Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
1
Pre-print Version: Danger, Danger! Evaluating the Accessibility of Web-based Emergency Alert Sign-Ups in the Northeastern United States
Brian Wentz (corresponding author)Department of Management Information Systems, Shippensburg University
1871 Old Main Drive, Shippensburg, PA [email protected], 717-477-1601
Jonathan Lazar Radcliffe Institute for Advanced Study, Harvard University and
Department of Computer and Information Sciences, Towson University8000 York Road, Towson, MD 21252
[email protected], 410-704-2255
Michael SteinExecutive Director for HPOD and Visiting Professor for HLS
Cabell Professor, William & Mary Law School, Harvard University1515 Massachusetts Avenue, Cambridge, MA 02138
[email protected], 617-495-1726
And
Oluwadamilola GbenroEdwin HolandezAndrew Ramsey
(students from Frostburg State University,101 Braddock Rd, Frostburg, MD 21532)
Abstract
People with disabilities need access to emergency-related information at the
same time that the general public receives that information. Many county and municipal-
level governments suggest that citizens sign up on a web page to receive emergency
alert information. While the messages being sent out via e-mail or text message might
be accessible, the sign-up processes are often inaccessible, preventing people with
disabilities for signing up for these important information services. In this paper, all of
the county-level emergency alert sign-ups in Massachusetts, New York, and Maryland,
2
were evaluated for accessibility. A total of 156 evaluations took place (6 evaluations for
each of the 26 counties evaluated). Of the 26 counties evaluated, 21 of them had
accessibility violations. Legal, policy, and design-related implications are presented in
the following discussion.
Keywords
Web accessibility, policy, disability, compliance, Section 508, Section 504, WCAG,
emergency alerts
1. Introduction
Many municipal, county and state governments offer emergency alert services,
where citizens can sign up to receive an e-mail or text message with information about
weather, flooding, or other emergency events. Access to this emergency information is
vital for public safety. The most convenient approach to register to receive emergency
alerts is through a web-based registration form. Some local and state governments use
third-party web interfaces to manage the registration and account information for their
citizens. It is important to evaluate the accessibility of the web-based registration
processes for people with disabilities, since historically, people with disabilities are often
not considered when electronically communicating emergency information to the public
(Waterstone and Stein, 2006). When planning for emergencies, the technologies used
by government and emergency respondents (such as GIS mapping), often do not
include any information about the location of people with disabilities, disability-related
barriers, or organizations that serve people with disabilities (Enders and Brandt, 2007).
3
It has been informally reported that many emergency alert systems have sign-up
processes that are inaccessible to many people with disabilities. In one high-profile
instance, the home page of FEMA, the Federal Emergency Management Agency, had
been inaccessible to blind people who use screen reader technology (Olalere and
Lazar, 2011), but it has since been fixed. The purpose of this paper is to discuss the
legal status of accessibility of web-based emergency alert sign-ups, and then evaluate
all of the county-level emergency alert sign-ups of three states in the Northeastern US.
1.1 Background Literature on Web Accessibility
People with various disabilities often use different types of assistive technology to
access web-based information. For instance, blind users may utilize a screen reader,
which will take what appears on the computer screen, and provide computer-
synthesized speech output. Deaf or hard of hearing users may utilize captioning or
transcripts instead of audio. People with motor impairments that limit use of their hands
may use a keyboard, but not use a pointing device (such as a mouse), may use an
adaptive keyboard or may use no keyboard at all, instead using speech recognition or
head tracking to control their computer (Lazar, 2007). Web site designers are not
expected to design different web site versions for each disability population, nor are
they expected to add different features for each disability group. A set of international
technical standards for making web sites accessible for people with disabilities, called
the Web Content Accessibility Guidelines (WCAG), has been in existence since 1999.
These technical standards cover all perceptual and motor impairments as well as some
cognitive impairments, and are internationally considered the “gold standard” for making
4
web sites accessible (Loiacano, Romano, and McCoy, 2009). Most countries have laws
or regulations related to disability access to Internet content, which either are technically
identical to the WCAG, or are derived from the WCAG with only minor differences
(Lazar and Wentz, 2012). For instance, in the United States, the US Access Board
defines engineering specifications (and the related regulations) for disability access,
including both physical architectural access and also access to web content. The first
versions of the technical specifications for US federal government web content, which
are the regulations for Section 508 of the Rehabilitation Act, were derived from WCAG
1.0. A new version of WCAG, 2.0 was officially released in 2008, and the US federal
Government is currently going through a rulemaking process to update the Section 508
regulations (Olalere and Lazar, 2011). In the most recent draft, the US Access Board
has indicated the new version of Section 508 (known as the “508 Refresh”) will refer
directly to the international standard WCAG 2.0.
Technical guidelines for web accessibility have existed for over a decade, and
there is currently a wealth of information available to web developers, that explain how
to make their web sites accessible. Furthermore, making web sites accessible is not
technically hard to do, especially for the simple web-based forms typically used for
emergency alert sign-ups. However, numerous studies have reported that U.S.
government web sites, at state and national levels, are inaccessible (Fagen and Fagen,
2004; Jackson-Sanborn, Odess-Harnish, and Warren, 2002; Jaeger, 2006; Lazar et al.,
2010; Loiacono, McCoy, and Chin, 2005; Olalere and Lazar, 2011; Yu and Parmanto,
2001). Governments around the world have had varying levels of success with web
5
accessibility (Goodwin et al., 2011), and the approaches that seem to lead to higher
levels of compliance include either massive automated monitoring of government web
site accessibility (Mirri, Muratoir, and Salomoni, 2011) or public posting of accessibility
results on a regular basis (Gulliksen et al. 2010).
