first time use measured by retrieval time of specified information
’90 % of student users without any prior training should be able to retrieve the shelf number of a book, given the author and the title within 10 times the length of time an expert user would take to complete the same task’
statement of the subset of users to which the specification applies

12. Example of a usability specification
In the context of a library information system e.g.
Efficiency of first time use measured by retrieval time of specified information
’90 % of student users without any prior training should be able to retrieve the shelf number of a book, given the author and the title within 10 times the length of time an expert user would take to complete the same task’
statement of criteria that will represent attainment of the specification

13. Example of a usability specification
In the context of a library information system e.g.
Efficiency of first time use measured by retrieval time of specified information
’90 % of student users without any prior training should be able to retrieve the shelf number of a book, given the author and the title within 10 times the length of time an expert user would take to complete the same task’
statement of pre-conditions of measurement

14. Testing specifications
Using a prototype, a number of people can be asked to complete the task and the following can be measured
Whether a person can find the information or not
The length of time it takes
One or more experts in the use of the system are given the same task and the time taken from them to complete the task is measured
If in a sample of 20 students, 18 or more complete the test task successfully within the specified time (ie within 10 times as long as an expert would take) then the specification has been achieved

15. Usability Engineering
extends the principles underlying usability specifications
provides techniques to direct resources to improve the system with respect to individual usability attributes
defines usability goals through metrics
sets planned levels of usability that have to be achieved
analyses the impact of possible design solutions
incorporates user defined feedback in product design
iterating through the 'design-evaluate-design' loop until planned levels are achieved 24

16. Impact Analysis Partition total time = task time + error time
error time represents potential saving if time spent in error can be removed by designing out the cause of the error
determine those errors which contribute the most to error time component
allocating resources to designing out individual errors gives then a known saving in task completion time 25

17. Strengths of Usability Engineering
agreeing on a definition of usability
setting this definition in terms of metrics and usability goals
putting usability on a par with other usability goals
providing a method for prioritising usability problems 26

18. and Weaknesses….
assumption that usability that usability can be operationalised
requirement that practitioner is familiar with laboratory methods
cost of conducting usability tests
unnaturalness of testing environment 27

19. ‘Laboratory’ vs. Real World
work context
e.g. lab. test - 6 pages
normal work - reports of over 30 pages
time context
lab test - time when task will be completed is prescribed
real world - individual has some control over scheduling
motivational context
clear differences between lab and real world
social context
lab test - no support
real world - social network of support - ‘ask my friends’ 28