Usable?...Or Not?...Factors Affecting the Usability of Web Sites
by Rakhi Rajani and Dr Duska Rosenberg
This article contemplates factors affecting the usabilty of web sites. It addresses issues from both the disciplines of psychology and computing and attempts to consider the relevance of these issues with regards to a study carried out at the Centre for Information Environments Research at Brunel University. The study was developed in order to investigate issues affecting the usability of web sites. A test site was created and investigations revealed some interesting findings, some of which are detailed below.
Despite being a complex graphical user interface (GUI), the World Wide Web is very different to the more 'conventional' interfaces we use to create documents and manipulate data. With the focus upon information display and instant retrieval in a multimedia environment, the Web is a domain which must be instantly usable. It is a domain that is often expected to support all facilitative human communication modalities and thus designers of Web pages must focus upon the person whose primary goal is often information gathering and not manipulation of data within the computer. Thus, properties of colour, sound, navigation, placement and others must be considered from a very different perspective. For example, in most task oriented systems, sound is used to warn and inform (unless specific music oriented packages are being used). In a web site, on the other hand, sound is more often used to introduce the site, aid the user in navigation and create an atmosphere (Rajani, R. 1998). This being the case, a beep to inform users of an error is acceptable, but to create atmosphere and convey ideas, sound must be more complex and carefully considered.
This is just one example to illustrate our argument that design of web based media must include psychological considerations alongside computer design aspects. Let us first consider the perspective of human psychology and perception. The web is a primarily visual interface and so it is important to consider how visual information is processed, manipulated and interpreted (even if it consists solely of text). We need to consider an enormous array of possibilities including the effects of the use of colour, illusory effects, positioning of information, size and distance, light, contrast and image size - to name but a few! In order to instantiate a solid ground for developing a highly effective and 'pleasant' interface, these considerations must be paramount.
It is clear that the communication on the Web is multi modal in nature. Visual clarity is important, so too are auditory issues such as the speed of speech, language barriers, and accentual differences. It is also necessary to mention the immense importance of written language, speed of reading, text size and colour, cognitive memory aspects, knowledge issues surrounding 'pre-programmed' states and the everyday navigational issues that users grow so accustomed to. With such a global user base of 'clients' with varying ability and computational and personal backgrounds, is it possible to create a truly usable interface?
The problem that is emerging out of such human complexity is - can we approach the design of such an interface within a single, uniform scientific framework? The human factors issues are so vastly complex that it is natural for designers to try and simplify the problem definition. Technologists will often attempt to overcome real-world complexity in the belief that the solution can be found. This sound engineering principle leads to the creation of "infallible" design ideas and all manner of visual aids and documentation that is thought to provide the answer. Human factors issues are sometimes buried under an enthralling array of novel design ideas.
There are, however, limitations in the user environment that cannot be overcome by imaginative design ideas. Difficulties may arise because of slow download time, incompatibility of browsers, machines and software, inexperienced user abilities and many more. To understand and overcome this, designers must also consider the computational issues of Web usability. In practice, these often hinder us in our quest to provide creative usable web sites that are dominated by human factors. Here we need to consider the computational design issues relating to Web design as outlined in this table (from Rajani, R. 1998).
It may be argued that consideration of all of these factors will lead to a Web interface that is instinctively usable. The use of respected design techniques that apply to conventional interface design are assumed to apply to web design as well. However, usability studies that focus on the human factors issues do not support such complacency.
Nielson's usability study of some of the web sites of influential companies revealed that high technology was not as important as a site that had a 'high touch' feeling (Nielson, J. 1994). Users were more comfortable with pages that had fewer colours, greater means of communicating with the web master, more pictures of real people and appropriate metaphors. On the whole, less complex interfaces were preferred.
Another study carried out at the Centre for Information Environments Research at Brunel University provided similar results. The research question was to consider how users conceptualise web sites and results were sometimes (but not always!) surprising.
The results of usability testing showed that users who had a task to fulfil when browsing a site went into the site, found what they wanted, and then left! There was some 'surfing' if they found interesting information on route, but in most cases, if they had a set task in their mind, they used the site for this purpose and no more. Thus, they had to be able to find things quickly. If there was any unnecessary information in the way, this would often 'put them off the scent' of the information they were looking for. Indeed, such unnecessary information introduced greater complexity and prevented users from finding what they were looking for even if it was right in front of them.
One significant finding concerns "banner blindness" (Benway, J. P. 1998). There were clear navigational instructions at the top of the pages which provided users with information that would have allowed them to locate their information with extreme speed and ease. However, very few people noticed the instruction and thus failed to use the navigational aid. The majority of users spent a considerable amount of time searching various pages and then, about 10 -15 minutes into searching the site would see the instruction and say "oh, I just noticed that". Others did not notice the instructions at all. This was even more surprising because each user had been given time to browse the site at their own leisure without having a task to accomplish. It appears that users will frantically try to find the information they are looking for at the start of a goal-oriented search for information. But, as time passes and they are unable to locate the information, their perception widens and they begin to look for more clues - and the clues they find are the ones that the designers feel are obvious - but that the user has completely ignored at first glance! The issue for a designer is thus how such 'clues', navigational aids and instructions can be made more 'eye-catching'.
