3. ACCESSIBILITY & DESIGN FOR VISUALLY IMPAIRED PEOPLE
3.1 Accessible web interfaces
There has been considerable research into accessible web interfaces. TIDE ACCESS is described in a paper by Stephanidis and Emiliani (1998) which stress its focus on the "design for all" concept. The SPEECH project (Zajicek and Powell 1997) built a conceptual model of the web for visually impaired users through development of the BrookesTalk web browser, which aims to present the contents of web pages for anyone using speech-only technology. Other development work includes that undertaken by the Trace Center (O'Briant 1999) and the W3C Web Accessibility Initiative (Casey 1999). For a review of this work see the REVIEL Project final report (Brophy and Craven 1999).
Early papers on accessibility (e.g. Astbrink (1996), Dixon (1996), Kerr (1997)) offer practical advice on accessible web design stressing that correctly applied HTML will help ensure universal access and should be provided as "a right, not a favour". Both Dixon and Astbrink provide advice on all the main accessibility points such as the handling of images, forms, tables etc. Kerr also offers practical advice for making web pages accessible, and although the paper concentrates mainly on the problems faced by blind users trying to access web pages using a screen reader, the article emphasises the fact that making web pages accessible for this group of people improves access for all. The paper covers technical solutions for accessing web pages such as access technology; then moves on to consider some general design principles such as the use of images, columns and forms; and finally covers more advanced design issues such as the use of JavaScript and PDF files.
Basic design principles are addressed by Poulter et al., concentrating mainly on the structure of files and hypertext links (Poulter et al. 1999). Ormes and Peacock focus on the need for public libraries to be aware of accessibility issues and summarise the main points such as the use of images, maps, tables, frames etc (Ormes and Peacock 1999). The paper goes on to describe the work of the W3C and describes how to check web pages for accessibility. The final report of the REVIEL project, published under the title The Integrated Accessible Library: A model of service development for the 21st century, includes a chapter devoted to "Making Electronic Formats Accessible" as well as another chapter on the "Accessibility of current electronic information services in the UK" which looked specifically at the JISC online services and also at a selection of UK Higher Education libraries' web sites (Brophy and Craven 1999).
A wide range of articles, books and reports can be found on the subject of web accessibility: see for example Waters (1997), Paciello (2000) and Nielsen (2000). Many organisations have also produced their own accessibility guidelines, usually based on the recommendations of the World Wide Web Consortium, perhaps tailoring this approach to make it more user friendly or to address specific issues. For example, Jenkins describes how the Web Accessibility Guidelines were implemented at IBM (Jenkins 1997), covering steps taken including information relating to general disability issues. The paper describes how different types of visual impairment can present different problems, for example with colour, depth and fonts and also covers hearing, mobility and cognitive impairments. Solomon recognises that whilst it is important to try to adhere to the W3C guidelines, some organisations "may not have the resources to allow developers to learn these standards end-to-end", therefore producing a basic guide for developers which covers the main accessibility issues can act as a starting point "rather than not thinking about accessibility at all" (Solomon 2002).
The development of an accessibility policy is described by Foley and Regan who view accessibility as an "important and timely issue" which needs to be addressed by any organisation that has a website. The paper provides advice on the identification and adoption of a policy as well as validation methods to check for accessibility based on the W3C guidelines. The authors conclude that "there are no quick fixes for accessibility, but careful and thoughtful planning can minimize many of the challenges of accessible design"(Foley and Regan 2002). Thatcher describes implementation of accessibility standards relating to testing for Section 508 compliance in the USA (see section 3.2), but he warns that when testing web pages for accessibility it is important not to rely solely on automated checking tools (such as Bobby and Lift) because although software tools provide an initial check, "human judgement is essential" (Thatcher 2002). User testing of websites is strongly advocated by Nielsen (2000), who provides advice on accessible design of websites based on usability studies, as well as guidelines for conducting usability tests.
Brazier and Jennings provide an insight into the users’ perspective, which can be compared with published guidelines and is based on consultations with users (Brazier and Jennings 1999). Also looking at specific issues rather than general design principles is a study commissioned by the RNIB (Buultjens et al. 1999). It examines font size and styles of print for visually impaired students to ascertain which are the most popular. Unsurprisingly the results showed a wide range of preferences but conclude that Helvetica N24 plain text is the most generally acceptable font size and style.
