301 Lake St .
Dallas , PA 18612
This paper has been translated into Serbo-Croatian by Jovana Milutinovich. The transation is available at http://science.webhostinggeeks.com/korist-od-pristupacnog-dizajna
Accessible design of web pages is commonly supported as increasing the potential market by including people with disabilities while offering little benefit to able-bodied visitors. Accessible design of the physical world, however, has been shown to be of benefit to all, not just to those with disabilities. This study tested the effect of accessible design on able-bodied web-users under three conditions: conventional graphical browsers, small-format browsers, and text-only browsers. 120 able-bodied subjects were asked to locate information on either an existing website or a fully W3C compliant copy, using one of three browser types. Time to answer 24 questions, number of correct answers, rated ease-of-use, pleasure of use, and likelihood of reuse were recorded. Results indicated that the accessibly designed site had equivalent usability in the graphical browser, but was significantly more usable in text-only and small format browsers. The results support accessible design as universal design, of benefit to all users, independent of disability.
Key words: Web accessibility – accessible design – usability – universal design
The World Wide Web may be the greatest “melting pot” in history. People from all over the world use the Web as an integral part of their jobs and their lives. In a Business-To-Business model, corporate purchasing and sales staff may use the World Wide Web to maintain contacts with their customers (Journal of Business-to-Business Marketing, 2005). In education, students use the World Wide Web to learn about the animals of the Australian Outback or to check the meaning of the word “ jejune” (Cart, 2000). Many people rely on RSS news-feeds and bloggers to keep up with rapidly changing events (Vamosi, 2004). Full participation in modern culture is very difficult without web access. In spite of this, much of the technology used to create and access the web has been developed without consideration for the needs of individuals with disabilities (Bartlett, 1999, Checkpoint 1.1; Heilman, 2005; National Center on Accessible Information Technology in Education, 2002; Pavka, 2002; Waddell, 1998).
A site that has been designed with consideration for the needs of individuals with disabilities is said to be “accessible.” An accessible website is one that provides access to all users whether a disability exists or not (Brewer, 2003, Slide 2 of 26). The guidelines for Section 508 compliance (General Services Administration, 2002) and those provided for web content accessibility by the World Wide Web Consortium (Henry, 2005) provide guidance for web authors in producing accessible web pages (Brewer, 2003).
Where design for accessibility is intended to benefit individuals with disabilities, “usable design” is intended to benefit all visitors. The international standard, ISO 9241-11, defines usability as “the extent to which a product can be used by specified users to achieve specified goals with effectiveness, efficiency and satisfaction in a specified context of use” (Usability Net, 2003). Among the key features of usability identified by Nielsen are learnability, efficiency, and satisfaction (2003b). Nielsen reports,
Studies of user behavior on the Web find a low tolerance for difficult designs or slow sites. People don’t want to wait. And they don’t want to learn how to use a home page. There’s no such thing as a training class or a manual for a Web site. People have to be able to grasp the functioning of the site immediately after scanning the home page for a few seconds at most (Neilsen & Norman, 2000).
While the ISO standard includes the qualifier “a specified context of use,” web developers may design their sites to be usable on the high-powered development system residing on their own desktop rather than the system that the potential user may have available. HTML, the design language of the Web, was specifically written to allow for presentation of documents through a range of browsers, and on a variety of displays (Graham, 1997). In spite of this, some designers write pages for the features of specific browsers rather than assuring that their pages will be usable on any system a visitor might be using. While not as frequent as in the past, it remains common for websites to include a page detailing the “specified context of use” for their site. (See, for example, http://www.pizzahut.com/browserinfo.asp and http://www.banksterling.com/aboutus/au_bestviewedwith.asp).
Among the “range of browsers” for which web sites should be designed are high-resolution graphical browsers, small format browsers, and text-only browsers. The vast majority of web access today is via the graphical web browser. By mid-2005, over half of web-site visits were by browsers with resolutions of 1024x768 or greater (W3Schools, 2005). However, there is an increasing trend to provide web access via cell-phone displays and wireless personal digital assistants (PDAs). These small format displays may have screens smaller than 320x320 pixels. While display fonts can be reduced to present more information in this small area, the visual acuity of the user limits such strategies. Text-only browsers find use at both the very low and very high technology ends of the spectrum. Web users in remote sites with slow connections may use a text-only browser, or may turn off graphics in a conventional browser to allow pages to load more quickly. Other, high-tech users may wish to browse websites “eyes-free” while driving a car or performing some other visually intensive task. As more and more locations get broadband access so that graphical browsers become widely usable, and cities adopt city-wide wireless service as a utility, the adoption of spoken browsers is likely to increase, so that text-only access is more likely to increase than decrease.
While designing web sites that are usable is clearly a highly desirable goal, the frequency of poor usability suggests that designers have little idea of how to make pages usable (Nielsen, 2004). Yet, usability information is available from a range of sources, and in varying degrees of authority and detail. Jakob Nielsen’s “Designing Web Usability” (Nielsen, 2000a) is based on over a decade of usability testing performed by his company. The Department of Health and Human Services maintains a website that provides guidelines for usability, along with the research supporting each specification (Department of Health and Human Services, 2005). A key element of usable design is to present information in the way the user will be using it (Katz-Haas, 1998; Lynch & Horton, 2002; Nielsen, 2000a, pp. 10).
