Transcript Slides - Course notes are available at York only

Access to the web for disabled
and older people
Helen Petrie
Advanced Topics in HCIT
© 2014 Department of Computer Science
Overview
• Why worry about accessibility?
• Understanding users with disabilities
relevant to Web and multimedia access
• some basic statistics and characteristics
• how they use computer technologies,
the Internet
• problems they have in using the Web
and multimedia applications
Advanced Topics in HCIT
© 2014 Department of Computer Science
Why worry about accessibility?
• The legal case
– national legislation
– European directives etc
– selling to the USA: ADA and Section 508
• The business case (particularly the older
market)
• The moral case
• The usability bonus
Advanced Topics in HCIT
© 2014 Department of Computer Science
Who has difficulties with the Web?
Important to remember that it is not just people with
visual impairments who have difficulty with the
Web/e-services:
Visual impairments Hearing impairments -
Physical impairments Cognitive impairments,
Advanced Topics in HCIT
© 2014 Department of Computer Science
blind
partially sighted
deaf
deafened
hard of hearing
dexterity problems
people with specific learning
difficulties
dyslexia
Visually impaired people – some stats and facts
• about 2 million people with severe sight problems
in the UK (source: Royal National Institute for the
Blind) in a population of 60 million (i.e. 1 in every
30 people)
• 82% over the age of 65, but 166,000 people of
working age and 24,000 children
• only 10% see “nothing”
• all want to use any vision they have
• Only 5% of totally blind people read Braille
Advanced Topics in HCIT
© 2014 Department of Computer Science
Some myths about blind people
• Very few have guide dogs (4,700 in the
UK)
• They do not have super acute hearing
or sense of smell (but they do make
much better uses of these senses than
sighted people typically do)
• They are not inherently musically
• They “watch” television as much as
sighted people of similar characteristics
(source: RNIB)
Advanced Topics in HCIT
© 2014 Department of Computer Science
Visually impaired people would like ..
• Also worth noting that visually impaired
people greatly appreciate audio description
of multimedia material (extra verbal
description of the visual aspects of the
material, carefully slotted into appropriate
quiet sections of the production) - currently
films, theatre, opera and television
• Our survey of 111 visually impaired people
for the RNIB found that this was one of the
most appreciated “technological” aids
Advanced Topics in HCIT
© 2014 Department of Computer Science
Colour vision deficiencies (“colour blindness”)
• About 10% of the population have some form
of colour vision deficiency: 8 – 12% of men
and 0.5% of women
• most common forms in perception of red and
green (see next slide)
• Aging affects our colour perception – the
average 80 year old has only 40% of the
colour vision of a 20 year old
• If three Caucasian males come and look at
your website, there’s a 22% chance that one
of them will have a colour vision deficiency
Advanced Topics in HCIT
© 2014 Department of Computer Science
Green weakness (Deuteranomoly)
• Affects 5 in 100
males
• Perception of green
as red
(missing/weak
green-sensitive
pigment)
• Poor discrimination
in the red/orange/
yellow/green region
of the spectrum
Advanced Topics in HCIT
© 2014 Department of Computer Science
Other forms of colour vision deficiency
• There are numerous other forms of colour
vision deficiency, all much rarer than
deuteranomoly
• Most severe – monochromaticism, no colour
vision at all
• None of these people see the world in what
people with full colour vision would think of as
black and white – some distinctions not
made, others made in different ways
• a monochromat may actually see distinctions
a full colour vision person does not see,
based on differences in saturation
Advanced Topics in HCIT
© 2014 Department of Computer Science
Blind people: technologies for computer and
internet access
• People with no vision use screenreaders to access
computer applications, including the Web
• Most common of these are JAWS, WindowEyes
and HAL (last in the UK only) - for Windows (a new
screenreader for Macs in currently under
development)
• Important to remember that these do not simply
“read the screen” (in spite of their name), but allow
the user to interact with the information
• e.g. provide list of links, try to make use of the
mark-up structure, provide a known set of
commands
Advanced Topics in HCIT
© 2014 Department of Computer Science
Blind people - technologies for computer and
internet access (cont)
• Specialist Web browsers for visually impaired
people have also been developed – PW
WebSpeak, IBM’s HomePage Reader
• Instead of using screenreader with standard
browser (e.g. Internet Explorer), can use a
specialist browser
• Haven’t taken off – people are used to the
commands of their screenreader, don’t want
to learn another set; screenreader developers
caught up with the specialist browser
developers
Advanced Topics in HCIT
© 2014 Department of Computer Science
Problems encountered by blind people when
using the Web (from our DRC study)
• Incompatibilities between screenreading
programs and web sites (poor coding ….)
