A Post by Michael B. Spring

(A list of all posts by M.B. Spring)

Links (July 23, 2009)

Over the last decade, I have worked with Armando Rotondi on a number of projects that explore the use of web technologies for various forms of medical interventions. Several of the studies are related to individuals with schizophrenia, and explore the delivery of Family Psycho Educational Therapy to individuals with schizophrenia. As a part of that study we designed a website, based on extensive usability studies with the subjects, that would be easy for them to use. We had done exhaustive literature reviews and found no clear guidelines for web accessibility for individuals with Severe Mental Illness (SMI). SMI would include not only schizophrenia, but traumatic brain injury, Alzheimer’s disease, attention deficit hyperactivity disorder, and other conditions. We are currently working on a project for the Veteran’s Administration in which we will do a controlled study of the usability of more than 4000 controlled website designs that will be varied across twelve dimensions. We are examining things such as reading level, number of links, link construction, page length, etc. It is pretty easy to begin to develop a list of design guidelines if you consider the following simplified summary of the ideas that are generally supported in the literature. Individuals with SMI may: Having had good feedback on the website we designed to be as flat as possible using very explicit links, we are now hoping to be able to experimentally define concrete metrics that can be used by designers to build websites that are more accessible to individuals with SMI. As one might guess, several of the working hypotheses have to do with links. Given the earlier work where we had developed a system that used links that were very explicit, and were generally longer, we decided that one condition in our current study would be the number of words in links –links of length 3 or less versus links of length 7 or more. This gave us a very concrete metric that could easily be communicated to designers if validated by the study. As we began to write the code that would generate the sites by adding and subtracting words, we noticed that something was wrong. The original site we had developed had many links such as: But we also had links such as: In designing the test sites, we set as one condition links of 3 words or less versus links with seven words or more. We realized that some of the links that we made longer were no longer as clear as they had been. Going back to think about how we had abstracted the dimensions to be tested, we realized that the links we had constructed were concrete and explicit. In general, that made them longer, but sometimes a very concrete explicit link was actually very short. For example, “What is Schizophrenia?” was concrete, explicit and short. The dilemma for developing guidelines is that “explicit and concrete” is harder to quantify than links of seven or more words. In addition, as we examined our generated pages we realized that we had not lengthened the links in a constant navigation bar we placed at the top of every page. The constant navigation toolbar is used in all the sites we have designed for various medical research projects. The sites have between five and seven items in this bar such as “Home”, “Library”, “Journal”, “Chat”. They contain what I think of as a safety net. (You can always get home.) In general, we try to use a number of pages that an individual would want to jump to from anywhere. Other researchers refer to these pages as landmarks. In general, the labels associated with these links are one or two words. They also contain the most important hub pages. For example, a lot of our work has to do with online support groups where users spend as much as 90% of their time. For these sites “Support Groups” is an element in the constant navigational toolbar. This leads to a more precise operational statement about the links designed for individuals with SMI. First, links that are designed to assist a user in answering a question or obtaining specific information should be as long as they need to be to explicitly articulate the content that will be found if the link is traversed. They should be constructed such that they are simple sentences if possible. Any words or phrases in the statement that add extraneous on confusing should be removed. Second, the site should provide a direct access/recovery mechanism that is constant across all pages on the site. The goal of this navigational feature is not discovery but recovery. The goal of the navigation bar is to provide direct access or recovery of a known state (e.g. the home page). It provides the ability to simply and quickly hyperjump to a location of regular and known value. The most obvious of these is the “Home” menu item. Others might include things such as “Help” or “Contact Us”. Direct access buttons would include items that a user might want to get to from any place on the site. Again, these will depend on the nature of the site and the purpose for which people use it. These links should be as few as possible, as clear as possible and as simple as possible. There may be as few as one or two and as many as six or seven.