The perceptions and experiences of web use by blind and visually impaired users have been summarised by Berry in a literature review on issues of visual impairment (Berry 1999). The paper describes a study undertaken using a group of blind and partially sighted students and staff to ascertain their experiences in accessing and using the web. Not surprisingly, poorly designed web pages were identified as a severe problem for blind people. Beginners and inexperienced users were more likely to become frustrated and switch off the computer. A study by the Nielsen Norman group also found that blind and visually impaired people experienced problems navigating the web and estimate that "the Web is about three times easier to use for sighted users than it is for users who are blind or who have low vision" (Coyne and Nielsen 2001 p.5).
3.2 Legislation
In the UK the Disability Discrimination Act (DDA) 1995 requires (under Part III of the Act) providers of "goods, facilities and services", such as libraries, to provide an equal level of services to all their customers. It states also that no extra charges can be imposed for service provision in relation to a person's disability, for example charging a fee to produce materials in alternative formats (Great Britain 1995.).
A summary of relevant legislation in the UK is provided by Oppenheim and Selby in a paper on access to the Web for blind and visually impaired people (Oppenheim and Selby 1999). The possible effects of the DDA are discussed, recognising that although the Act requires equal access to goods services and facilities, it is still unclear with whom responsibility lies, for example whether "an Internet Service Provider is responsible in this respect for the content of the web pages it carries" (Oppenheim and Selby 1999 p.337) or the actual creator of the page. Of greater concern to institutions is whether responsibility lies with the creator, the institution, or the individual of an institution providing links to resources from the institutions website. This still remains something of a grey area, although since technology plays such a big part in everyday life it is an issue that needs addressing. Orme suggests that lessons could be learned from US legislation, where steps have been taken to ensure universal access in Amendments to Section 508 of the Rehabilitation Act (1998) (http://www.section508.gov). The Act now makes it "unlawful for Federal Agencies to purchase inaccessible systems" (Orme 2001), although there are still areas which need further clarification. For example it is generally accepted that contrasting colours should be used for pages, but "Section 508 does not have such a provision requiring adequate contrast" (Thatcher 2002).
Implementation of measures outlined in the DDA under "goods, facilities and services" is required to take place between 1999 and 2004, by which time service providers must have taken "reasonable steps" to meet the needs of disabled people and to be providing an equitable service to all. As providers of services, libraries must ensure their buildings are accessible and must consider alternatives to service provision for those who cannot physically visit the library. Libraries should also provide alternative means of accessing resources, such as reading aids and alternative formats. This will help break down barriers to the information society, in particular to information that is available electronically.
Exemptions under the goods, facilities and services section of the DDA originally excluded the educational provision of institutions, such as universities. The Special Educational Needs and Disability Act (2001) (Great Britain 2001) (Part 4 of the DDA) will be implemented in September 2002 and although covering services already included in Part 3 of the DDA will affect all education and training provision in higher education, including services such as libraries. The new legislation "require(s) UK universities to provide access to assistive technologies such as Braille readers and speech encoders (and) requires that material placed on the Web is accessible" (Doyle and Robson 2002). However, there are some areas of the new Act that require further clarification. For example, it is still unclear whether an individual providing an accessible web site, but with links to inaccessible resources is in breach of the Act. A Draft Code of Practice to the Special Educational Needs and Disability Act 2001 has been produced by the Disability Rights Commission (http://www.drc-gb.org/drc/documents/post16cop.doc), which should help clarify such issues. As well as the Code of Practice, a comprehensive guide to the provision of accessible curricula has been produced by the University of Wales Cardiff, the Learning and Teaching Support Network and TechDis, and includes details about universal design and web accessibility (Doyle and Robson 2002).
3.3 Design considerations for human-computer interaction.
3.3.1 The users
Users can be categorised in many different ways: novice or expert; occasional or frequent; literate or illiterate; well-equipped or resource-starved; and so on - and each of these categories represents a spectrum rather than an absolute. Few of the possible categories will be mutually exclusive. While visual capability can form a useful categorisation - or spectrum of capability - it is important to recognise that this is but one more way of categorising individuals and that the same multiple underlying differences would characterise any population being studied.