• Incorrect or non-existent labelling of links,
form elements and frames
• Unclear and confusing layout of pages
• Images not labelled (with ALT text) or labels
unhelpful
• Confusing and disorienting navigation
mechanisms
Advanced Topics in HCIT
© 2014 Department of Computer Science
Partially sighted people - technologies for
computer and internet access
Partially sighted people may use a screen
magnification program such as ZoomText or
use the accessibility settings in their browser
to enlarge text, change colours of
text/background etc
People with colour vision deficiencies largely
use no assistive technologies, so special care
needs to be taken to ensure appropriate use
of colour in applications and websites
Advanced Topics in HCIT
© 2014 Department of Computer Science
Problems encountered by partially sighted people
when using the Web (from our DRC study)
• Inappropriate use of colours and poor
contrast between content and background
• Incompatibilities between screen
magnification programs and web sites (poor
coding ….) [B]
• Unclear and confusing layout of pages [B]
• Confusing and disorienting navigation
mechanisms [B]
• Graphics and text size too small
Advanced Topics in HCIT
© 2014 Department of Computer Science
Deaf and hard of hearing people – stats
and facts
• There are about 9 million deaf and hard of hearing
people in the UK (source: Royal National Institute of
Deaf People)
• 8.8 million have an acquired hearing loss - they have
become hard of hearing usually through aging
• 120,000 people are prelingually deaf - born deaf or
became deaf early in life; may have difficulty with
written English
• 50,000 people use a sign language as their preferred
language; see themselves as a linguistic and cultural
minority, rather than a disability group; English is a
second language
Advanced Topics in HCIT
© 2014 Department of Computer Science
Deaf and hard of hearing people –
technologies for accessing the Internet
• Currently unlikely to use any assistive
technology on the Web
• Greatly appreciate signing and subtitling of
film, theatre, opera and television
• People from the Sign language community
really want interpretation “from the deaf
viewpoint”
• Even some signing is appreciated - in our
MultiReader Project we provided signing of
the navigation icons, which was greatly
appreciated
Advanced Topics in HCIT
© 2014 Department of Computer Science
Problems encountered by hearing impaired people
when using the Web (from our DRC study)
• Unclear and confusing layout of pages [B,
PS]
• Confusing and disorienting navigation
mechanisms [B, PS]
• Lack of alternative media for audio-based
information and complex terms/language
• Inappropriate use of colours and poor
contrast between content and background
[PS]
• Graphics and text size too small [PS]
Advanced Topics in HCIT
© 2014 Department of Computer Science
Dyslexia
• causes difficulties in learning to read, write and spell
• Short-term memory, mathematics, concentration, personal
organisation and sequencing are not dyslexia, but may cooccur with this
• usually arises from a weakness in the processing of
language-based information
• Biological in origin, it tends to run in families, but
environmental factors also contribute
• can occur at any level of intellectual ability
• not the result of poor motivation, emotional disturbance,
sensory impairment or lack of opportunities
• effects can be largely overcome by skilled specialist
teaching and the use of compensatory strategies
Advanced Topics in HCIT
© 2014 Department of Computer Science
Typical characteristics of dyslexia
•
•
•
•
•
•
•
•
bright in some ways with a 'block' in others
late in learning to talk, or with speaking clearly
particular difficulty with reading or spelling
put figures or letters the wrong way e.g. 15 for 51, 6
for 9, b for d, was for saw
spell a word several different ways without
recognising the correct version
confuse left and right
answer questions orally but have difficulty writing the
answer
have trouble with sounds in words, e.g. poor sense of
rhyme
Advanced Topics in HCIT
© 2014 Department of Computer Science
People with dyslexia – stats and facts
• About 10% of the population have some form of
dyslexia
• About 4% of the population are severely affected,
including 375,000 schoolchildren in the UK
• Dyslexia causes problems with reading, spelling,
writing, orientation and navigation (including on web
sites)
• Reading problems: words may appear to jump
around the page/screen, bleed down the page,
disappear, the relationship between the written form
and the spoken form is very difficult to grasp
Advanced Topics in HCIT
© 2014 Department of Computer Science
People with dyslexia – technologies
• People may use software such as SpeakOut
and textHELP!
