Top | About SiteAnalyzer | Past Reports | Feedback
 

The Accessibility of Disabilities Sites

I surveyed over 200 web sites belonging to organizations working in support of people with disabilities. I used as my sample the home pages provided by these organizations to an established list of disabilities resources. Discarding those pages for which the URL provided failed, I ran each page through a well accepted routine for determining compliance with the current standards for web page accessibility.

In summary, I found that nearly 65% of the sites had one or more accessibility errors. Nearly 80% had some sort of browser compatibility error that might lead to an access problem. 20% of the sites took more than 30 seconds to download, though most took much less. No site had more than three access errors, indicating that whatever problems do exist would be easy to repair.

This sample is not at all representative of nonprofit web sites as whole. The sample was deliberately selected because it would be likely to have the lowest number of problems, given the issue focus of the organizations. I will discuss the value of these numbers and of web accessibility tools in general.
 

Purposes | Methods | Results | Discussion | Conclusion
 

Purposes

This project had two objectives: (1) To test the Bobby Tool for evaluating web site accessibility and thus expand my tool kit of evaluation methods. (2) To evaluate a highly targeted group of web sites for compliance with such standards, especially in light of forthcoming U.S. requirements for sites of organizations which receive government funding. (I have learned, since the publication of this report, that there is no such requirement and that threats of such a requirement were probably either overstated or inaccurate.)
 

Methods

I chose as my sample the 253 URLs indexed as of April 18, 1999 at the World Association of Persons with Disabilities Resources under the categories of aging, employment, mental health, organizations, physical disabilities, sensory disabilities, and services. I chose this index because it represented a large enough sample of organizations which, presumably, take accessibility issues rather seriously.

My software collected the page content from each URL and passed it to Bobby 3.0b3, an application developed by the Center for Applied Special Technology to test for most of the requirements of the World Wide Web Consortium's Web Content Accessibility Guidelines. Out of 253 pages, 183 returned useful data. Others may be completely "inaccessible" due to bad links.

I defined my primary measurement of accessibility as the inverse of the number of accessibility errors discovered by Bobby. I considered browser compatibility errors and download time as secondary issues affecting accessibility of the page.

Accessibility Errors: The W3C specifies a very long list of requirements for full accessibility. There are several that are not captured by the Bobby tool. Some of the errors that are captured include: failing to provide alternate text for all images, failing to provide alternate text for applets, or using server-side image maps without providing similar functionality within the page.

Browser Compatibility Errors: Browser compatibility errors help to identify whether HTML tags and their attributes are not compatible with certain web browsers or the W3C HTML specifications. A browser error means that a page may not render correctly on certain browsers. These errors are especially troublesome if the user depends upon standard HTML for assistance in parsing the data.

Download Time: Download time is calculated based upon a 28.8 kbps connection with a half second latency for each file and image. High download times are less accessible for all people, but are particularly onerous for those already dealing with enormous delays due to their disabilities.
 

Results

Of the 183 valid samples, nearly two thirds (64.5%) had at least one accessibility error. None had more than three such errors. More than three quarters (78.7%) of the pages had at least one browser compatibility error, with the highest number of such errors being 27.
 

There was a wide range of download times for the pages tested. A large number of pages (42.1%) downloaded in less than ten seconds. Half that many (21.9%) took more than thirty seconds to download. The mean download time was was 17.91 seconds.
 

Discussion

With the existence of the WWW Consortium's standards, measuring accessibility is far less problematic than measuring other factors of web success. This study concerns itself with the three linear measurements of accessibility listed above.

Browser errors and download time are not officially part of the W3C accessibility requirements. The sites surveyed did most poorly in the area of browser errors. But, in my opinion, nonprofits are hardly to blame. The failure of the major browser companies to adhere to W3C standards has led to almost insurmountable fragmentation. More importantly, most of these errors are not fatal to the functionality of the site.

Download times concern me. The 22% of the sites that take more than 30 seconds to download are sites which are losing people, especially if subsidiary pages continue to be hard to download. Fortunately, most of the sites are reasonably speedy, indicating a minimal use of gratuitous graphics and applets.

I admit some surprise to discover that a substantial majority of sites contained full fledged accessibility errors, especially given the deliberately skewed sample I selected. I worry about the accessibility of other nonprofit web sites.
 

Conclusion

Most of these sites throw up at least one barrier to access by people with disabilities. This presents at least two problems for the sites in question: (1) Such errors make their site less usable to the very people they want to serve. (2) If funders, such as the U.S. Federal Government, increasingly require sites to be accessible, these organization may be putting their organizational financial position at risk.

As in my previous studies, I opted to take a shallow look at a large number of sites. But I have no reason to believe that the subsidiary pages would be any improvement over the front page. If these organizations were considering accessibility, that would be reflected on their front page.

Unlike my previous studies, where I have suggested deeper data gathering as the best next step, the most intriguing extension of this study would be to broaden the sample of nonprofits far beyond those organizations serving people with disabilities.

 
Your advice, criticism and feedback would be most appreciated. Please contact us with any thoughts you have to share on this topic.

Thank you,

-- Michael Gilbert

Published: 1999