There are a number of fundamental principals of usability, agreed upon by most authors. Among these are: cross-platform design, separating meaning and presentation, using headings, and using simple, consistent layout and navigation.
The concept of cross-platform design is that, since the web-designer cannot specify the browser and system characteristics of the visitor, s/he must design in a way that allows the page to be usable with a variety of display devices. While the Department of Health and Human Services recommends designing pages for 800x600 screens (2005), Nielsen advises designing pages that are resolution independent (2000a, pp. 29).
The separation of meaning and presentation does not mean that the preferred presentation cannot reinforce meaning, only that it cannot be the only source of some part of the meaning. In The Non-Designer’s Web Book (Williams & Tollett, 1998, pp. 110), designers are told that proximity helps viewers make association between page elements, and that this should be used in web design. However, proximity should not be the only cue that items are related. When some of the information of a web page is provided only by its position on the page, meaning and presentation are conflated. This can present problems when the page is rendered in a format that changes the physical layout of the content, or in a non-visual format.
The use of headings and subheadings assist web usability in two ways: they help identify the level of importance of content, and the help make pages “scannable.” When information is presented under a first or second level heading, it is identified as more important than information that is presented under a level four or five heading. This allows the reader to place importance levels on the content of a page. Because the headings are formatted differently from the content, and in ways that contrast across levels, they are easily identified within the body of the page, allowing the reader to move immediately to the key areas. This is particularly important to usability because most web users do not “read” web pages, but skim them looking for “nuggets” of information (Nielsen, 2000a, pp. 104).
Consistency of layout and navigation helps usability by allowing the user to learn site structure and navigation once, and then to apply what has been learned over the entire site. The layout and navigation should be as intuitive as possible, which means that it should work as expected. Jakob’s Law states that “users spend most of their time on other sites” (Nielsen, 2000b). The more closely the site design and navigation emulates the de facto standards of the web, the easier it will be to use.
While the ability of the web user to accomplish goals is clearly a primary goal of web design, it is often not achieved. Nielsen finds that basic usability guidelines are followed in less than 40% of web sites (2003a). This is also shown by the observation that, in typical usability studies, success rates for able-bodied users are below 70%, and for teens, elders, or people with disabilities, substantially lower (Nielsen, 2001, 2002, 2005a, 2005b).
As with usability, accessibility standards are available, if poorly followed. The Section 508 requirements of the Rehabilitation Act (General Services Administration, 2002) provide guidelines for minimal accessibility for web sites. The more comprehensive guidelines of the W3C (Chisholm et al., 1999) provide instruction on how to make site access possible (Level A compliance), less difficult (Level AA compliance), and convenient (Level AAA compliance). However, the W3C guidelines have been criticized as being “at once too vague and too detailed” (Clark, 2003). The basic concepts of accessibility are not complex, but attempting to create a comprehensive set of guidelines may be. Among the basic elements of accessibility are simplicity, organization, and equivalency.
The W3C recommends that sites be written with the clearest and simplest language appropriate for a site's content (Chisholm et al., 1999, Checkpoint 14.1). The language used should be easily understood by the site’s target audience. This does not require “dumbing down” a technical site where specific training is required to understand the content. However, a site that is intended to inform the general public should be written using language that recognizes that almost half of the adult population reads at the lowest two levels of literacy (Kirsh et al., 2002).
The recommendation that sites be organized so that they are readable even if not rendered with the intended style (Chisholm et al., 1999, Checkpoint 6.1) recognizes that different browsers may present pages differently. If a page is organized logically in its basic layout, the loss of visual formatting and positioning will not make it incomprehensible.
The recommendation of equivalent alternative presentation of information (Chisholm et al., 1999, Checkpoint 1.1) is intended to assure that an individual using a browser that does not render graphics or multimedia can still obtain the information presented by a site. In general, though not in all cases, this is assured through using “alternative text” descriptions.
In general, the proponents of accessible design have supported the technology as benefiting users with disabilities, and increasing access by including this portion of the population in potential customers (Arch & Letourneau, 2002; Byrne, 2001; Clark, 2003; Heilman, 2005; Pavka, 2002; Waddell, 1998). In addition, authors have argued that accessible design is mandated by law, and shows social responsibility. A few authors have asserted that web accessibility also increases usability (Moss, ND), but provide little objective evidence to support the assertion.
Considering the striking similarities between the requirements for accessibility and usability, the assertion is not obviously flawed. Both standards suggest device independent design. Both standards suggest simplicity. Both standards suggest the use of logical organizing markup (headers).
In other areas of accommodation, it has been demonstrated that accommodations for individuals with disabilities also improve access by able-bodied individuals. Curb cuts, installed to provide access by individuals in wheelchairs, have also improved access for those pushing children in strollers, or those transporting wheeled luggage. Amplified handsets on public telephones, provided to assist users who are hearing impaired also aid those with intact hearing when the levels of background noise are high. In general, accommodations for people with disabilities have been shown also to aid able-bodied persons working in sub-optimal situations. The similarities between the standards for accessibility and usability suggest that a website that has been designed for accessibility should also provide a higher level of accessibility, especially when rendered on a sub-optimal browser or via a sub-optimal web connection.
