might address the inaccessibility of an individual app, but a
population perspective on app accessibility might reveal the causes of
systemic problems and suggest potential solutions. An example
would be discovering that a widely-used interface toolkit was
responsible for inaccessible widgets used in many apps.
We propose an epidemiology-inspired framework for the examination
of mobile app accessibility. We emphasize that this metaphor
supports the social model of disability [29]. App accessibility is a
community responsibility, as captured by our multi-factor framing
(see Figure 1) that guides how different parts of the community can
contribute to app accessibility. As more companies invest resources
into accessibility and more researchers investigate app accessibility,
it becomes increasingly beneficial to have a conceptual framework
from which to guide thought and action. Conceptual frameworks
(e.g., [7]) give a common vocabulary to ground discussion, guide
efforts to improve accessibility with known strategies, and
illuminate opportunities not previously considered. We acknowledge
that the concepts in our framework are numerous, but we believe
that this is indicative of the richness of the framework and of its
potential to inspire and inform thought and action.
Adapting a model from epidemiology [34], Figure 1 illustrates
many factors that act upon an app during its creation, distribution,
maintenance, and usage. These factors range from intrinsic factors
that are tightly encapsulated within each individual app to extrinsic
factors that indirectly but influentially affect app populations.
Example factors, listed from intrinsic to extrinsic, include source
code, visual design, development and testing tools, operating
systems, assistive technologies used, app popularity, company and
government policies, and public opinions. As this framing
exemplifies, apps do not exist independently of one another or of
their environments. A natural extension is to recognize that neither
do their accessibility strengths or weaknesses. Understanding how
these factors interact and influence the accessibility of apps over
time can help in improving app accessibility through development
of preventative measures and post-release repairs [37].
Developing an understanding of how a variety of factors contribute
to app accessibility requires recognizing the value of varying levels
of analysis, from individual entities to populations at specific
moments and over time. Many well-established scientific disciplines
have benefitted from longitudinal population-level analyses, such
as ecology [33], oceanography [22], and computer security [9]. As
stated, we chose epidemiology [20] as our metaphor for our app
accessibility framework. We construct our epidemiology-inspired
framework and, although no metaphor is perfect or without limitations,
we advance the claim that the study of app accessibility can benefit
from epidemiology’s well-developed language and approach to
collecting, analyzing, and acting upon longitudinal multi-factor
population-based data. To the best of our knowledge, ours is the
first attempt to frame app accessibility as a “population science.”
To put our conceptual framework through its paces, we apply it in
an analysis of accessibility barriers in popular Android apps
available on the Google Play Store. We analyze a sample of 100
apps for nine determinants, or causes, of a variety of
“inaccessibility diseases” using Google’s Accessibility Scanner
[19]. We present the prevalence of different determinants,
motivated by the objective of “Determining the Extent of the
Disease” in the population (see Section 4.2). We then reflect on
how our framework and preliminary data informs future work.
Our research contributions are twofold:
• A novel conceptual framework for monitoring, analyzing, and
acting upon longitudinal multi-factor large-scale data on mobile
app accessibility. Our framework highlights wide-ranging
intrinsic and extrinsic factors that influence app accessibility,
motivates the collection and analysis of large-scale data, and
guides opportunities for enhancing treatments for app
“diseases” of inaccessibility.
• Empirical results from a framework-guided analysis of a
stratified sample of 100 apps from the Google Play Store.
Motivated to determine the extent of the disease in the
population, we found high prevalence with 100% of apps
having an “inaccessibility disease” based on the nine
determinants scanned for.
2. REAL-WORLD EXAMPLES
We present two real-world examples of population-level factors
that influence app accessibility and how they fit into the
epidemiology-inspired framework. The examples present elements
that are infectious agents that carry inaccessibility within Android
Studio [2] and Android’s Floating Action Button design tutorial
[17]. These examples both exemplify how factors apart from
developer-written source code affect the likelihood of an
accessibility barrier.
2.1 Android Studio App Designer
Android Studio is the development environment released by
Google for creating Android apps and is one of the most popular
tools for Android developers. Due to its widespread use, the
accessibility of the Android features it provides has a large impact
on the accessibility of the whole population of Android apps.
Android Studio includes a drag-and-drop WYSIWYG Layout Editor.
The editor provides basic widgets including Image Buttons for
common functionality, such as a star icon for “favoriting.” When
an icon button is dragged onto an interface, Android Studio
generates the layout code in a separate file that defines the button,
its layout size, and other basic features. A notable omission within
the generated code is the Content Description, the field a screen
reader uses to describe an icon button. If that field is not manually
added by the developer, the app will have an inaccessible button for
people using a screen reader. As an approach to addressing this,
when the content description field is non-existent or has no content,
Android Studio will issue a warning that provides options to:
(1) guide the developer to add the field, or (2) set a flag to ignore
all warnings of that type.
This example illustrates the transmission of an inaccessible button
disease from an infectious agent, the icon button, to a host app. The
determinant, or cause of the disease, is the missing content
description. Our epidemiology-inspired framework then motivates an
in-depth analysis to evaluate the existing preventative treatment of
the warning to determine if it is sufficient at preventing the spread
of the disease.
2.2 Android Floating Action Button Design
Floating action buttons are a part of Google’s Material Design for
Android, a guide for a more unified design in Android apps [24].
Floating action buttons are already being adopted in popular apps
such as Skype, Gmail, Facebook Messenger, and Dropbox. Floating
action buttons, however, are potential infectious agents carrying
inaccessibility. These buttons are typically separate from standard
menu bars, “floating” in a visually prominent location, such as in
the bottom-right corner of the screen, highlighting the most
important action. Android provides design guidelines for how to
employ these buttons in interfaces, including outlining how the
buttons should look, act, animate, and function. Such buttons might
become accessibility barriers to people who are blind or have a
Session: Interaction Techniques and Frameworks
ASSETS'17, Oct. 29–Nov. 1, 2017, Baltimore, MD, USA