Given the gap between existing knowledge and technical ability, and actual
practice, Vint Cerf, the president of the Association for Computing Machinery, even
wrote an article asking “Why is Accessibility So Hard?” (Cerf, 2012). Numerous reasons
have been presented as possible explanations for such a low level of government web
accessibility. These explanations include: a gap of almost 10 years in compliance
activities at the federal level, a lack of a requirement to document activities related to
accessibility compliance, clear technical guidelines, but no guidelines related to process
or procedures, and accessibility compliance responsibilities being added on to
government employees who already have full-time jobs (no resources or time provided
for compliance activities) (Olalere and Lazar, 2011). Often, there is more expertise
about IT accessibility at federal and state levels, as compared to local levels of
government, such as towns, cities, and counties. Yet, the average citizen interacts more
often with their local government (for water bills, fire and ambulance service, public
schools, public libraries, trash collection, etc.) than their state of federal government
(Lazar and Wentz, 2012). Although state IT accessibility can sometimes rival federal IT
accessibility (Yu and Parmanto, 2011), at no point has there been any documentation of
local (city or county) IT being superior in accessibility. So, in some ways, it is not
surprising that local governments may have challenges in IT accessibility.
6
While there is a lot of published research about inaccessible web sites in general,
there are no published studies about accessibility of emergency-related information via
electronic means. However, in the following section, there are references to research
relating to the more general topic of legal issues related to providing emergency-related
information to people with disabilities.
1.2 Background literature on emergency information access for people with
disabilities
Both the Rehabilitation Act and the Americans with Disabilities Act (ADA) prohibit
state and local governments in the US from discriminating against individuals with
disabilities. Section 504 of the Rehabilitation Act bans “any program or activity receiving
Federal financial assistance” from excluding equal participation by people with
disabilities in funded programming (US Department of Justice, 2012). Title II of the ADA
declares that “public services and programs must be accessible to people with
disabilities” (US Department of Justice, 2008).
Consequently, courts have found that not including people with disabilities in
disaster preparation and evacuation plans violates both those federal laws. For
example, in two recent federal court cases, a California district court held that the City of
Los Angeles violated both federal laws by failing to adequately serve the needs of some
800,000 individuals with disabilities through its emergency preparedness program (US
Department of Justice, 2011), and a New York district court certified a class action
against the City of New York on behalf of some 900,000 people with disabilities who
were not sufficiently accommodated within disaster plans (US District Court, 2012).
7
Nevertheless, state and local governments have been sorely remiss as far as
including people with disabilities in disaster preparedness. In April 2005, for instance,
before either hurricane Rita or Katrina, the National Council on Disability released a
report that examined the disaster experiences of people with disabilities and concluded
that access to emergency public warnings did not satisfactorily include individuals with
visual or hearing impairments. The report noted specific examples of such failures—
including the lack of closed captioning during the September 11 attacks—and
underscored that, although emergency e-mail and wireless network alerts can be
helpful, they were not being used (Frieden, 2005). This situation existed despite an
Executive Order issued by then-President Bush requiring State and local governments
to design and implement emergency evacuation plans for persons with disabilities (Lord
and Stein, 2010).
The glaring gap in inclusive preparedness for the disability sector tragically
manifested in grievous harm following hurricanes Rita and Katrina. People with
disabilities were not adequately warned of the impending disasters, were not taken
sufficiently into account as part of emergency evacuation plans, and were not
accommodated post-disaster in government sponsored relief efforts (Waterstone and
Stein, 2006). The federal government has subsequently responded to this egregious
oversight by establishing an Interagency Coordinating Council on Emergency
Preparedness and Individuals with Disabilities, periodic reviews of emergency
preparedness by the Department of Homeland Security, and the creation by President
Obama of a disability focal point position at FEMA (Federal Emergency Management
Agency). The change in FEMA is obvious when you consider that since 2010, the Office
8
of Disability Integration and Coordination has grown from one disability coordinator to a
staff of over 70 disability integration advisors working to improve coordination and
communication for people with disabilities, before, during and after emergencies. While
FEMA is a U.S. federal-level agency, it is important to clarify that the emergency alerts
referred to in this report are implemented at the local government level, not the federal
level (Federal Emergency Management Agency, 2013).
2. Research Methods
There are typically two different approaches for evaluating the accessibility of
web sites accurately: expert inspections and user testing. Expert inspections involve
experts in accessibility, using a structured method to inspect a series of web pages
against guidelines. Usability (user) testing involves people with disabilities attempting to
complete representative tasks. User testing is generally more effective for assessing
accessibility, especially when the site focuses on performing transactions involving a
series of steps, such as signing up for an e-mail account, submitting an employment
application, purchasing from an e-commerce site, or signing up for a service. Expert
reviews are more effective for assessing the compliance of individual web pages, with
laws, regulations, or guidelines. User tests involve attempting to complete tasks, while
expert reviews involve checking code for every specific interface guideline.
Because signing up for emergency alerts involves a series of steps to complete a
transaction, user testing would typically be the most appropriate evaluation method.
However, there were a number of complications with using traditional usability testing
for this evaluation project. In usability testing, users with disabilities would attempt to
9
perform tasks, but it is always important to protect the anonymity of the participants in
the usability testing. In other words, no one who participated in usability testing should
have their participation publicly known. In most types of usability testing, the identity of
participants is protected by using fake names, postal addresses, phone numbers, and
e-mail addresses anytime that personal information is requested (Lazar, Olalere, and
Wentz, 2012). However, most usability testing does not involve signing up for services
on government web sites where it could be legally problematic to provide false
information. Furthermore, many of these emergency alert sign-ups are limited to
residents of the respective counties or states, which means that if you are not a county
resident or do not have a work address in the county, you may not sign up for
emergency alerts.