This study was conducted to test the hypothesis that when websites are constructed to comply with W3C requirements for accessibility, they are more usable by able-bodied users as well. The specific research questions for the study were:
This study used a quasi-experimental design with random assignment of participants to experimental conditions.
The participants in this study were 120 adults without identified disability. The participants ranged in age from 18 to 74, with a mean age of 33 years. They were predominantly female (77.5% female, 22.5% male), and had educational experience ranging from high school to advanced college degrees, with an average of 15.7 years of formal education. All of the participants had visual acuity (with or without glasses) sufficient to read 12 point, Times Roman font, hearing adequate to participate in a conversation in a quiet room, and had sufficient motor skills to use a standard mouse and keyboard, and to sit in an armless chair for at least 60 minutes without assistance.
This activity required participants to use a computer to access a specific website. The study used a number of computers, at sites in Pennsylvania, New York, and Georgia. All computers had at least 900 MHz Pentium IV processors, with at least 128 MB of RAM memory. All computers used broadband internet connections so that connection speed did not influence the results. To control for differences in screen size, the monitors on the computers, regardless of dimensions, were set to 800x600 pixel resolution.
This study compared access to two web sites on three available browsers, representing conventional graphical browsers, text-only browsers, and small format browsers.
The study used Firefox, version 1.0.5 as the graphical browser. This browser was selected for the study as it is highly standards-compliant, and would display a compliant web-site correctly. In addition, it would be less likely to render non-standard extensions that might be used when a site is authored for a specific browser rather than to comply with web standards.
Text-only browsers are used in spoken-web interfaces over telephones or screen-readers for blind web browsers, and are often used by individuals with slow internet connections. For purposes of this study, we used Lynx, version 2.8.3 as the text-only browser. This version displays pull-down menus appropriately, but, as with many text browsers, “linearizes” tables. Lynx uses the cursor control keys to navigate through a site, and does not respond to mouse clicks. The controls available on Lynx emulate the controls that might be available through a spoken browser. Lynx is often recommended to web developers as a way to experience their work as it would be rendered to a blind person.
Initially, we had intended to use the Palm Blazer web browser via wireless connection for the study. We changed this plan for two reasons. First, the Blazer browser, like the Lynx browser, did not render the Flash navigation used in our study site. Since we would find out the effects of lost navigation in the text-only browser, and we wanted to also discover the effects of a small screen, we considered abandoning the Blazer browser. Second, we found that the early release of the Blazer, combined with a Palm T5 hand-held, was unstable, and would frequently lock up the handheld, requiring a reset.
Because of these factors, we created a “Windows wallpaper” that included a colored rectangle 480x320 pixels, the resolution of the Palm T5 screen. We then sized the Firefox browser to just cover this rectangle, and used this constrained window to emulate the small-screen display while having the full stability of a PC-based browser.
For our “conventional web site,” we wanted a site of significant complexity (>100 pages) which used commercial web design techniques. The study site should be intended to convey information to visitors, including the general public. After searching for an appropriate site, we selected the site of the Pennsylvania Bar Association (http://www.pabar.org). This site is maintained by a professional web design company, and includes a statement that the content “is copyrighted and may be reproduced in any form including digital and print for any non-commercial purpose so long as this notice remains visible and attached hereto.” Because web sites are dynamic, we replicated the site onto our server, so that the content would not change over the course of the study.
Once the conventional web site was selected, we replicated the look-and-feel of the site in a second, accessible version. The accessible version pages were tested with the online accessibility checker, Cynthia Says, to assure that our accessible site complied with all A, AA, and AAA level recommendations of the W3C. Site changes for accessibility included replacing the Flash-based navigation with a list/CSS navigation that provided the same visual experience, the replacement of font-tag based, visual markup with header-based logical markup, replacing many visually formatted lists with HTML list markup, and rewording many “Click here” links with meaningful text. In addition, a number of PDF files were rewritten as HTML.
The accessible site did not include all changes that might have improved usability. The logical organization of the site and of pages was not changed in our redesign. While it is recommended that pages be relatively short to facilitate loading and reading, the site included a number of pages that were many screens long. Since we could not assure the preservation of the authors’ intent if we divided the pages, we left page-lengths as they were in the original site. We also did not convert all PDF files to HTML, as some large portions of the site were PDF versions of newsletters. We did change a number of PDF pages to HTML to show the feasibility of doing this.
Both sites were hosted on our web server, so that differences in access rate would be controlled.
Elapsed times were recorded using a SportLine Jumbo Sports Timer 234 or equivalent.
Participants in this study were given a set of 24 fact-based questions to answer from the web site Each question included a “hint” about where to look for the answer to the question. The questions were designed to be somewhat ambiguous, to reflect the likelihood of a site visitor not knowing the precise terminology that would be used for a topic. For example, a question on rental security deposits included the response “first and last months rent” while the target page used the wording “two times the monthly rent.” Similarly, the hints would take the participant to a page that either contained the answer, or that contained an explicit link to the answer. The questions were arranged so that the participant would not receive all questions that could be answered from a section of the site in sequence. Rather, the participant would, if answering the questions in the order presented, move to other portions of the site, then return to areas previously visited. This provided some insight into the “learnabilty” of the site.
