How accessible is Facebook for users with disabilities? We have set up three tests to find out. In our first test, we look at the Facebook registration process.
This is a translation of the German article covering part 1 of our Facebook accessibility test. The article provides a summary of the problems identified. See also part 2 Account and Profile Settings. Part 3 will follow in July and look at the core functions of Facebook, finding 'friends' and networking.
We were covering the German registration process, which may differ from other language versions of Facebook. The detailed test report is available only in German.
Facebook start page (www.facebook.com)
To keep an accessibility test of a complex web application like Facebook manageable, we have decided to conduct three tests instead of one, focussing in turn on different processes:
It is clear that even these three separate tests cannot fully cover Facebook. However, we believe that this approach will give a fair impression of the accessibility of Facebook for an average user.
Many people use Facebook without relying on the Facebook web site. Instead, they use apps on smartphones oder interfaces offered by other social media platforms like Twitter. While we are aware of these alternatives, we do not cover them in our tests: we just focus on the Facebook web site.
The test instrument used, the German BITV-Test, is an expert test based on requirements of the German regulation BITV, which mirrors WCAG requirements. (There is an overview of the BITV-Test which describes it in some detail.)
Carrying out the BITV-Test usually does not involve practical tests with users. This time, we have also talked to several blind users of Facebook to put our test results into a broader context.
The Facebook registration process is relatively straightforward. Still, altogether six pages or views have to be passed through in order to create a profile. Our test of the registration process focused merely on the creation of the profile, it did not cover the input of additional personal data or configuration settings (this will be the focus of further tests in June and July).
We have selected three pages from the registration process:
The overall test result arrives at just 74.75 of 100 possible points, which in terms of the BITV-Test means that the Facebook registration process is "badly accessible". A range of problems contribute to the low score.
In the course of registering, solving a CAPTCHA is mandatory. As an alternative to the graphical CAPTCHA, an audio CAPTCHA is available. The blind Facebook users we interviewed mostly used Webvisum (www.webvisum.com/) to solve the CAPTCHA. Webvisum ist eine Firefox extension for solving graphical CAPTCHAs (probably based on human assistance). Audio CAPTCHAs seem to be rather unpopular with blind users due to the difficulty of extracting the information from the noisy background. Naturally, this is no a Facebook-specific problem.
When using the keyboard instead of the mouse, focus visibility is non-existent or very weak. The focus order is also often disorienting.
An example for bad focus order is the lightbox that takes care of uploading a profile image. After calling up the lightbox, the focus is set to the cancel button ("Abbrechen"). The next tab moves the focus into the dimmed page content underneath the lightbox. To use the upload function you either have to tab backwards from "Abbrechen" or all the way forwards through the dimmed page content. This is likely to cause problems for many keyboard users.
In contrast, the many important popout menus in the navigation area at the top of pages (Profil, Konto and Benachrichtigungen) are relatively easy to use with a keyboard.
The semantic structure provided by headlines ist insufficient and gives a somewhat random impression. On none of the pages investigated, the heading structure is adequately used to support blind users in navigating content and functions. The use of WAI-ARIA (see below) cannot (yet) compensate these deficiencies.
Text alternatives are often missing or deficient. This is especially true for the frequent icons implemented as background graphics (such as icons für notifications). For users of custom color schemes, these icons and many important functions with them disappear without a trace. This is mainly a problem for users with visual impairments; screen reader users can rely on an accessible text replacement.
Looking at "normal" graphics, i.e., decorative and layout graphics, many have no
alt attribute at all (instead of the required empty
Text scaling does not work in Internet Explorer since the text size has been defined using
px. In Firefox, text scaling does not pose problems apart from minimal content overlaps. Zoom magnification works equally well in both browsers.
Error identification in response to user input shows critical flaws. If a user offers a password that is deemed too short by Facebook, the input error is displayed below the form. This causes problems for users who magnify content in the browser or use screen magnifying software, and equally for Screenreader users: both may simply not become aware of the error message.
All accessibility problems are presented in more detail in the German test report.
Due to time constraints we were unable to include the mobile version in our test. A cursory assessment indicates that no major accessibility problems should be expected. Looking at the HTML code, it appears that no semantic elements are used for structuring page content. The low complexity of the mobile pages suggest that this is likely to be a minor problem.
In comparison to the desktop version, the mobile site has a much simpler structure. For this reason, it is the preferred choice for many blind users.
Facebook uses WAI-ARIA markup, both for document landmarks and for marking up diverse roles and states. Currently WAI-ARIA is not sufficiently well supported by many Browsers and assistive technologies (compare our article at "A list apart" from November 2010: The accessibility of WAI-ARIA). For blind users with WAI-ARIA support, the accessibility of the application is significantly better.
The accessibility of the Facebook registration process is not great: the test results in just 74.75 of a total score of 100. At least the registration process does not pose insurmountable barriers. Obviously, since we have focused exclusively on the initial registration process for new users, the result does not allow an overall conclusion regarding the accessiblity of Facebook.
Part 2 the Facebook test looks at Account and Profile Settings, while the third part will look at important functions in the day-to-day use of Facebook: Finding friends and customising settings (including pivacy settings).
Internet address: www.facebook.com
Date of testing: 30. May 2011
Test result: 74.75 of 100 Points (badly accessible)