Therefore, this accessibility evaluation used a hybrid approach, using aspects of
both a usability test and an expert inspection. Most emergency sign-up pages have
multiple areas of content. Only the steps specifically involved in signing up for
emergency alerts were evaluated in this project; the other content on each web site was
not evaluated. The interface experts (researchers) could not actually sign up for the
services using false identification, since that would potentially be against the law, so the
evaluation process stopped short of submitting the registration information. A task-
focused inspection (often known as a cognitive walkthrough) was conducted (Brajnik et
al., 2012; Blackmon et al., 2002; Wharton et al., 1994), in which only the aspects related
to the task were inspected, except in this case, the researchers did not complete the
task of actually submitting information.
10
A team of evaluators, trained on how to conduct accessibility evaluations,
inspected the steps involved in signing up for an emergency alert, to determine if those
steps were compliant with Section 508 of the Rehabilitation Act. In reality, the
researchers did not conduct a full Section 508 inspection of the web pages. The focus
of this project was identifying violations specific to the sign-up process. Because of this,
the data collected in this project will not determine if the entire web pages are compliant
with Section 508. The focus of the task-based inspection included the following
research questions:
1. Is the link to the sign-up page accessible?
2. Are the descriptions accessible?
3. Are all of the form fields marked-up properly?
4. Are all buttons labeled/marked-up properly?
5. Is the notification for which fields are required accessible?
6. Is there a CAPTCHA? If so, is there an audio version?
7. Is any progress indicator accessible?
8. Are there any alternative means to register for the alert service?
For the inspections, the four most popular web browsers (IE, Chrome, Firefox
and Safari) were used. The following combination of operating systems, web browsers
and screen readers were used for the evaluations:
Windows 7 with Firefox 17 and JAWS 13 screen reader
11
Windows 8 with Chrome 23 and JAWS 14 screen reader
Windows 7 with Internet Explorer 9 and JAWS 13 screen reader
Windows 7 with Internet Explorer 9 and JAWS 14 screen reader
Mac OS 10.7.5 with Safari and VoiceOver screen reader
Blind users typically access web sites through the use of text-to-speech (or
screen reader) software, which reads the content of a web page in an audible manner to
the user in a linear fashion. Examples of screen reader software include JAWS,
WindowEyes, NVDA, and VoiceOver. Screen reader software is the dominant method
of access because Braille literacy is extremely low among blind people (Schroeder,
2006).
The researchers chose to evaluate the emergency alert sign-ups in three states:
Maryland, Massachusetts, and New York. This was done because the researchers were
based in Massachusetts and Maryland and also have partnerships in New York. The
intent was to evaluate a sample of states in the northeastern part of the US and to be
able to use the data to influence improvements in accessibility in those specific states.
Furthermore, all three states are high-value targets for terrorist attacks (considering the
April 2013 Boston marathon bombing, as well as previous attacks in New York City and
Washington, DC.) There are a total of 100 counties (or county-equivalents) within these
three states (24 in Maryland, 14 in Massachusetts, and 62 in New York). The process
for selecting which counties should be included in the evaluations included the following
steps:
12
1. Exclusion of any counties within the three states that did not have county-
level sign-ups available to the public (as of October 2012)
2. For the remaining counties that did have county-level sign-ups, exclusion
of any counties where the sign-up processes were not web-based (such
as PDF or e-mail only)
3. For the counties that did have web-based sign-ups, exclusions of any
counties where the web site sign-up process (or links) were not functional
(as of October 2012)
The conclusion of this selection process resulted in a 26 emergency alert sign-up
processes being evaluated, by each of six evaluators (for a total of 156 individual
evaluations).
In contrast to the human evaluations that took place in this study, evaluations
conducted with automated software tools (such as Deque WorldSpace, Odellus
ComplyFirst, or SSB Technologies InFocus) might be able to point out where potential
violations exist. However, automated evaluations are often not as accurate as human
evaluations, primarily because the automated tools cannot determine whether the
context is appropriate (e.g. that “button one” is not descriptive alternative text for a
graphic). Multi-stage human inspections of web pages, involving screen readers (such
as JAWS or VoiceOver) are considered to be the most accurate form of accessibility
evaluation (Mankoff et al., 2005), and that accuracy further increases when multiple
individuals evaluate the same interfaces and then combine their results into a meta-
evaluation (Lazar et al., 2010). Screen readers, while designed primarily for blind users,
are very helpful in identifying accessibility violations because they help point out where
13
web page components are not accessible for keyboard-only use, which also impacts
how people with motor impairments utilize web pages. Furthermore, in these
evaluations, coding inspections on the web pages were performed if any clarifications
were needed.
When each evaluator completed their individual evaluation of the emergency
sign-up processes, the evaluators met as a group and compared their results. The
approach of using multiple human evaluators to produce one meta-evaluation increases
the reliability and accuracy of the inspection and has been previously utilized in many
web accessibility evaluations (Lazar, et al., 2010; Lazar et al., 2012; Lazar et al., 2011;
Wentz et al., 2012). If there was a disagreement in the results comparison, the group of
evaluators re-visited the sign-up process in question and formed a consensus on the
nature or lack of a violation. Due to the dynamic nature of web-based content, it is
important to note that these evaluations took place in November and December 2012. It
is possible that the accessibility of the web-based content and sign-ups have changed
since December 2012.
The highlighted violations that resulted from the evaluations are based on the 16
guidelines set forth in Section 508 of the US Rehabilitation Act (1194.22), identified as
paragraphs “A” through paragraph “P” that focus on web site accessibility. Most US
states have technical guidelines that are identical to the technical guidelines in Section
508 (although the legal requirements and remedies may differ). Table 1 lists only the
four “paragraphs” of the Section 508 guidelines that were violated by the interfaces
evaluated and provides a short description of each guideline (note that the descriptions
are from Lazar et al., 2010, not from the law).