The “experience questionnaire” asked participants about their perceptions of the navigation, font, links, and scrolling of the site through open-ended questions. Additionally, this questionnaire used three visual-analog scale items on which the participants recorded the site’s ease-of-use, pleasure-of-use, and the participant’s likelihood of reusing the site, should they need similar information in the future.
Prior to the beginning of data collection, the researchers prepared a set of 120 envelopes, each containing a specification for a web site version and a browser version. These envelopes were shuffled into a storage box, and used for random assignment to groups.
At the beginning of each trial, the participant selected an envelope from the prepared set. This envelope determined the site and browser to be used. The researcher used the full-screen Firefox browser to orient the participant to the site, demonstrating the basic content and navigation features of the site. If the participant was using a small screen or text-only browser, this configuration was also set up, and the home page of the target site loaded. After orientation to the site on the full-screen browser, the participant was introduced to the site and navigation on the non-standard browser. The participants were instructed that they could use the navigation features of the pages, and the “find” feature, if present, in the target browser, but that they were not allowed to use a search engine to find the answers on the site. (Both versions of the site had the included search feature disabled, and participants were monitored so that they did not use a general web-search tool.)
Once the participants were oriented to the site, they were given the content questionnaire. They were instructed: “When I say go, please answer these questions as quickly and as accurately as you can. If you cannot find the answer to a question, just skip over it and go on to the next question. You will notice that each question has a ‘hint’ to help you locate the answer. Are you ready? Go.” At the word “Go,” the stopwatch was started. If, in the course of the trial, the participant had questions about the browser or the features of the site, these were answered, but they were not guided to answers to the questions. When the participant answered the last question, or indicated that they could find no more answers, the stopwatch was stopped.
Once the content questionnaire was completed, the participants were asked to complete the experience questionnaire, describing their experience using the web site Each participant used one browser, and visited only one site.
The data collected in this study include both quantitative and qualitative components. The quantitative measures were the time to complete the content questionnaire, the number of correct responses, and the participant’s ratings of ease-of-use, pleasure-of-use, and likelihood of reuse of the sites. In addition, we calculate the “time per correct response” as elapsed time / number correct. The first-pass analysis of these values were obtained using multiple-ANOVAs to determine the effect of browser, the effect of site, and interactions of site with browser. Within-site and within-browser comparisons were made using t-tests and ANOVA for more specific comparisons.
Qualitative information for each site/browser combination was obtained through the experience questionnaire. The participant responses to the open ended questions were analyzed for emergent themes using a priori categorization.
This study tested three hypotheses:
Hypothesis 1 : A web site that complies with W3C A, AA, and AAA requirements can be as usable as a conventional web site for able-bodied users of standard graphical browsers.
This hypothesis was confirmed. Using the conventional graphical browser, visitors spent slightly longer on the conventional site (mean elapsed (conventional) = 3229 seconds, mean elapsed (accessible) = 2941 seconds, p > .05), answered the same number of questions correctly (mean correct (conventional) = 20.1, mean correct (accessible) = 19.65, p > .05) and took slightly longer for each correct response (mean time per correct (conventional) = 164 seconds, mean time per correct (accessible) = 160 seconds, p > .05) than users of the accessible site. In addition, the sites were rated equal in ease of use (mean ease of use (conventional) = 82.3, mean ease of use (accessible) = 87.2, p > .05, in pleasant to use(mean pleasant (conventional) = 73.45, mean pleasant (accessible) = 83.0, p > .05, and in the likelihood of being reused by the visitor (mean reuse (conventional) = 81.95, mean reuse (accessible) = 87.68, p > .05). (See Figure 1)
These findings indicate that an accessible web site can, with attention to the look and feel, replicate the functionality of a conventionally designed web site for users of the conventional, full-screen graphical browser.