When considering users and their needs, good system design will take account of users’ characteristics in relation to the tasks they are performing as well as considering "user characteristics independently of tasks" (Hackos and Redish 1998 p.23). Indeed there is considerable evidence to suggest that task is one of the most important of all variables (Griffiths et al. 2002). A task-oriented view would suggest that a number of specific questions need to be answered during the design and evaluation of any system. Examples include the users’ perceived relationship with their work, how the users envisage using the system (e.g. occasionally, frequently, within particular contexts), previous experience performing similar tasks and motivation to perform the specific task (Hackos and Redish 1998). Clearly in order to answer such questions it is also important to establish who the users are, or are likely to be.
In any research study there is a trade-off between the number of variables which can be investigated and the depth of investigation. While not neglecting the characteristics of the user samples selected, NoVA has been primarily concerned to explore use in depth with small numbers of users rather than superficially with a large sample. As a result, no claims are made for significance of results in respect of user variables other than visual impairment. Other studies which have explored user characteristics in depth need to be consulted for this perspective.
3.3.2 Information gathering as a physical process
Human beings gather data through the five senses: vision, hearing, taste, smell and touch. It is generally accepted that for most people vision is the most important and the most used and is the one they will concentrate on first. For this reason the design of human-computer interfaces tends to concentrate on vision first and foremost, and at present is concerned almost exclusively with only three of the five senses, vision, hearing and touch. Taste and smell do not generally feature in serious interface design.
Vision is a complex function, which not only involves the physical capture of light (including shade, colour and contrast) by the eye itself but also interpretation by the brain. Theories of perception take us well beyond the scope of NoVA (although see section 3.3.3 below for some brief comments on this topic), but it is important to note the range of physical disabilities which can result in visual impairment. So, for example, the degree of visual acuity and visual field will impact on the quality of sight, a loss of visual acuity causing objects to become blurred. Restricted field of vision will have an effect on the user’s ability to view several pieces of information on one screen - in frames for example. Colour, hue, brightness, contrast and ambient light will each have a different effect on different users, and for those with visual impairment they may effectively result in exclusion.
For someone who is blind or severely visually impaired, and effectively unable to process information visually, the most important senses will probably be hearing or touch. Sound can be used as a locator and as an alerting device, not just for visually impaired people but also for sighted people as "people respond more quickly to auditory signals than to visual signals" (Faulkner 1998 p.27). Auditory output is increasingly moving "beyond the regular error beep" (Lowgren 1993 p.91) used in many applications and of course speech output is now in the mainstream (though perhaps not for web-based applications). However, a number of research projects have looked specifically at the production of auditory sounds to depict graphical icons to enable access to graphical user interfaces by blind and visually impaired people (see 3.3.5 below).
Touch can be a useful way of presenting information for someone who is blind or deaf-blind, or for users working in a very noisy environment. Touch has been important in the development of keyboards: research has found that users can be affected by the feel of a system and touch has been used to provide feedback on "simulated physical manipulation" (Lowgren 1993 p.92). Tactile images, graphs and maps are also used to display visual information to someone who is unable interpret the screen visually. Tactile or "feelable" mice have also been developed which make different vibrations according to different areas on-screen such as the tool bar or the edit box.
In essence these approaches all help to make information available by using a range of sense-data and thus compensating for impairment. However, the success of this approach is entirely dependent on whether both the content and the structure of the information object can be successfully translated in this way. Users are participants in this process and are able to adapt themselves to the available technology and systems; NoVA was interested in how this adaptation works in practice, and how successful users are in developing strategies which use available technology to overcome potential and actual access problems.
3.3.3 The interpretation of information through cognitive processes
As noted above information is interpreted not only through the senses, but also through cognition. Simply seeing, hearing or touching provides primary data but for users to be able to turn data into information they need to have a "cognitive framework upon which to hang their understanding" (Rowley and Slack 1998 p.25). Cognition is "what enables the perception of objects and events" (Faulkner 1998 p.12).