14
Table 1. Description of the Four Paragraphs of the Section 508 Web Accessibility
Guidelines that were violated by the Emergency Sign-up Forms
(A) Text Equivalent (have a text equivalent for any graphical elements)
(C) Use of Color (color should not be used as the only method for identifying
elements of the web page or any data)
(L) Scripting Languages (make sure that equivalents for any non-accessible
scripting are included, e.g., for those who are not using pointing devices)
(N) Online Electronic Forms (all forms must be properly labeled and accessible)
It is important to note that a number of the web sites use an external service provider or
software vendor for their emergency alert sign-ups. That is, the emergency alert web
pages were not developed in-house by the county government, but rather, were
acquired through a government procurement process. Examples of common
interfaces/systems that were used (along with the number of counties included in this
study that utilized them) include:
Cassidian Communications (2)
CODERED (7)
Connect-City (Blackboard product) (1)
CooperNotification (3)
Everbridge (4)
Hyper-reach (3)
Nixle (1)
15
3. Results
Of the 26 emergency notification sign-up forms that were evaluated, 21 had one
or more accessibility violations during the sign-up process. The aspect of the
emergency notification sign-up process that had accessibility violations on the most
sites was the indication of which form fields are required (14 sites, as illustrated by the
fifth column in Table 2). The use of color alone (par. C), inaccessible scripts (par. L),
and poorly labeled form components (par. N) were the primary problems related to the
required sign-up fields. Inaccessible progress indicators were the second most
problematic violation (13 sites, the 6th column on Table 2). The lack of adequate
alternate text (par. A) was the primary problem with the graphics-based progress
indicators. There were also problems with general form field accessibility (11 sites, as
illustrated by the fourth column in Table 2). A lack of alternate text for graphical form
components (par. A) and poorly-labeled form fields (par. N) were the primary problems
with general form field accessibility.
Table 2 provides a matrix of which sites had Section 508 paragraph violations in
the various aspects of the emergency alert sign-up process. Allegany County (MD),
Charles County (MD), West Chester County (NY), Wicomico County (MD), and the
state-wide NYAlert.gov (NY) were the only sites that had no accessibility violations
related to their emergency alert sign-up process.
16
Table 2. Matrix of Section 508 Paragraphs Violated by Sign-up Process Components (26 counties evaluated; population = 100 counties)
County Name: Sign
-up
Link
Dire
ctio
ns
Form
Fie
lds
Req
uire
d Fi
elds
Prog
ress
In
dica
tor
But
tons
CA
PTC
HA
Boston area, MA ADukes Co., MA N A L A APlymouth Co., MA N A L A AWorcester Co., MA AAllegany Co., MD No violationsAnne Arundel Co., MD N A L A ABaltimore Co., MD LCalvert Co., MD N A L A ACecil Co., MD N A L A ACharles Co., MD No violationsDorchester Co., MD AFrederick Co., MD AHarford Co., MD A C AHoward Co., MD N NMontgomery Co., MD N NPrince George’s Co., MD N NWashington Co., MD AWicomico Co., MD No violationsFranklin Co., NY CGreene Co., NY A N A L A ASuffolk Co., NY N A L A ATioga Co., NY A CWayne Co., NY A CWest Chester Co., NY No violationsNYC Area ANYalert.gov (used by other NY counties) No violations
17
Cassidian Communications designed the interface for Baltimore County (MD)
and Wicomico County (MD). The CODERED system was used by Anne Arundel County
(MD), Calvert County (MD), Cecil County (MD), Dukes County (MA), Plymouth County
(MA), Greene County (NY), and Suffolk County (NY). Connect-City from Blackboard
was only used by Harford County (MD). CooperNotification was used by Howard
County (MD), Montgomery County (MD), and Prince George’s County (MD). Everbridge
was used by Frederick County (MD), Washington County (MD), Worcester County
(MA), and the Boston area. Hyper-reach was used by Franklin County (NY), Tioga
County (NY), and Wayne County (NY). Nixle was only used by Allegany County (MD).
Charles County (MD), Dorchester County (MD), the New York City area, West Chester
County (NY), and the NYalert.gov site (for other counties in NY) all appeared to use
some type of proprietary interface design. The most widely used product (CODERED)
was also the interface with the most accessibility violations, and this interface claims to
be in use by thousands of users in all 50 US states (Emergency Communications
Network, 2012).
One example of an accessibility violation that impacted all eight sites that used
the CODERED interface was the “continue” and “edit” buttons used during the
registration process which were lacking alternate text (the continue button would read
as “continue-ns.png” to screen reader users). Another example of a violation on the
CODERED interface was the progress indicator at the top of the screen, which was
lacking alternate text to indicate to non-sighted users the progress status. Figure 1
shows a screenshot of the progress indicator, which is lacking alternate text.
18
Figure 1. The Inaccessible Progress Indicator on the CODERED Interface
One example of an accessibility violation that impacted the Baltimore County
interface (designed by Cassidian Communications) was the inaccessible pop-up
calendar script that is illustrated by Figure 2.
Figure 2. The Inaccessible Calendar Script from Cassidian Communications
19
The CooperNotification system violations included form fields on the second
page of the sign-up process that had text fields that were not properly labeled and read
as “secondary text field.” Also, when a required field was missed, the form reloaded, but
only a sighted user would readily notice the notification that shows up at the top of the
page. The notification used to indicate which fields are required was inaccessible, since
the focus for screen reader users was once again at the beginning of the form (just
below the critical information regarding required fields). Figure 3 shows a screenshot of
this accessibility violation.
Figure 3. Required Form Field Violation on CooperNotification Interface
20
The proprietary sign-up form for the New York City area had a violation relating
to an inaccessible progress indicator (missing alternate text). On the interface for Hyper-
reach (used by three counties in New York), the only indication that fields were required
was through the use of red text, as indicated by Figure 4.