• But often don’t use assistive technology for
the Web, so need to be able to adapt
appearance of pages
• Short lines of text, different colour
combinations and highlighting of text can help
deal with some of these problems
• Text-to-speech synthesis can help people
with severe dyslexia
Advanced Topics in HCIT
© 2014 Department of Computer Science
Problems encountered by people with dyslexia
when using the Web (from our DRC study)
• Unclear and confusing layout of pages [B,
PS, HoH]
• Confusing and disorienting navigation
mechanisms [B, PS, HoH]
• Inappropriate use of colours and poor
contrast between content and background
[PS, HoH]
• Graphics and text size too small [PS, HoH]
• Complicated language or terminology
Advanced Topics in HCIT
© 2014 Department of Computer Science
People with physical impairments
• People who have difficulty using a keyboard
and particularly a mouse (or other pointing
device) have problems using the Web
• These physical impairments can result from
many conditions from RSI to spinal injuries
• People may use special keyboard and
pointing devices
• People may use switch access and head
mounted pointing devices (see next page)
Advanced Topics in HCIT
© 2014 Department of Computer Science
Input and pointing devices for people with
physical impairments
Advanced Topics in HCIT
© 2014 Department of Computer Science
Problems encountered by people with physical
impairments (from our DRC study)
• Confusing and disorienting navigation
mechanisms [B, PS, HoH, Dys]
• Unclear and confusing layout of pages [B,
PS, HoH, Dys]
• Graphics and text size too small [PS, HoH,
Dys]
• Inappropriate use of colours and poor
contrast between content and background
[PS, HoH, Dys]
Advanced Topics in HCIT
© 2014 Department of Computer Science
The Disability Rights Commission Formal
Investigation
• I lead a team that conducted the most detailed study
in the work to date on web accessibility
• we were appointed to conduct the research for a
Formal Investigation of Web accessibility for the
Disability Rights Commission
• Accessibility of web sites potentially comes under the
Disability Discrimination Act (1995) – although the
original act doesn’t mention web sites, the DRC’s
Code of Practice (2002) made it clear that provision
of information via the Web is a service
Advanced Topics in HCIT
© 2014 Department of Computer Science
Overview of the FI research
• User Panel of 50 people – blind, partially
sighted, deaf, hard of hearing, dyslexic
• Focus groups with User Panel members
• Automated testing of 1000 home pages
• In-depth automated testing and user testing
of 100 websites
• Small controlled study with blind and sighted
users
• Surveys of Web developers and owners
Advanced Topics in HCIT
© 2014 Department of Computer Science
Sample of 1000 websites
•
Took a representative sample of websites of
interest and importance to disabled people
in Great Britain
•
Five main categories:





Government and information
Businesses (SMEs to multinationals)
E-commerce (banking, travel, retail…)
Entertainment and leisure
Web services (ISPs, portals, search engines,
chat rooms …)
Advanced Topics in HCIT
© 2014 Department of Computer Science
Automated testing
• Automated testing of the home pages of
the 1000 sites
• Criteria: WCAG 1 - those items in the
Guidelines which can be checked
automatically
This covers 12 out
of 65 Checkpoints
Used WebXM from
Watchfire
Advanced Topics in HCIT
© 2014 Department of Computer Science
Results of automated testing
• 19% of home pages (192) passed the
automatic Priority 1 checks, so less than 19%
would be fully Priority 1 compliant (WAI A
Conformance)
• 51% of government Website home pages
passed automatic Priority 1 checks
• Significantly better than the rest of the sample
• No other differences between the sectors
Advanced Topics in HCIT