The cognitive framework is only one of the models which needs to be elucidated if the process of understanding data, and turning it into information and knowledge, is to be understood. Again this takes us well beyond the scope of NoVA itself, but it is worth noting here that one of the ways in which systems designers seek to aid cognition is by using analogies. So, a dustbin (or trash can if the software is of US origin) is used to depict the action of deleting files. This can be useful because familiar actions from other environments provide powerful aids to cognition in unfamiliar ones. However, the approach must be used with great care. For example, the analogy of a card catalogue may be too limited to map onto the reality of an OPAC and may mislead the user into thinking that all actions will be analogous - the debates which at one time raged over the concept of the "main entry" in computer-based catalogues form a good example of the confusion that can arise.
Among the other cognitive models which need to be considered, learning styles may be particularly important in the context of electronic information systems, though this is an area that has attracted only limited research. Rowley and Slack (1998) use Kolb’s learning cycle to demonstrate a perspective on adult learning based on four learning styles: activist, reflector, theorist, pragmatist - all of which can be mapped onto the learning cycle (Rowley and Slack 1998 p.32). Brophy (2001) points out that "learning models can be related to personality types" and that "influences range from past experiences, habits of thought, basic personality, education, specific environmental demands from the learner’s chosen profession, current job, and specific task in hand".
As an example of the relevance of this approach, research into cognitive styles and their influence on information seeking behaviour undertaken by Kim suggests that the Field Dependent (FD) and the Field Independent (FI) styles are likely to impact differently on the success of the user searching in a non-linear web environment. FD users tend to navigate in a linear way whereas FI users move freely around the system, trying a "higher number of access attempts than FDs" (Kim 2001 p.235). User testing found that FIs tended to be more successful in their searching than FDs if they were novice searchers, although experienced FD and FI users tended to have similar information seeking behaviour. If users whose natural inclination is to adopt an FI approach are forced into FD type searching by their visual impairment, there is a clear danger of disadvantage. Again, however, this introduces another type of variable into the analysis and would require further research.
3.3.4 Human-computer interfaces
A further area of research which is of considerable relevance is that of human-computer interaction (HCI) and in particular its influence on the design of interfaces. Marchionini defines an interface as a communication channel which "serves as an intermediary between the user and the database" (In: Church 1999 p.6). HCI itself has been defined by Booth as "the study of the interaction between humans and computers" (Booth 1989 p.4). The following areas of HCI research are of particular importance:
· Users’ methods of interaction and how these affect communication
at the interface.
· More broadly, how users interact with computer systems as a whole.
· How well systems meet users’ task requirements and information
needs.
· User centred design and development: improving the design and development
of systems by placing the user at the centre of the design process.
· The impact of the computer on individuals within an organisation and
on the organisation as a whole.
Dillon & Watson have suggested that further research is needed into user differences and how these could influence interface design. Different interface types for instance may work best for particular groups of users (Dillon and Watson 1996 p.620): an obvious example is the provision of separate interfaces for beginners and experts, though this is now becoming subsumed into research on personalisation. Aspects such as domain knowledge, general computing experience and more specific application experience are all of relevance. This approach is stressed by Church who suggests that "an evaluation of the receiver’s traits should be an initial step in instructional design" (Church 1999 p.6) and also by Green who, referring to the cognitive approach, states that "it starts from considering the mental life of the user to reveal what he or she knows about the interface and how that knowledge is put to use" (Green 1986). Other variables could include the user’s background and the types of web pages a user typically visits (James 1998 p.27).
The nature of HCI has evolved over the years. Marchionini and Komlodi (1998 p.92) track the development of HCI from one of an interface designed for specialists who acted as professional intermediaries to satisfy users’ information requirements, to the interface designed for generally computer literate end-users performing their own information seeking tasks. The current emphasis is on universal access for all whether computer literate or novice.
3.3.5 HCI Design for visually impaired users
In various fields there are examples of design for disabled users leading to access for all - such as the development of the talking book for blind and visually impaired people led to books on cassette tape, and a well known example is that of curb cuts, designed for wheelchair users but of benefit to people with shopping trolleys and pushchairs.