Figure 4. Required Form Field Violation on Hyper-reach Interface
There were several examples of overall poor design that would result in
accessibility problems but were not actual violations. For example, the Cassidian
Communications sign-up forms for Baltimore and Wicomico Counties (MD) have no
indication that an asterisk is used to indicate required form fields (other than a visible
21
red color) until the end of the forms are reached. Figure 5 shows a screenshot of the
Baltimore County sign-up form.
Figure 5. Screenshot of the Baltimore County Sign-up Form
Washington County (MD) and Worcester County (MA) provide an audio
CAPTCHA, however the CAPTCHAs seemed to be of some proprietary nature
(designed for use with the Everbridge system) and were extremely difficult to use.
Tioga, and Wayne Counties (NY) used the Hyper-reach system, which had a graphical
22
form submit button with alternate text but no label, which caused it to not work properly
for some web browser/screen reader combinations (such as Internet Explorer and
JAWS). The New York City area notification sign-up had a form with items (such as the
combo box for SMS carrier) in the wrong tab order, which could cause problems for
screen reader users. Suffolk County (NY) provided two sign-up links, but the first
(graphical) link did not work. Table 3 shows a matrix of the violations that were
consistently seen on third-party interfaces which were used by various counties/states.
The interfaces implemented by Cassidian Communications and Nixle were the only
ones that did not seem to have a consistent Section 508 violation.
Table 3. Matrix of Section 508 Paragraphs Consistently Violated
System Name: Sign
-up
Link
Dire
ctio
ns
Form
Fi
elds
Req
uire
d Fi
elds
Prog
ress
In
dica
tor
But
tons
CA
PTC
HA
Cassidian CommunicationsCODERED N A L A AConnect-City from Blackboard C ACooperNotification N NEverbridge AHyper-reach CNixle
It should be noted that several sites provided users with an alternate means of
signing up for emergency alerts. Howard County and Prince George’s County (MD)
allowed users to sign up with a Google, Yahoo, or OpenID account, and Montgomery
County (MD) provided users with an option to sign up via text message. The Boston
23
area, the New York City area, Suffolk County (NY), and the state-wide NYalert.gov all
provided a phone number for users to call if they wish to sign up over the phone.
4. Discussion
4.1 Technical Implications
From a technical point of view, all of these emergency alert sign-ups are simple
web-based forms. The coding is not complex, and the accessibility solutions are not
complex. Unlike, for example, a complex web accessibility challenge like creating
accessible equivalents of geo-spatial interactive maps for blind users (Weir et al., 2012),
accessibility solutions for web-based forms are relatively easy. Referring to the
screenshot presented earlier in the paper as Figure 1, the progress indicator for the web
sites using the CODERED system, the actual code is as follows:
<img src="images/1.png" width="36" height="36" alt="1" align="middle" />PROVIDE
<img src="images/2w.png" width="36" height="36" alt="2" align="middle" />VERIFY
<img src="images/3w.png" width="36" height="36" alt="3" align="middle" />SUBMIT
Note that there are really two problems with this code. The images are identified with
alternative text of “1” “2” or “3” and there no way to determine which page you are on.
This can be seen with the names of the images, which are “1.png”, which is a darkened
image, and “2w.png” and “3w.png”, which are lighter, and the fact that the image is
lighter, is the only representation that you are on that specific page. To make that code
accessible, it should be changed to the following:
24
<img src="images/1.png" width="36" height="36" alt="You are currently at step 1, to
provide your information" align="middle" />PROVIDE
<img src="images/2w.png" width="36" height="36" alt="Verify" align="middle"
/>VERIFY
<img src="images/3w.png" width="36" height="36" alt="Submit" align="middle"
/>SUBMIT
To provide another example from the most commonly used interface evaluated,
one of the first screens on the CODERED interface provides a section for users to
select the types of alerts that they wish to receive. Under “Alert Types” there is a choice
for “Emergency Notifications” and also “General Notifications.” “Emergency
Notifications” is selected by default, and the user has the option of selecting or
deselecting “General Notifications.” While this seems like an obvious and non-
problematic design by all visual appearances (refer to Figure 6), a screen reader user
only hears the option to select the “General Notifications” and only hears the alternate
text “Receive community notifications” next to the text for “Emergency Notifications” with
no indication that the box in the graphic depicted is already checked. The current code
for that form: <img src="images/checkmark.png" alt="Receive community
notifications" /> could easily be modified to more accessible alternate text such as <img
src="images/checkmark.png" alt="Emergency Notifications Checkbox –selected by
default, cannot be changed" /> or similar. This is a good example of an instance where
an automated accessibility evaluation tool would consider this code to be accessible
25
(since the graphic DOES have associated alternate text), when it is really not accessible
in its context.
Figure 6. Screenshot of the Alert Types Selection Problem on CODERED
As another example not previously mentioned in the paper, the following form (Figure
7), to complete the registration for the alerts in Howard County, Maryland, requires that
you provide a home address.
26
Figure 7. Screenshot of the NotifyMeHoward Sign-up Form, Requiring a Home Address
However, when you are listening to the page using a screen reader and you hear the
edit box for the form field (which requires a home address), the screen reader says,
“supplementary text field”, not indicating what the edit box is asking for (although an
expert user might be able to navigate around the page and determine that the box is
located near text for home address). The actual page code is as follows:
<input type="text" name="info_value_2" maxlength="100" value="" size="40"
alt="Supplementary Text Field">
Furthermore, although the home address field is required, the fact that the field is
labeled as “supplementary” strongly hints that the field is not required. Yet, both of these
problems could be resolved quickly by changing that one line of code to:
<input type="text" name="info_value_2" maxlength="100" value="" size="40" alt="Home
Address Field (required)” >
This is not a complex technical fix! This is one of the most basic accessibility fixes that
can be done and would not even take 10 minutes to perform.
