Accessibility of mobile applications: evaluation by users with visual impairment and by automated tools

Providing accessible mobile applications to people with visual disabilities demands appropriate evaluation techniques and tools to identify problems during the design of such systems. Automated accessibility evaluation tools are important to support evaluation tasks and to make evaluators more productive to perform repetitive analyses. However, automated tools cannot find alone all problems that users encounter in accessibility evaluations of mobile applications. Despite previous investigations on the coverage of accessibility problems encountered by automated tools on websites, there is little knowledge about the relationship between the problems encountered by those tools and problems faced by users with visual impairments in mobile applications. This paper presents a study comparing issues encountered by the automated tools MATE (Mobile Accessibility Testing) and Accessibility Scanner with a set of 415 instances of accessibility problems encountered in a previous user study involving six blind and five partially-sighted users on four mobile applications. The results showed that 36 types of problems were encountered only by users, tree types of problems were encountered both by users and by the tools, and 11 types of problems were encountered only by the automated tools. The results show the kinds of relevant problems that automated tools can identify, aiding in the early identification of such problems. The study also contributes to determining the types of problems that are only encountered by evaluations with users, reinforcing the importance of involving users in accessibility evaluation and characterizing the problems in mobile applications that can go unnoticed if automated tools are used alone.