Chong states that "each step in the evolution of computer technology posed a different set of problems for the blind" (Chong 1994 p.53) although he goes on to acknowledge that some of the problems have been addressed, if not overcome. Graphical User Interfaces (GUIs) have posed particular problems for visually impaired users, not least because "they cannot interact with the computer using a mouse and pointing system" (Boyd et al. 1990) although this mode of interaction is often simply assumed by designers.
A considerable number of research projects have explored the development of systems which overcome the problems of the GUI. In some cases this is achieved by providing a dual access system which can be used by both blind and sighted people simultaneously. Systems developed include Soundtrack (Edwards 1996), Outspoken (Edwards 1996 p.48), Mercator (Mynatt and Weber 1994), GUIB (Gill 1993), (Mynatt and Weber 1994), (Petrie and Morley 1995), (Edwards 1996), HOMER (Savidis and Stephanidis 1995), Emacspeak (Raman 1996), DAHNI (developed for the TIDE ACCESS project) (Morley et al. 1998), (Petrie and Morley 1997) and the development of the audio HTML access (AHA) framework (James 1998).
Features include the use of non-speech auditory icons to help users to navigate around a page. These are everyday sounds used to substitute for a graphical icon - such as the sound of a dustbin lid crashing as an analogy for the waste bin icon or delete key. Earcons is another approach which uses sounds or tones such as a "bing" sound to denote movement from frame to frame. This helps with more abstract concepts such as the scrollbar which are difficult to translate (Mynatt and Weber 1994). The use of non-speech sounds is felt by developers to be more useful than speech because they are less time consuming to listen to and can hold a great deal of information in a relatively short processing time (Petrie and Morley 1998 p.207), although the users may have more difficulty in learning the meaning of auditory cues and not everyone finds them easy to learn and remember. Mynatt and Weber found that "although the auditory icons were readily learned, the intuitive nature of the interface suffered from the subjects’ frustration with identifying the auditory cues" (Mynatt and Weber 1994 p.169).
Auditory and vocal cues are used in the Audio HTML Access (AHA) framework. This is based on the idea that "usable audio renderings can be produced in the same way that usable visual renderings are ... directly from the HTML mark-up of pages" (James 1998 p.27) and follow a number of set principles:
· Vocal source identity: number of voices, context switches.
· Recognisability: sound identity, identity of prominent features and
identity of metaphors.
· Distraction: number of sounds, signalling tendency (e.g. to denote the
importance of tags), length of sounds, aesthetics (e.g. loud or soft cues, the
use of abrasive or annoying cues).
Different sounds have been chosen to relate to the expectations of a variety of users. For example, children will be provided with different sounds to those designed for adults (another example of personalisation). It is generally recommended that, in the same way that prolific use of colours or fonts on a page can be confusing or too "busy" and therefore should be avoided, "the total number of speakers (i.e. different sounds) should be small" (James 1998 p.28).
The problem of controlling the mouse and locating the cursor has been addressed in many research projects. Substitutes for the mouse have been developed, including tactile displays (Ebina 1999), touch tablets (Mynatt and Weber 1994), (Petrie and Morley 1995) and Braille characters (Petrie and Morley 1995), (Edwards 1996), (Ebina 1999 p.41), although there are differing opinions as to whether the mouse should be replaced or simply eradicated (Mynatt and Weber 1994). Non-visual methods for locating the cursor have been developed to help prevent users from losing their bearings within a webpage or parts of a page. These include an overview of information such as quick navigation through the links on a page, headings for each section of a page together with quick navigation through headings (Petrie and Morley 1997 p.49).
The Emacspeak interface is designed to provide spoken feedback into the application i.e. not relying on using a separate screen reader to access what is on screen: "a screen reader speaks what is on the screen without conveying why it is there" (Raman 1996) and will therefore provide more richly spoken feedback. The idea behind this system is to allow the use of either the visual application or the spoken one (or the two combined).
Work has also been undertaken to develop an interface that can be used by sighted and non-sighted people simultaneously rather than using two separate systems. Part funded by the TIDE project, the HOMER tool was developed to enable the use of integrated interfaces which would be accessible to blind, visually impaired and sighted users.
