An Epidemiology-inspired Large-scale Analysis of Android App Accessibility
An Epidemiology-inspired Large-scale Analysis of Android App Accessibility
复制标题
DOI：
10.1145/3348797
复制
发表时间：
2020-04-01
影响因子：
2.4
通讯作者：
Wobbrock, Jacob O.
中科院分区：
文献类型：
--
作者：
Ross, Anne Spencer;Zhang, Xiaoyi;Wobbrock, Jacob O.
移动的应用程序（app）中的无障碍可能会使有障碍或使用辅助技术的人难以使用这些应用程序。Ross等人的流行病学启发的框架强调，各种各样的因素可能会影响应用程序的可访问性，并提出大规模分析作为了解可访问性障碍的普遍性的强大工具（即，疾病）。利用这个框架，我们对免费Android应用程序进行了大规模分析，探索了可访问性障碍的频率以及可能导致障碍流行的因素。我们测试了9，999个应用程序的七个可访问性障碍：少数TalkBack可聚焦元素，缺失标签，重复标签，无信息标签，可编辑的TextView与contentDescription，完全重叠的可点击元素和尺寸过小的元素。我们首先测量了所有相关元素类和应用程序中每个可访问性障碍的普遍性。缺少标签和尺寸过小的元素是最普遍的障碍。作为衡量应用程序中障碍传播的一个指标，我们评估了五个最重复使用的元素类别，以发现缺少标签和尺寸过小的元素。Image Button类是最容易出现障碍的高重用元素类之一; 53%的Image Button元素缺少标签，40%的元素尺寸过小。我们还调查了可能导致某些类别的元素中的高障碍流行率的因素，根据先前的知识，我们的分析以及重用和障碍倾向性的指标选择示例。这些案例研究探讨：（1）少数TalkBack可聚焦元素的可访问性障碍如何与应用程序类别相关（例如，教育，娱乐）和用于实现应用程序的工具，（2）基于图像的按钮中基于标签的障碍的流行，（3）影响Radio按钮和复选框的标签和大小的设计模式，以及（4）第三方插件元素大小的可访问性影响。我们的工作描述了Android可访问性的当前状态，建议改进应用程序生态系统，并演示了可用于进一步应用程序可访问性评估的分析技术。
Accessibility barriers in mobile applications (apps) can make it challenging for people who have impairments or use assistive technology to use those apps. Ross et al.'s epidemiology-inspired framework emphasizes that a wide variety of factors may influence an app's accessibility and presents large-scale analysis as a powerful tool for understanding the prevalence of accessibility barriers (i.e., inaccessibility diseases). Drawing on this framework, we performed a large-scale analysis of free Android apps, exploring the frequency of accessibility barriers and factors that may have contributed to barrier prevalence. We tested a population of 9,999 apps for seven accessibility barriers: few TalkBack-focusable elements, missing labels, duplicate labels, uninformative labels, editable TextViews with contentDescription, fully overlapping clickable elements, and undersized elements. We began by measuring the prevalence of each accessibility barrier across all relevant element classes and apps. Missing labels and undersized elements were the most prevalent barriers. As a measure of the spread of barriers across apps, we assessed the five most reused classes of elements for missing labels and undersized elements. The Image Button class was among the most barrier-prone of the high reuse element classes; 53% of Image Button elements were missing labels and 40% were undersized. We also investigated factors that may have contributed to the high barrier prevalence in certain classes of elements, selecting examples based on prior knowledge, our analyses, and metrics of reuse and barrier-proneness. These case studies explore: (1) how the few TalkBack-focusable elements accessibility barrier relates to app category (e.g., Education, Entertainment) and the tools used to implement an app, (2) the prevalence of label-based barriers in image-based buttons, (3) design patterns that affect the labeling and size of Radio Buttons and Checkboxes, and (4) accessibility implications of the sizing of third-party plug-in elements. Our work characterizes the current state of Android accessibility, suggests improvements to the app ecosystem, and demonstrates analysis techniques that can be applied in further app accessibility assessments.