A final example of how easy improving accessibility would be is illustrated by the
Connect-City interface from Blackboard which tells the user that the “fields in bold are
required.” This could be changed to something as simple as the word “required” in
27
parenthesis after each required field or stating that the fields with asterisks are required.
Many of the other violations could be corrected in a similar manner to the progress
indicator example above. Simply adding understandable alternate text via the “alt”
attribute would provide a solution. Since the developers of these interfaces are using
templates which are used by many customers, this would be a relatively simple fix
which would have a broad impact, as these fixes would immediately have an impact on
hundreds of county emergency systems.
4.2 Policy implications
There are currently no other published research studies about the accessibility of
emergency alerts or online emergency information for people with disabilities, so
comparing our results with the results of others, or comparing results over time, would
not be possible. However, the literature review at the beginning of the paper discusses
the comparison with other government information presented online, which often is
inaccessible for people with disabilities. Another comparison could be made between
the easy fixes that would be required in the case of these simple emergency alert sign-
ups and the much higher level of complexity involved in other emergency information
dissemination. For instance, at the US federal level, FEMA is undergoing the process of
modernizing and updating the emergency alert system, creating the Integrated Public
Alert and Warning System (IPAWS), allowing for disseminating information via
television, radio, telephones, and other methods. IPAWS is technically far more
complex than the emergency alerts described in this paper. Yet, efforts are being made
to ensure accessibility of this updated IPAWS infrastructure.
28
Because the technical fixes described in this paper are quite easy, the question
arises, why would these problems exist? There are some potential reasons. Due to the
nature of disability rights laws, often individual people with disabilities (or advocacy
groups consisting of multiple people with disabilities) are in a position where they need
to complain about problems, rather than having government agencies be responsible for
acting proactively (Wentz, Jaeger, and Lazar, 2011). For instance, it is likely that the
counties that have accessibility problems on the emergency alert sign-ups did not
check, in advance, the systems that they acquired for accessibility. Procurement
processes (when government agencies purchase or acquire hardware or software,
when money is spent), are appropriate times to enforce accessibility, and there has
been some level of success with using procurement processes to enforce accessibility
at the federal level (Olalere and Lazar, 2011).
The advocacy efforts attempted by one of the co-authors of this paper can serve
to highlight some of the challenges involved in improving accessibility of emergency
alerts. The inaccessibility of a county-level emergency alert sign-up was originally
brought to the attention of one of the authors of this paper by a disability advocate in
their county. This disability advocate had already complained twice to the county office
of emergency affairs, which twice indicated that they did not understand the problem,
but would contact the company providing the service. A co-author of this paper
identified the specific problem, wrote up the technical solution (which was of similar
simplicity to those provided earlier in the paper), and provided the corrected code to the
county office of emergency affairs. The county office of emergency affairs indicated that
they had no ability to change the code, since the code was not located on their
29
government servers, but was located instead on the server of the software service
provider. The county further indicated that they would pass along the corrections to the
company, but when asked, indicated that the procurement contract did not mention
accessibility in any way. The software company did follow-up, indicating that they had
run their software through an automated software tool for testing accessibility, and the
tool indicated that the web pages were indeed accessible (which relates to the problem
described earlier in the paper, where automated tools can determine the context or
usefulness of labeling, only to determine if labeling is present). While the co-author of
the paper indicated to the county office of emergency affairs that an automated tool
cannot, by itself, determine if a web page is really accessible, it seems that the
knowledge to perform any other type of accessibility validation was lacking at both the
county level and at the software provider. To put it more bluntly, the county government
wanted to hear that the software is accessible, and the software provider wanted to say
that the software is accessible, and these claims from disability advocates, stating that
the software is inaccessible, seem surprising and confusing to both software provider
and government, neither of whom are familiar with the topic area.
Because of this, there are often multiple policy failures. For example, the
software providers fail to ensure accessibility of their emergency alert sign-ups, and the
county governments fail to accurately check if the emergency alert sign-ups are
accessible. While ideally more expertise on accessibility would reside at the county
level, an easier approach to improve compliance would be to involve local disability
advocates in the procurement processes and discussions at the county level. The
disability advocates would generally be happy to provide assistance and advice on the
30
accessibility of technology that is potentially being acquired at the county level. Once
software or service procurement contracts have been executed without requirements for
accessibility, it is very unlikely that the accessibility problems will be resolved.
Furthermore, the costs for retrofitting accessibility features into software are much
higher than including accessibility features in initial design (Wentz, Jaeger, and Lazar,
2011).
31
4.3 Legal implications
Because these emergency alert systems are being provided by companies
through government procurement processes, these procurement processes should be
used to enforce accessibility. The Rehabilitation Act prohibits recipients of federal funds
from discriminating against people with disabilities, and the Americans with Disabilities
Act prohibits discrimination in state and local government, so private companies under
contract with any level of government to deliver emergency alert services, may not
exclude people with disabilities by rendering those notices in inaccessible formats.
Unfortunately, although state and local government agencies themselves can easily
require service providers to comply with the law by making accessibility a stated
contractual term, private enforcement of an equal right to inclusion by people with
disabilities through lawsuits can be difficult and costly. A more expedient and efficient
way of guaranteeing legally required access begins with state and local governments
consulting with local disability rights stakeholders, advocates, and organizations from
the early stages of program design. In doing so, the needs and priorities of the group
ultimately affected will be taken into account and avoidable and costly errors curtailed
before they are entrenched. Often the government agencies just want the problem to
“go away”, so rather than investigate, if they hear that the problem is resolved, even
when it is not, they are likely to believe what they are told.