The method used in this study was mostly observation of users working on their tasks. User response was found to be influenced by the task requirements. Many users stated that content surpassed appearance when they were specifically looking for information, in spite of the fact that the 'appearance' of the pages was designed to provide users with visual and often graphical 'landmarks' to which they could relate information. This is in contrast to user response at all other times (such as when browsing), when they did use the appearance of a site/page to orient themselves.
As for navigation, most users looked to the bottom of pages for a list/row of links to other places. If these links were not there then they would stop for a second or two and re-familiarise themselves with the page before attempting to find another path. As expected, users brought with them their own knowledge and experience of navigation around a site, and if this information was distorted they would have to "reinitialise their receptors" of new information and then continue. In fact, user actions resembled those of someone looking for a drinking glass in someone else's kitchen - if they are unable to ask someone, they will often first look in a place where they themselves might keep their drinking glasses and then in every other 'logical' place.
In terms of concentration and 'living in the moment', it seems that users did just that. They lived in the moment and knew where they were within the site at a given point in time. The majority of users were barely able to remember where they had been for about 3-4 pages before hand but could vaguely remember other parts of the site according to images shown. However, they concentrated on 'the now' and did not concern themselves with where they were going next, unless they had to choose from a variety of options at the beginning of their task. That is, if an opening page had a list of options, they would create a mental road map of routes they would try, but once in the site they tended to live in the moment.
Generally, users tended to stay looking at a page if it contained interesting information, but more so if that information was coupled with a simple page layout, a 'light' and unfussy colour scheme and clear navigational aids (for example to show them a way out of the page!). If navigational aids were confusing, they would spend more of their time planning how to move on as opposed to concentrating upon content. It was as if they felt 'trapped' or 'imprisoned' within the page.
What came to light in this study was that users were possibly impressed by novel ideas and high technology, but not necessarily concerned that the capabilities of the technology were used to the full. If the technology was there they found it interesting and fun to able to use multiple channels in the interface. However, they found unexpected use of sounds for example, disturbing (perhaps because normal interfaces are not so inclined?). This was particularly noticeable when the deliberately pre-programmed interface sounds suddenly started playing or voices talking (as if in a normal conversational interruption). Then users literally 'jumped' out of their seats and were truly startled, even though such sounds would be considered quite normal in a situation that did not involve the use of a computer - the reactions shown here were not the same as say suddenly turning on the TV to find the volume at top setting and then also jumping out of your seat (although they were similar). Many users immediately tried to turn off the sound, showing their annoyance even if it was particularly relevant to the topic on the page. When given the choice however, most users would try out the sound, but would often turn it off after a few minutes. Their comments show that they appreciated the existence of the sound channel even if they did not wish to use it in a given task situation.
Therefore, the generally held assumptions that sites rich in colour and animation with high tech sound etc to provide a truly 'human' experience are good, have been shaken. The multimedia technology is a powerful aid to designing usable interfaces, but sometimes it is too powerful. It is true that multi-modal interfaces are capable of achieving greater degrees of ease of use, but this achievement is not directly dependent on imitations of human behaviour and perception.
It thus also seems that the design considerations pertaining to technology (as listed in the the table) are perhaps of more significant importance. The results of our study show that people tend to use the various channels on offer selectively, as they would in normal communication. It is as if they are accustomed to the fact that computing technology is not communicative in the human sense, but they are familiar with the communicational aids of the computer and so are often happy to use these as they find them.
In conclusion, there is no single and no right answer as regards the usability features of multi-modal interfaces, as their usability is dependent upon contextual influences. In some cases, to comprehend information, all human senses must come into play, whereas at other times, only some are more prevalent. Thus, is it really possible to demand 'ultimate' web usability when the influences that surround the issue are so vast and varying?
Rakhi Rajani (firstname.lastname@example.org) is a PhD student (researching Human Computer Interaction and Neuropsychological Rehabilitation, Visual Language and Computer Mediated Communication) in the Department of Information Systems and Computing at Brunel University in London, UK. For further information on web usability, take a look at Rakhi's internet site on the topic at http://www.rakhi.dircon.co.uk/webindex.htm--best viewed with MSIE4!
Dr. Duska Rosenberg (email@example.com ) is a lecturer in the Department of Information Systems and Computing at Brunel University in London, UK. Her research interests include Virtual reality; Natural Language; Situated communication. Take a look at Duska's Web site at: http://www.brunel.ac.uk/~csstddr.
Copyright © 1999 by Rakhi Rajani and Duska Rosenberg. All Rights Reserved.