For someone with a disability such as a visual impairment, searching and navigation will have limitations ranging from mild to severe, depending on the level of disability. Basic principles recommended by Carey and Stringer (2000) can be applied to the design of navigation and search screens to enable easier use by people with limitations:
· Synchronicity and integrity: information presented in a variety
of formats (audio, visual, interactive) should be self-sufficient.
· Degrees of simplicity and complexity: users should have control over
the way information is delivered to them and should be able to define the level
of information required (basic, advanced etc).
· Multiplicity: establish a system to enable users to customise hardware,
software and information systems to meet their individual needs.
· Over-ride: users should be able to over-ride the presentation of the
original document on a temporary basis in order to render it accessible, but
this should be done without changing the original document.
Requirements of a navigation system for people with disabilities may include:
· The conceptual model: the ease of which the user can understand
the purpose and function of a system.
· Usability: the ease of which the system can be operated - either by
the end user of by an intermediary system of device.
· Durability: the ability of the system to handle different materials
and platforms.
· Aesthetics: resulting from a mix of the above requirements.
Carey and Stringer see the desktop metaphor becoming less useful due to the increasing complexity of computer functions and requirements. New models are emerging, but many have elements which make them difficult for disabled people to understand. The authors believe that navigation should be the starting point of any model and that a bi-directional network structure is a more useful model to adopt using a requirement that each level (or node) of the network structure has no more than nine links "and the maximum number of steps from any given start point not just the central node, to any other node does not exceed 9" (Carey and Stringer 2000 p.15). This model could be applied to web pages although it is recognised that "most web pages have more than nine links" (Carey and Stringer 2000 p.20) but it is felt that this could be overcome in part by observing user behaviour as to which links are most popular and grouping these into sets. The thinking behind this bi-directional model is in part based on observation of the interface layout of a phone-pad or a 3x4 grid, this addresses the durability requirement as this model can be adapted to different platforms such as mobile phone, web page or Palmtop.
The Towel project at Manchester University (http://www.man.ac.uk/towel/) aims to devise a framework for "identifying travel objects and registering them as either cues to aid travel or obstacles that hinder travel for visually impaired users" (Goble et al. 2000) - the aim being to minimise obstacles and maximise cues by using travel as a metaphor. This framework will then be fed into the design of better user agents and of improved web design to aid visually impaired (and other) web users. The authors suggest that whilst the work of the W3C and the Web Content Accessibility Guidelines cover issues of content presentation and navigation there is still much work to be done on how visually impaired people actually move around websites.
The Towel project identifies a number of issues relating to travel and mobility which visually impaired persons will need to address in order to achieve their travel task. For example, they need to be able to have advanced knowledge of any obstacles on the route (in other words a preview of what is ahead). Goble et al. (2000) describe this as "granularity" or "environmental cue availability" and suggest that the lack of a preview may cause people to bump into objects. As well as a preview of what is ahead, visually impaired people also need to have some form of a mental map of the route. This may be aided by the use of appropriate or "egocentric" terminology to describe points along the way (i.e. with less emphasis on the spatial and visual). It is also noted that people with visual impairments are also more able to relate to environments with regular features. Harper and Green (2000) suggest that a man-made environment such as a city is easier to relate to than a natural one, i.e. countryside. Finally, spatial information needs to be described in a more appropriate way to aid navigation and location and tracking of position.
3.4 Information seeking behaviour
3.4.1 General research into information seeking behaviour
Research into models of information seeking behaviour up until the mid-1970s is identified by Wilson as being mainly "document-focused studies" and "attempts to explore information needs" which were generally concerned with system use rather than user behaviour (Wilson 2000 pp. 51-53). The shift towards a "person centred" approach rather than "system centred" has developed since the 1980s and the work of Ellis, Dervin, Kuhlthau and Wilson are often cited with this shift.
Ellis' model, for example, includes features of information seeking behaviour rather than prescribed stages. This takes into account variables such as the task in hand and the individual. The model begins with "starting" and finishes with "ending" but suggests features such as chaining and browsing which may occur at different intervals during the process. Kuhlthau's model, which is related, has been developed to include feelings associated with various stages and activities such as initiation and selection (Wilson 1999 p.255).
Wilson presents a broader model of information behaviour in which:
information seeking behaviour arises as a consequence of a need perceived by an information user, who, in order to satisfy that need, makes demands upon formal or informal information sources or services which result in success or failure to find relevant information (Wilson 1999 p.251).