© 2014 Department of Computer Science
Results of automated testing
• only 6 (0.6%) of all home pages passed Priority
1 + Priority 2 automatic checks
• But only 2 (0.2%) passed both automatic and
manual checks at Priority 1 + Priority 2 (AA
Conformance)
• No home pages passed Priority 1 + Priority 2 +
Priority 3 (even automatic checks) (AAA
Conformance)
• No substantial differences between
government/other areas on AA and AAA results
Conclusion: basic technical accessibility very poor
Advanced Topics in HCIT
© 2014 Department of Computer Science
Measuring accessibility in more detail
• also developed two more detailed measures of website
accessibility:
• Designer metric = number of different Checkpoints
violated
• (relates to how many different things the designer needs
to address)
• User metric = number of instances of Checkpoint
violations
relates to the number of problems a user may experience on
a page; of course an individual disabled user isn’t
affected by all checkpoints and unlikely to read the entire
page, but measures the potential number of problems
Advanced Topics in HCIT
© 2014 Department of Computer Science
Example of the two new metrics
Different Checkpoints violated on a particular page:
1.1 provide alt text on images
12.1 title each frame
Designer metric = 2
Instances of Checkpoint violations on a particular page:
no alt text on images
10
frames not titled
3
User metric = 13
Advanced Topics in HCIT
© 2014 Department of Computer Science
Results on the metrics
Designer metric
a mean of 7.7 different Checkpoints violated/home page
plus a mean of 39 different Checkpoints warnings
[Government websites: 6.0 (+ 35 warnings)]
User metric
a mean of 108 instances of violations per home page
plus a mean of 239 instances of warnings
Government websites: 60 instances of violations (+ 195
warnings)
Conclusion: no wonder disabled people have
problems with the Web
Advanced Topics in HCIT
© 2014 Department of Computer Science
In-depth automated and user testing of 100
Websites
Selected 100 websites from the 1000 on
the basis of a number of measures:
the 5 categories
use of different Web technologies
accessibility level on automated
testing
Advanced Topics in HCIT
© 2014 Department of Computer Science
More automated testing
• Automated testing of whole site or the
first 500 pages in the site
• In total we conducted automated testing
on 39,000 web pages
Advanced Topics in HCIT
© 2014 Department of Computer Science
User evaluations
• Fairly standard user testing – concurrent verbal
protocols
• Because of the amount of data that we wanted to
collect in the time available, couldn’t do all the testing
with a researcher in the lab
• One session in the lab evaluating 2 – 3 web sites,
and teaching participants how to conduct an
evaluation on their own
• Each participant evaluated 7 – 8 web sites alone,
making a total of 10 evaluations per participant
Advanced Topics in HCIT
© 2014 Department of Computer Science
User evaluations
• target was 1000 tasks = 50 users x 10 sites x 2 tasks
• Each Web site evaluated by at least 5 users, one
each from the different user groups
• 913 tasks actually attempted, logged and analysed
• We compared data collected in the lab to data
collected alone, and found no differences in effects,
although the quantity of data collected when working
alone was smaller
Advanced Topics in HCIT
© 2014 Department of Computer Science
Results: Success at tasks
• Overall, panel members were successful on (only) 76% of
the tasks
• but also significant differences between impairment
groups:
blind successful on 53%
partially sighted: 76%
dyslexic: 83%
Deaf/hearing impaired/physically impaired: 85%
Advanced Topics in HCIT
© 2014 Department of Computer Science
Ease of task ratings
Difficult
Blind
PS
Dyslexic
Deaf/HH
PI
61%
48
43
36
29
….