Hypothesis 2 : A web site designed to follow the guidelines for accessible design will be more usable to visitors using a text-only browser, such as would be used to access the web through a “web reader” service over a standard telephone or by a visitor with a slow dial-up connection having graphics turned off (estimated to be 20 to 30% of web traffic), than a conventionally designed web site
This hypothesis was strongly confirmed by the results of this study. Using the text-only browser, visitors spent significantly less time on the conventional site (mean elapsed (conventional) = 2246 seconds, mean elapsed (accessible) = 3932 seconds, p < .05), were able to answer significantly fewer questions successfully (mean correct (conventional) = 2.5, mean correct (accessible) = 18.72, p < .05) and took substantially longer for each correct response ((mean time per correct (conventional) = 951 seconds, mean time per correct (accessible) = 240 seconds, p < .05) than users of the accessible site. While users of neither site enjoyed using the text-only browser, the accessible site was rated significantly higher in ease of use (mean ease of use (conventional) = 10.1, mean ease of use (accessible) = 48.9, p < .05, as more pleasant to use(mean pleasant (conventional) = 11.3, mean pleasant (accessible) = 34.8, p < .05), and in the likelihood of being reused by the visitor (mean reuse (conventional) = 7.4, mean reuse (accessible) = 36.39, p < .05). (See Figure 2)
Hypothesis 3 : A web site designed to follow the guidelines for accessible design will be more usable to visitors using a small format browser, such as would be found on a cell-phone or PDA, than a conventionally designed web site
This hypothesis was partially confirmed. Using the small format graphical browser, visitors spent significantly more time on the accessible site (mean elapsed (conventional) = 2577.65 seconds, mean elapsed (accessible) = 3784.05 seconds, p < .05), but answered significantly more questions correctly (mean correct (conventional) = 12.75, mean correct (accessible) = 20.15, p < .05) and were able to get each correct response significantly more quickly (mean time per correct (conventional) = 289.92 seconds, mean time per correct (accessible) = 191.28 seconds, p < .05) than users of the conventional site. In addition, the accessible site was considered to be significantly easier to use (mean ease of use (conventional) = 49.2, mean ease of use (accessible) = 73.4, p < .05). However, while the ratings for pleasure of use (mean pleasant (conventional) = 47.75, mean pleasant (accessible) = 73.40, p > .05, and in the likelihood of being reused by the visitor (mean reuse (conventional) = 57.50, mean reuse (accessible) = 79.10, p > .05) both favored the accessible site, they did not rise to the level of significance. (See Figure 3)
Both the graphical and the small-format browser offered full access to the features of the conventional site. In spite of this, the conventional site was significantly less usable for able-bodied users of a small-format browser than for users of the conventional, graphical browser. While users spent the same amount of time on the site (mean elapsed (conventional, graphical) = 3229, mean elapsed (conventional, small format) = 2577.65 , p > .05), they were able to locate fewer answers (mean correct (conventional, graphical) = 20.1, mean correct (conventional, small format) = 12.75, p < .05), and spent more time per correct response (mean time per correct (conventional, graphical) = 164 seconds , mean time per correct (conventional, small format) = 289.92 seconds , p < .05). This difference was reflected in poorer ratings for ease of use (mean ease of use (conventional, graphical) = 82.3, mean ease of use (conventional, small format) = 49.2, p < .05), pleasure of use (mean pleasant (conventional, graphical) = 73.45, mean pleasant (conventional, small format) = 47.75, p < .05), and likelihood of reusing the site to locate similar information (mean reuse (conventional, graphical) = 81.95, mean reuse (conventional, small format) = 57.50, p < .05).
Users of the conventional graphical and small format browsers had a very different experience on the accessible site. Users of the accessible site took significantly longer to locate answers when using the small format browser (mean elapsed (accessible, graphical) = 2941 seconds , mean elapsed (accessible, small format) = 3784.05 seconds, p < .05), but had equivalent numbers of correct responses (mean correct (accessible, graphical) = 19.65, mean correct (accessible, small format) = 20.15, p > .05), and similar times per correct response (mean time per correct (accessible) = 160 seconds , mean time per correct (accessible, small format) = 191.28 seconds , p > .05). In addition, their ratings for ease of use (mean ease of use (accessible, graphical) = 87.2, mean ease of use (accessible, small format) = 73.4, p > .05), pleasure of use (mean pleasant (accessible, graphical) = 83.0, mean pleasant (accessible, small format) = 73.40, p > .05) and likelihood of using the site to find similar information (mean reuse (accessible, graphical) = 87.68, mean reuse (accessible) = 79.10, p > .05) were not significantly different.
Both the conventional and accessible web sites were found to be “easy to navigate” by 85% of users. Among the features that were cited as enhancing navigability of the site were the navigation menu on the left side of the page and having headers on the page to clearly delineate sections and topics. One unexpected finding was the number of individuals who mentioned having navigational links in alphabetical order as a benefit when present, and a difficulty when not. Web designers commonly arrange navigational links in order of perceived importance, but users, who may not be aware of the relative importance of features, seem to prefer the predictable, alphabetical ordering of links in a navigation system.
While 80% of users of the conventional web site found the links easy to locate, an additional 10% (90% total) found the links on the accessible site to be easy to locate. Users found “links that were part of a sentence describing what they contained” to be helpful. Since many “click here” links of the conventional site were rewritten as meaningful text in the accessible site, this preference may account for the difference in ease of use of links. Almost 10% of participants complained about links that opened pop-up windows or PDF files. One user indicated, “When I click on a link that opens a PDF, I usually just back out.”
Both of the sites in our study used a single-spaced, sans-serif font (preferentially, Arial), and the preponderance found it to be easily read. Interestingly, while 80% of users of the conventional site found the font easy to read, 20% found it too small to be read comfortably. By comparison, 85% of users of the accessible site found the font easy to read, and 10% found it too large. On both sites, several participants found that long pages of dense text were difficult to read, which agrees with the recommendations for both usability and accessibility.
On both the conventional and accessible web sites, most users of the conventional browser (80% conventional, 75% accessible) found scrolling to be easy. Two users of the accessible site found scrolling side to side to be a problem, although neither site required lateral scrolling in the conventional browser, so this must reflect general rather than specific experience. Roughly 20% of participants in each site found scrolling through long pages to be a hindrance to usability. One respondent indicated, “I would have preferred typing in a search as compared to scrolling the page especially when the lists of items are very long.”