Dervin (1973) identified four elements, which comprise the information seeking environment of individuals:
· The individual information seeker.
· His or her information needs.
· Available information providers.
· Possible resolutions.
Of these four elements, six linkages comprise the information seeking network of the individual:
· Individual/information providers.
· Individual/information solutions.
· Information needs/information providers.
· Information needs/information solutions.
· Information needs/solutions to information needs.
· Information providers/solution to information needs.
With each link, barriers may arise denying effective information access to the individual. Dervin identifies barriers which might arise to deny effective information access, including societal, institutional, physical, psychological and intellectual.
At much the same time of Dervin's work, a study by Warner et al. (1973) related information seeking variables directly to a person's income, age and education. It was suggested that these variables influence whether a person tends to seek information more or less often and whether they tend to call upon a wider range of services. In a study of adults and the role of public libraries Zweizig found that both demographic and non-demographic variables affected information seeking patterns. Non-demographic examples could include the amount of book reading undertaken, the amount of community involvement, past use of professional sources, how open minded people were, knowledge of the library and the perceived credibility of the library as an information provider. Demographic examples are similar to those of Warner and include the level of education, gender and age (Zweizig 1973).
The above examples conducted in the 1970s looked at information needs of populations. Later research concentrated more on the information seeking individual as the basis for understanding and assessing individual information needs. For example Chen and Hernon identified physical barriers which are imposed when the individual is unable to "make contact with the appropriate information providers due to some physical consideration" (Chen and Hernon 1982 p.19). The example given was the absence of an alternative to library stairs for the "handicapped individual". This barrier to physical access would then have been a major consideration when providing access to information providers, i.e. the librarian.
Apart from the interface itself, important factors for consideration in the usability of networked information are the individual and the task (Nielsen 1993). In a review of research on Web search behaviour conducted between1995 and 2000, Hsieh-Yee identifies similar patterns emerging from 40 different studies (Hsieh-Yee 2001), although he points out that many of the studies lacked external controls. Comparisons between expert and novice searchers revealed that expert searchers have a more flexible approach to searching and browsing, tend to use more of the features offered and can cope better with problems that might arise. Novice searchers tend to approach a task in a more chaotic way and often rate success of a search with the ease of use and number of results returned. This last point was also identified in a study undertaken of students’ searching behaviour that which indicated that "the amount of time and effort required from the user matters more than traditional recall and precision" (Craven and Griffiths 2001).
Large scale unobtrusive user-based studies conducted by Spink and by Wolfram using the Excite search engine revealed that "users submit largely unique queries and engage in similar, non-persistent browsing habits" generally not viewing results past the first two pages. Fewer than 10% of those conducting searches used Boolean operators (Wolfram 2000) and where they were used it was found that "every second user made a mistake" (Spink and Xu 2000). Because data was gathered anonymously, both of these studies were unable to ascertain user motivation for searching and browsing. However, these and other studies revealed that users tend to submit short, basic queries which are rarely modified, though often to good effect, for as Jansen concludes, "the use of more complex queries appears to have a very small impact on the results retrieved" which as a result "may not be worth the increased effort required to learn the advanced searching rules or the increased risk of making a mistake"(Jansen 2000).
Marchionini states that it should be possible to present information in different forms which "better suit the needs of the information seeker" (Marchionini 1995 p.191) and that "the provision of minimal, standardised and culturally diverse interfaces will lead to rapid changes as more users acquire habits and preferences" (Marchionini 1995 p.178-179). The rapid development of networked information and communications technologies provides opportunities for radical changes in the services which, increasingly, users - including those who need to use "accessible" formats and systems in order to overcome visual or other disability - must be able to access and use.