Easy
32%
51
53
63
64
So, partially sighted and dyslexic people also
disadvantaged, although not as severely as
blind people
Advanced Topics in HCIT
© 2014 Department of Computer Science
Accessibility problems the users encountered
585 instances of problems were encountered
55% (319) would probably have been avoided if
WCAG had been followed, but 45% not (266)
Guidelines are NECESSARY but not
SUFFICIENT – amazing result to the Web
community, shouldn’t really surprise the HCI
community
Advanced Topics in HCIT
© 2014 Department of Computer Science
Relationship between WCAG conformance and
user testing
• We investigated the relationship between the
WCAG conformance (total number of
checkpoint violations/warnings, number of
Priority 1,2 and 3 violations/warnings,
Designer measure, User measure) and the
user testing results (success/fail, ratings of
ease of use …)
• No significant correlations at all
• This does need further work - currently data
mining this data set
Advanced Topics in HCIT
© 2014 Department of Computer Science
What is the relationship between accessibility
and usability?
“pure” accessibility – one or more disabled
groups can’t access information, able-bodied
people can
no ALT text, graphical text
blind people can’t access that
information, able-bodied people fine
Both accessibility and usability – both disabled
and able-bodied people affected, disabled
people more than able-bodied
Navigation schema on the Olympics Games
Web site
Advanced Topics in HCIT
© 2014 Department of Computer Science
Olympic Games 2004 Web site
Advanced Topics in HCIT
© 2014 Department of Computer Science
Two conceptual problems
Relationship between accessibility and
usability
Relationship between conformance to
WCAG and ability of disabled people to
use websites
Advanced Topics in HCIT
© 2014 Department of Computer Science
Definition of web accessibility
Advanced Topics in HCIT
© 2014 Department of Computer Science
The usability bonus
• We conducted a small, conducted study of
web sites
• High accessibility – performed well on WCAG
and disabled user testing
• Low accessibility – performed poorly on
WCAG and disabled user testing
• Asked a group of blind people (using
screenreaders) and a matched group of
sighted people to perform a series of
standard tasks on the web sites
Advanced Topics in HCIT
© 2014 Department of Computer Science
task completion times
Advanced Topics in HCIT
© 2014 Department of Computer Science
Results in words
• Sighted people were 35% faster on the high
accessibility web sites than on the low
accessibility web sites
• Analysis of the problems the disabled users
encounter will also reveal many/all of the
usability problems of the site
• So, investigating accessibility will yield a
usability bonus as well
Advanced Topics in HCIT
© 2014 Department of Computer Science
Conclusions
• The characteristics and needs of people with
disabilities are often different from the stereotypes we
have
• The more we understand about people’s real needs
and the ways they currently use technologies, the
easier it is to provide accessible web sites and eservices
• No-one is expecting developers to spend huge
amounts of time on this, but a little investment of time
will be well repaid
• Understanding accessibility at either a conceptual
level or a practical level is a lot more complex than
you might imagine
Advanced Topics in HCIT
© 2014 Department of Computer Science
Resources
Terminology about people with disabilities:
http://hub.eaccessplus.eu/wiki/Terminology_for_writing_
about_people_with_disabilities
Disability Rights Commission (2004). The Web: Access
and inclusion for disabled people. London: The
Stationery Office.
Petrie, H. and Bevan, N. (2009). The evaluation of
accessibility, usability and user experience. In C.
Stephanidis (Ed). The Universal Access Handbook.
London: Taylor and Francis.
Petrie, H. L. and Kheir, O. (2007). The relationship
between accessibility and usability of websites. ACM
CHI 2007 Conference on Human Factors in Computing
Systems. New York: ACM Press.
Advanced Topics in HCIT
© 2014 Department of Computer Science
More resources
Power, C., Freire, A., Petrie, H. and Swallow, D. (2012).
Guidelines are only half the story: accessibility problems
encountered by blind users on the web. Proceedings of
the 30th International Conference on Human Factors in
Computing (CHI 2012). New York: ACM Press.
Freire, A.P., Petrie, H., and Power, C. (2011). Empirical
results from an evaluation of the Accessibility of
websites by dyslexic users. In Proceedings of the
Workshop on Accessible Design in the Digital World
2011, Lisbon, Portugal, September 5, 2011.
Advanced Topics in HCIT
© 2014 Department of Computer Science