The inability of a text-only browser to render a Flash navigation system strongly influenced the experience of users of the conventional web site in our study. 95% of participants commented that navigation was difficult or impossible. The remaining participant found that “using the arrows was easy once I got used to them, yet finding the information necessary to answer the questions was next to impossible.”
By comparison, only half of participants found the accessible site difficult to navigate in the text-only browser. Many of these were clearly responding to the fact that navigation was not what they were familiar with, commenting that they couldn’t “click on” a link to activate it, but others were frustrated by their inability to view PDF files natively. (In Lynx, a PDF file is downloaded to the user’s computer, and must then be opened by a separate program.)
The main page of both sites had a number of links that led to answers to the test task. On both sites, the links were presented in the same color and font by the Text-Only browser. In spite of this, 70% of users of the conventional site found the links difficult to locate on the page, and difficult to open. By comparison, 55% of users of the accessible site found the links easy to locate, once they became familiar with the browser. The participants specifically cited the colors used as an aid to locating the links, while users of the conventional site found the colors to hinder locating the links.
The body text of both sites was also presented in the same font in the text only browser. Yet, 70% of the users of the conventional web site found the font used by the browser to be difficult to read, indicating that it was “too small,” and had “not enough contrast.” By comparison, 55% of users of the accessible site found it to be easy to read, with most of the others reporting that the display font was too small. Several respondents in each site found that large blocks of text with no graphics (or white space) were difficult to read.
A sizable proportion of users of both sites (50% of users of the conventional site, 35% of users of the accessible site) found scrolling difficult or “time consuming.” At least some of this difficult stemmed from the use of the cursor keys rather than a mouse to scroll. Users commented “I had a problem every time I used the back key,” and that scrolling was “more time consuming.” In a spoken browser, such as a telephone reader service, it is unlikely that users would downgrade the site for inability to use a mouse.
In the small format browser, both sites were fully functional, and used the same fonts as in the large format browser. A primary difference between the two sites that could affect the user experience was the use of proportional layout throughout the accessible site, so that side-to-side scrolling was not necessary. This difference resulted in a different user experience for participants in the study. On the conventional site, 55% of participants found the navigation system to be difficult. Users reported that ”it was frustrating navigating through the web site and constantly using up and down and side arrows.” Others said that “the size of the screen made the site difficult.” Even those who found the site “easy” to navigate felt that “you had to scroll a lot to see everything.” Users of the accessible site had a much easier time. 70% of these participants rated the site as easy to navigate. Of those who felt that the site was difficult, the primary concern was the small screen size and the need for scrolling through long pages of information.
Although the display of link text was the same on the two sites, the links of the accessible site were written to contain meaningful text rather than “click here.” This change, combined with the fitting of text to the screen resulted in a different experience for users compared with the conventional web site Where half of the users of the conventional site found the links to be difficult to use, only 25% of participants on the accessible site had difficulty using the links. Many of these individuals reported difficulty with long pages where the navigation elements did not fit within the small screen.
As with the links, the page text used in the small format browser was the same for the conventional and accessible web sites. In spite of this, nearly 40% of comments about the font of the conventional site indicated that it was difficult to read, while only 6% of responses indicated difficulty with the font of the accessible site. Many of the negative responses about the fonts were actually related to the screen size, indicating that the participant could not enlarge the screen, and that a smaller font would have been desirable to see more of the page at once. As with the text-only browser, participant’s frustration with one feature (screen size, in this case) affected their perceptions of another feature (font readability).
One area where accessible design did not change the user experience in this study was in the perception of scrolling. Almost three quarters of respondents on both sites complained about having to scroll up and down to find information on the screen. As noted earlier, our site redesign did not include following the usability and accessibility recommendations for using many smaller pages in place of a few large ones. It appears that sizing pages to minimize scrolling will significantly enhance the visitor experience.
The findings of the study indicate that designing web sites for accessibility does not reduce the usability for able-bodied visitors using conventional graphical browsers, and makes them more usable for able-bodied users in small format browsers (such as PDAs and cell phones), and in text-only browsers, such as might be used with a slow connection or with a reading browser used over the telephone. The current study had sufficient power to validate our hypotheses, but showed a trend in the typical browser that warrants further study. While none of the comparisons of usability for the typical browser rose to the level of significance, all of them favored the accessible web site Since the accessible site was designed to emulate the look, feel, and performance of the conventional site as closely as possible, it is not clear what features of the accessible site produced this difference. The lack of statistical significance appears to be due to the high degree of variability in our participant performance. However, it appears that, with a larger sample, the results for the typical browser would also show a higher degree of usability for accessible web sites
As noted earlier, the participants who found the font difficult to read in the typical browser had different complaints. Those who found the font difficult to read in the conventional site found it to be too small. On the other hand, those who found it difficult to read on the accessible site, where the same font size was used, found it to be too large. This may reflect a difference in the cognitive load of the two sites. If a visitor feels stressed by the amount of information being conveyed at once, they may feel that the font is too small, and a larger font would make interpreting the information easier. On the other hand, if the visitor is not fully engaged by the information being presented, a smaller font would allow more to be taken in at once. The higher ease-of-use and comfort-of-use ratings, along with the feeling that the font was too large, may indicate that the cognitive overhead of the accessible site was lower than that of the conventional site.