3.4.2 Information seeking behaviour of blind and visually impaired people
There is little published work specifically on the information seeking behaviours of blind and visually impaired people, although some examples such as the study by Oppenheim and Selby (1999) and the work of Jakob Nielsen have emerged in recent years. Marchionini and Komlodi refer to a number of GUI/HCI research projects which have been undertaken to help enable universal access, but these are focussed more on product development than on the question of how people search (Marchionini and Komlodi 1998 pp. 103-104). Williamson et al. conducted a literature search on the information seeking behaviours of blind and visually impaired people (Williamson et al. 2000). Although they identified a number of studies conducted on the information needs of blind and VIPs (Astbrink 1996; RNIB 1998; Williamson 1995; Williamson 1998), they concluded that their search revealed "no major study of the information needs and information seeking behaviour of this group of people (sight impaired citizens)...".
As a follow up to their literature review, a study focussing on blind and visually impaired people revealed that sight impaired users need to be provided with a variety of methods for meeting their information needs. Like all people, sight impaired people have a "different set of strategies for finding the information they need for their daily living" (Williamson et al. 2000). Problems experienced when using the Internet were, unsurprisingly, related to the poor design of some web sites. However positive attitudes were expressed about the potential of the Internet for information seeking coupled with users’ desire for independence, which "appears to be rooted in the freedom to choose".
The question of how people develop search strategies and perform information seeking tasks is not explored in any depth in the above report but it is addressed in the Power of Nine report by Carey and Stringer (Carey and Stringer 2000). Five key characteristics in information handling are identified: access, apprehension, navigation, interaction, and expression. Although it is not clear whether these are derived from empirical studies.
Carey stresses that the navigation element needs further investigation as clarity of structure is particularly useful for the blind and visually impaired user. Users are able to find the information they want more easily if the sequence of the layout is consistent, but increasingly information is displayed out of sequence or in a way that loses sequence altogether (e.g. frames, where the user may be unsure which frame should be examined first).
In the Web environment, studies into usability have identified content organisation and navigation paths as the most important factors to aid web mobility (Goble et al. 2000). Other factors include link effectiveness and differentiation and destination prediction.
Goble et al. define a web page in terms of the page itself and the "user agent" i.e. the method used to present information to the user (Goble et al. 2000). For sighted people the user agent may refer to the web browser, but for the visually impaired person this may also include assistive technology such as a magnifier or a screen reader. The user agent could be a screen reader, which reads out the text on screen word by word and line by line (i.e. in a serial way) and forces users to listen to large blocks of text rather than allowing them to move around the page between headings and frames. Specialist browsers (BrookesTalk for example) allow more flexibility as they use the HTML elements to form headings, lists etc and can also cope with frames. Although assistive technologies enable visually impaired people to access information from a screen, Goble et al. (2000) found that moving around the page in a serial way does not support travel or mobility. Likewise specialist browsers tend to concentrate on sensory translation, but pay little attention to mobility within a page.
3.5 Conclusions
The background documentation analysed in this section has been used to inform the directions taken by the NoVA project. The range of issues to be considered is very wide, including accessible web interface design, legal and ethical requirements of the provision of digital resources, and user behaviour in web environments.
Although issues relating to the design of accessible web interfaces are covered extensively in the literature, gaps still exist in other areas. For example, much work is evident in the development of specialist systems such as speech output devices and specialist web browsers to improve accessibility. However, there is less evidence of research and development work on mainstream computer accessibility undertaken by HCI specialists, which should lead to implementation at the design stage rather than as an afterthought. Current and impending legislation (both in the UK and other countries) relating to the provision of services to disabled people may force designers and developers of systems to take accessibility issues more seriously. Many organisations are now developing their own accessibility policies to help ensure they meet legal requirements.
General information seeking models, such as those developed by Wilson and Ellis, have been used by HCI designers to aid the development of user-friendly interfaces. However, whilst there is evidence of work on the information needs of blind and visually impaired people, there is less work specifically on information seeking behaviours of this group (although in recent years the work of Nielsen’s team (Coyne and Nielsen 2001), and research such as the Towel project (Harper et al. 1999), have started to address user behaviour and usability issues). Learning and cognitive styles may also have a bearing on the way visually impaired users approach an information seeking task and how they navigate their way around websites.
In order to make recommendations for accessible digital library system design taking into account areas of both accessibility and usability, further understanding is needed of serial searching in non-serial digital library environments. This formed the core of the NoVA project’s work. Later sections of this report will address these issues.
[ Previous Section: Adapting for Visual Impairment ] [ Next Section: Digital Libraries]