4.4 Study Limitations
While this study illustrates an exploratory examination of the lack of accessibility
on emergency alert sign-up forms for county web sites, future research projects could
32
address the limitations of the current study. One limitation in the discussion of the
implications of these findings is that the counties themselves had not yet been
contacted by the authors for comment on these accessibility problems. It might be
interesting to interview those responsible for ensuring accessibility (the ADA compliance
officer?) in each county. Each of the seven third-party companies mentioned previously
in this report were contacted (by the researchers) via publicly available contact
information on their web sites, to inquire about the accessibility of their services. Of the
three companies that responded to the researchers, all claimed to evaluate their web-
based sign-up process for accessibility compliance. The company responses were
somewhat vague and further detail was not forthcoming. The other companies did not
respond to our e-mailed request for information. Future research could pursue avenues
of discussing the accessibility evaluation processes in place by those vendors, if we
could get those vendors to share their information openly. It might also be interesting to
submit freedom of information act requests, to gain access to the procurement
contracts, and determine if the appropriate disability compliance language was included
in the contracts.
In addition, it is important to note that the models of local government are very
different in the three states evaluated. In Massachusetts, much government power is
concentrated at the town or city level, with very little control at the county level.
Maryland is the exact opposite, with much power concentrated at the county level, and
except for Baltimore city (which is not part of any county, and in a way, is considered to
be its own county), very little power is located at the town or city level. It might be
33
interesting to examine if/how government structure (strong county-level governance vs.
strong town-level governance) influences IT accessibility.
5. Conclusion
Access to emergency-related information is of the utmost importance to citizens,
yet the recent history demonstrates that people with disabilities were often left out of the
loop. Despite clear legal requirements, emergency preparedness plans often do not
properly include communication to people with disabilities. Delays in receiving such vital
information can be a life-or-death distinction. Despite the fact that the technical barriers
to accessibility in emergency alert sign-ups are very easy to solve, many problems
remain. The technical solutions to solve these problems are easy. The law and policy
solutions to address the technical problems, can be more challenging. But the first step
is bringing awareness to the problem.
This paper provided a discussion of the importance of accessible emergency
alert sign-ups, as well as accessibility inspections of 26 county and municipality
emergency alert sign-ups in Massachusetts, New York, and Maryland. Of the 26 sign-up
processes that were evaluated, 21 of the processes had accessibility violations. Since a
majority of the emergency alert sign-up system software are from external vendors,
government procurement processes need to be used more effectively to enforce
accessibility. Decision-makers in emergency preparedness should be required to be
properly trained, about the role of information technology accessibility, in
communicating emergency information to people with disabilities. The vendors also
need to receive proper training in implementing technical solutions. Once all
34
stakeholders are more aware of the problems, enforcement mechanisms need to be put
into place. Earlier in the paper, it was described how, despite being aware that there
was a problem in one county, the county officials wanted to just ignore the problem. If
either federal-level officials, or state-level officials, performed compliance and
enforcement activities, this could potentially improve the situation by bringing a potential
penalty into place. For instance, Minnesota recently implemented a law that charges a
penalty for inaccessible government documents or technologies, $500 per violation up
to a maximum of $15,000 (State of Minnesota, 2013). Providing a mechanism for a
financial penalty, specific to the accessibility of documents or technology, is an
enforcement mechanism that few other states have, and is the type of idea that might
improve the situation of inaccessible emergency alert sign-ups.
6. References
Blackmon, M., Polson, P., Kitajima, M., and Lewis, C. (2002). Cognitive Walkthrough for
the Web. Proceedings of the SIGCHI Conference on Human Factors in
Computing Systems (CHI ’02). ACM, New York, 463-470.
Brajnik, G., Yesilada, Y., and Harper, S. (2012). Is Accessibility Conformance an
Elusive Property? A Study of Validity and Reliability of WCAG 2.0. ACM
Transactions on Accessible Computing (4)2, 28 pages.
Cerf, V. (2012). Why is accessibility so hard? Communications of the ACM, 55 (11). 7.
Emergency Communications Network. (2012). Experience. Available at:
http://www.ecnetwork.com/about/experience.php
35
Enders, A. and Brandt, Z. (2007). Using geographic information system technology to
improve emergency management and disaster response for people with
disabilities. Journal of Disability Policy Studies 17(4), 223-229.
Fagen, J., & Fagen, B. (2004). An accessibility study of state legislative websites.
Government Information Quarterly, 21, 65-85.
Federal Emergency Management Administration (2013). Integrated Public Alert and
Warning System. Available at: http://www.fema.gov/integrated-public-alert-
warning-system
Frieden, L. (2005). Saving Lives: Including People with Disabilities in Emergency
Planning. Available at: http://www.ncd.gov/publications/2005/04152005
Goodwin, M., Susar, D., Nietzio, A., Snaprud, M. and Jensen, C. (2011). Global Web
Accessibility Analysis of National Government Portals and Ministry Web Sites.
Journal of Information Technology and Politics, 8 (1). 41-67.
Gulliksen, J., Axelson, H., Persson, H., and Goransson, B. (2010). Accessibility and
public policy in Sweden. Interactions, 17 (3). 26-29.
Jackson-Sanborn, E., Odess-Harnish, K., & Warren, N. (2002). Website accessibility: A
study of six genres. Library Hi Tech, 20(3), 308-317.
Jaeger, P. (2006). Assessing Section 508 compliance on federal e-Government
websites: A multi-method, user-centered evaluation of accessibility for persons
with disabilities. Government Information Quarterly, 23(2), 169-190.
Lazar, J. (ed.) (2007). Universal Usability. Chichester, UK: John Wiley and Sons.