Participants using both the conventional and accessible web sites required more time to answer questions when using the small format browser than when using the conventional graphical browser. This difference is likely due, in large part, to the reduced amount of information that can be displayed at once in a small format window. More scrolling is required to find information on a page than in a large-format browser. Hence, the increased time-per-correct-answer in both sites is likely related to the browser display, and not a function of the site.
From the comments made in the experience questionnaire, it appears that the small-format browser experience was strongly affected by the presence of a large monitor. Half of the subjects in this study made explicit comments about their frustration with not being able to expand the size of the browser window. If the physical dimensions of the screen were fully occupied by the browser, this negative reaction might have been lessened. Alternatively, it is possible that small screens are not really suitable for web browsing. This question could be answered in a further study, but is beyond the scope of the current research.
As with the small format browser, it appears that users of the text-only browser were strongly affected by comparisons of the browser to the more familiar graphical browser. Although users of the conventional web site were severely limited by inability to use the site’s graphical navigation system, even users of the accessible site complained about navigation. Among all users of the text-only browser, the ease of navigating with a mouse compared with use of the keyboard arrow keys was commented on by 27 of 40 respondents. In this case, we were using the Lynx browser to provide an experience similar to what would be obtained using a voiced browser over the telephone, and the use of the computer as a means of access may have affected the responses. The user of a telephone to access a web site will not expect to have all of the amenities provided by a full graphical browser, so will not be as likely to complain about the browser experience. In spite of this effect, the accessible site was clearly more usable than the conventional site for users of the text-only browser.
The usability of the conventional site was strongly affected by the method used to access it. Because the text-only browser did not provide access to the navigation features, the conventional site was essentially unusable. Of the 24 questions, only two subjects were able answer as many as five. The typical user found the two answers directly accessible from the home-page, and were then unable to find any more. However, users of the small-format browser also experienced reduced usability of the conventional site. As measured by success in locating information, the conventional site was only half as usable in the small format browser as in the conventional full-screen browser.
The accessible site, on the other hand, was relatively unaffected by the method used to render it. While users of the text-only browser rated the site as less easy to use, less pleasant to use, and less likely to be reused than did the visitors using the conventional or small-format browsers, their time to locate the answers and the number of correct responses did not differ significantly from the participants using the typical browser. This suggests that the ease of use and pleasure of use scores were more related to the browser than the web site
These finding are an important part of the case for accessible design. Many authors on accessible design, including Christian Heilmann (2005) and Joe Clark (2003) have supported the notion that accessible design has little impact on the able-bodied user of a web site The results of this study clearly show that accessible design, even when the look and feel of the site is changed only minimally, have a significant effect on the able-bodied user when using a sub-optimal browser (text only or small format). Even in the conventional graphical browser, the mean scores consistently favored the accessible site, but did not rise to the level of significance because of the large variability in the sample. It is likely that, in a larger study of accessible versus conventional sites in graphical browsers alone, these scores would achieve significance.
An interesting finding is that users of the conventional web site, in a sub-optimal condition, were more critical of features than users of the accessible site. For example, when using the text-only or small format browser, the font size and colors were exactly the same between the two sites. In spite of this, all users of the conventional site said that the font was too small and difficult to read in the text-only site, while only half of the users of the accessible site found the font difficult, and roughly half said that the font was easy to read. It appears that the frustration of using the conventional site resulted in a more negative interpretation of all of the features. It is likely that this same negative spill-over would be directed to the organization owning the site.
These results suggest that accessible design should not, in fact, be considered as an aspect of accommodation for individuals with disabilities, but as an implementation of universal design. Historically, accessible design has been justified because it allows individuals with disabilities to use the site and become customers. The more reticent justify accessible design as legally required, and something that must be done whether or not it makes sense. However, this study shows that accessible design, as with other implementations of universal design, also benefits able-bodied users of a web site when visiting via a sub-optimal browser. As an added benefit, accessible web design accommodates the needs of those with functional limitations, brought about by disability or by the processes of normal aging.
Usability studies commonly find success rates in completing web tasks, such as purchases, to be on the order of 50%. The overall success rate of users of the conventional site in our study, across all three browsers, was 49%, suggesting that our results closely match the real-world experience of web users. Users of the accessible site in this study, however, achieved an overall success rate of 80% across all three browsers. Accessible design produced a 30% improvement in usability for able-bodied users.
Since this increased usability comes at minimal cost, this study provides strong support for businesses creating accessible web sites As able-bodied users increase their access to the web through their cell phones, or through web-reader services, the importance of designing sites for these sub-optimal delivery channels increases, and accessible design provides guidelines for such methods of delivery of web content.
