Perhaps the most interesting session was on usability testing with Nora Dimmock from the University of Rochester. They have a usability program that works with programmers and designers on digital projects. The usability group deals with 5-6 projects each year; new homegrown digital products, evaluating and suggesting changes to vendor products, and revising existing systems. The workshop went through a number of usability techniques including card sorting in which you give 21 users cards with one element from your website in each card (for instance a library website might have journals, books, articles, hours, staff, news, events, ask-a-librarian, departments, and policies) and ask the user to group them in a way that makes sense to them. Also the reverse, when you put flip charts on the wall with headings and ask users to put postit notes with when they would expect to find under each heading on your website for instance. Both these tests reveal the users’ mental model of the service. Heuristics, when you ask expert users, like librarians, to critique a digital product based on specific criteria. This latter one has the added advantage of enabling librarians to have input into web design etc. something that is politically useful since librarians have very strong, but usually conflicting, ideas about website design and like to be consulted. Finally, she took us through assessment, when you have a user complete 5-6 tasks on a website (e.g. "Find a book on fish farming.”) and see where they click and how long it takes them. This can be combined with the Talking Out Loud Protocol in which the subject articulates their thinking while they are working on the task.
Rochester used to then write copious reports on the results of the tests. Now they just record them using Morae and send the file to the designers and programmers.
Oh, one final point. They often save expensive programming and designer time by creating paper mock-ups of the website and having users work through those. This means that big problems can be fixed before it ever gets to the programming stage.
OK, I lied: one more point. In a few years Rochester may not even have a stand alone website. They are considering moving to widgets to be inserted in appropriate places, often at the users discretion, through the users’ personal “infosphere.”