Lazar, J., Beavan, P., Brown, J., Coffey, D., Nolf, B., Poole, R., Turk, R., Waith, V.,
Wall, T., Weber, K., and Wenger, B. (2010). Investigating the Accessibility of
36
State Government Web Sites in Maryland. In Langdon, P., Clarkson, P., and
Robinson, P. (Eds.), Designing Inclusive Interactions--Proceedings of the 2010
Cambridge Workshop on Universal Access and Assistive Technololgy. London:
Springer-Verlag, 69-78.
Lazar, J., Olalere, A., and Wentz, B. (2012). Investigating the Accessibility and Usability
of Job Application Web Sites for Blind Users. Journal of Usability Studies. Vol. 7:
No 2, 68-87.
Lazar, J. and Wentz, B. (2012). Ensuring Accessibility for People with Disabilities. in
Buie, E. and Murray, D. eds. Usability in Government Systems: User Experience
Design for Citizens and Public Servants, Morgan Kaufmann Publishers,
Waltham, MA, 191-204.
Lazar, J., Wentz, B., Akeley, C., Almulhim, M., Barmoy, S., Beavan, P., Beck, C., Blair,
A., Bortz, A., Bradley, B., Carter, M., Crouch, D., Dehmer, G., Gorman, M.,
Gregory, C. Lanier, E., McIntee, A., Nelson R., Ritgert, D., Rogers R.,
Rosenwald, S., Sullivan, S., Wells, J., Willis, C., Wingo-Jones, K., and Yatto, T.
(2012). Equal Access to Information? Evaluating the Accessibility of Public
Library Websites in the State of Maryland. In P. Langdon, J. Clarkson, J.
Robinson, J. Lazar, and A. Heylighen (Eds.), Designing Inclusive Systems:
Designing Inclusion for Real-world Applications (London: Springer, 185-194).
Lazar, J., Wentz, B., Bogdan, M., Clowney, E., Davis, M., Guiffo, J., Gunnarsson, D.,
Hanks, D., Harris, J., Holt, B., Kitchin, M., Motayne, M., Nzokou, R., Sedaghat,
L., and Stern, K. (2011). Potential Pricing Discrimination Due to Inaccessible
Web Sites. Proceedings of the INTERACT 2011, 108-114.
37
Loiacono, E., Romano, N. and McCoy, S. (2009). The state of corporate website
accessibility. Communications of the ACM, 52 (9). 128-132.
Lord, J. and Stein, M. (2010). Ensuring Respect for the Rights of People with
Disabilities, in the Human Impact of Natural Disasters: Issues for the Inquiry-
Based Classroom 77 (V. Pang, W. Fernekes, and J. Nelson, Eds.).
Mankoff, J., Fait, H., and Tran, T. (2005). Is your web page accessible?: A comparative
study of methods for assessing web page accessibility for the blind. Proceedings
of the ACM Conference on Human Factors in Computing Systems, 41-50.
Mirri, S., Muratoir, L. and Salomoni, P. (2011). Monitoring Accessibility: Large Scale
Evaluations at a Geo-Political Level. Proceedings of the ACM Conference on
Accessible Computing (ASSETS), 163-170.
Olalere, A., and Lazar, J. (2011). Accessibility of U.S. Federal Government Home
Pages: Section 508 Compliance and Site Accessibility Statements. Government
Information Quarterly 28(3), 303-309.
Schroeder, F. (2006). Braille Usage: Perspectives of Legally Blind Adults and Policy
Implications for School Administrators. Baltimore, MD, National Federation of the
Blind: Available at: https://nfb.org/braille-usage
State of Minnesota. (2013). Statutory Basis for Accessibility. Available at:
http://mn.gov/oet/policies-and-standards/accessibility/accstatutory.jsp
US Department of Justice (2012). A Guide to Disability Rights Laws. Available at:
http://www.ada.gov/cguide.htm
US Department of Justice (2008). Americans with Disabilities Act, 42 U.S.C. § 12132.
Available at: http://www.ada.gov/pubs/adastatute08.htm
38
US Department of Justice (2011). Enforcing the ADA: A Status Report from the
Department of Justice (January-March 2011). Available at:
http://www.ada.gov/janmar11.htm
US District Court (2012). Brooklyn Center for Independence of the Disabled v.
Bloomberg, 287 F.R.D. 240. Available at:
http://www.nysd.uscourts.gov/cases/show.php?db=special&id=239
Waterstone, M. and M. Stein (2006). "Emergency Preparedness and Disability." Mental
and Physical Disability Law Report 30(3), 338-339.
Wentz, B., Jaeger, P.T., and Lazar, J. (2011). Retrofitting accessibility: The legal
inequality of after-the-fact online access for persons with disabilities in the United
States. First Monday. Volume 16, Number 11 - 7 November 2011.
Weir, R., Sizemore, B., Henderson, H., Chakraborty, S., and Lazar, J. (2012).
Development and Evaluation of Sonified Weather Maps for Blind Users. In P.
Langdon, J. Clarkson, P. Robinson, J. Lazar, and A. Heylighen, (eds.) Designing
Inclusive Systems: Designing Inclusion for Real-world Applications. London:
Springer-Verlag, 75-84.
Wentz, B., Cirba, M., Kharal, N., Moran, J., and Slate, M. (2012). Evaluating the
Accessibility and Usability of Blogging Platforms for Blind Users. In P. Langdon,
J. Clarkson, P. Robinson, J. Lazar, and A. Heylighen (Eds.), Designing Inclusive
Systems (London: Springer, 43-52).
Wharton, C., Rieman, J., Lewis, C., and Polson, P. (1994). The Cognitive Walkthrough
Method: A Practitioner’s Guide. In: Nielsen, J., Mack, R. (Eds.), Usability
Inspection Methods. Wiley, New York, 105-140.
39
Yu, D. and Parmanto, B. (2011). U.S. state government websites demonstrate better in
terms of accessibility compared to federal government and commercial websites.
Government Information Quarterly, 28 (4). 484-490.