Arch, A., & Letourneau, C. (2002). Auxiliary benefits of accessible web design. Retrieved Nov. 11, 2005, from http://www.w3.org/WAI/bcase/benefits.html
Bartlett, K. (1999). Web accessibility and users with disabilities. Retrieved July 16, 2003, from http://aware.hwg.org/why/essay_kb_01.html
Brewer, J. (2003). Online overview of the web accessibility initiative. Retrieved Nov. 8, 2005, from http://www.w3.org/Talks/WAI-Intro/Overview.html
Byrne, P. (2001). Benefits and challenges of web accessibility. Retrieved Nov. 11, 2005, from http://www.mcu.org.uk/articles/benefits.html
Cart, S. (2000). The learning innovations center: How might you benefit. Retrieved Nov. 22, 2005, from http://www.schoolworld.asn.au/lic/intro2.html
Chisholm, W., Vanderheiden, G., & Jacobs, I. (1999). Web content accessibility guidelines 1.0. Retrieved May 25, 2004, from http://www.w3.org/TR/WAI-WEBCONTENT/
Clark, J. (2003). Building accessible web sites Indianapolis: New Riders.
Department of Health and Human Services. (2005). Research based web design & usability guidelines. Retrieved Nov. 10, 2005, from http://usability.gov/guidelines/
General Services Administration. (2002). § 1194.22 web-based intranet and internet information and applications. Retrieved Nov. 8, 2005, from http://www.section508.gov/index.cfm?FuseAction=Content&ID=12#Web
Graham, I. S. (1997). Html 4.0 sourcebook. New York: John Wiley & Sons, Inc.
Heilman, C. (2005). 10 reasons clients don't care about accessibility. Retrieved Sept. 20, 2005, from http://www.digital-web.com/articles/ten_reasons_clients_dont_care_about_accessibility/
Henry, S. L. (2005). Web content accessibility guidelines (wcag) overview. Retrieved Nov. 8, 2005, from http://www.w3.org/WAI/intro/wcag.php
Journal of Business-to-Business Marketing. (2005). About the journal, Journal of Business-to-Business Marketing: Haworth Press, Inc.
Katz-Haas, R. (1998). User-centered design and web development. Retrieved Nov. 10, 2005, from http://www.stcsig.org/usability/topics/articles/ucd%20_web_devel.html
Kirsh, I. S., Jungeblut, A., Jenkins, L., & Kilstad, A. (2002). Adult literacy in america: A first look at the findings of the national adult literacy survey. In U. S. D. o. Education (Ed.) (pp. 201): U.S. Government Printing Office.
Lynch, P. J., & Horton, S. (2002). Web style guide: Basic design principles for creating web sites (2nd ed.): Yale University Press.
Moss, T. (ND). The secret benefit of accessibility part 1: Increased usability. Retrieved Nov. 11, 2005, from http://www.hotlib.com/articles/show.php?t=The_Secret_Benefit_Of_Accessibility_Part_1:_Increased_Usability
National Center on Accessible Information Technology in Education. (2002). What is assistive technology? Retrieved July 29, 2004, from http://www.washington.edu/accessit/articles?109
Neilsen, J., & Norman, D. (2000, Jan. 14, 2000). Web-site usability: Usability on the web isn't a luxury. Information Week, Jan. 14.
Nielsen, J. (2000a). Designing web usability: The practice of simplicity. Indianapolis, Indiana: New Riders.
Nielsen, J. (2000b). End of web design. Retrieved Nov. 11, 2005, from http://www.useit.com/alertbox/20000723.html
Nielsen, J. (2001). Beyond accessibility: Treating users with disabilities as people. Jakob Nielsen's Alertbox Retrieved Feb. 10, 2005
Nielsen, J. (2002). Usability for senior citizens. Jakob Neilsen's Alertbox Retrieved Feb. 10, 2005, from http://www.useit.com/alertbox/20020428.html
Nielsen, J. (2003a). The ten most violated homepage design guidelines. Retrieved Nov. 11, 2005, from http://www.useit.com/alertbox/20031110.html
Nielsen, J. (2003b, August 25, 2003). Usability 101: Introduction to usability. Retrieved Nov. 8, 2005, from http://useit.com/alertbox/20030825.html
Nielsen, J. (2004). Top ten mistakes in web design. Jakob Neilsen's Alertbox Retrieved Feb. 10, 2005, from http://www.useit.com/alertbox/9605.html
Nielsen, J. (2005a). Lower-literacy users. Retrieved April 25, 2005, from http://www.useit.com/alertbox/20050314.html
Nielsen, J. (2005b). Usability of websites for teenagers. Jakob Nielsen's Alertbox Retrieved Feb. 10, 2005, from http://www.useit.com/alertbox/20050131.html
Pavka, A. (2002, April 10). Accessible by design. Digital Web Magazine.
Usability Net. (2003). International standards for hci and usability. Retrieved Nov. 8, 2005, from http://www.usabilitynet.org/tools/r_international.htm#9241-11
Vamosi, R. (2004). Rss: News you choose. Retrieved Nov. 22, 2005, from http://reviews.cnet.com/4520-10088_7-5143656.html?tag=nav
W3Schools. (2005). Browser statistics. Retrieved Nov. 8, 2005, from http://www.w3schools.com/browsers/browsers_stats.asp
Waddell, C. D. (1998, June 17, 1998). Applying the ada to the internet: A web accessibility standard. Retrieved July 16, 2003, from http://www.rit.edu/~easi/law/weblaw1.htm
Williams, R., & Tollett, J. (1998). The non-designer's web book. Berkeley, California: Peachpit Press.