Rehearsal-based interfaces such as Marking Menus or FastTap are designed to enable smooth transitions from novice to expert performance by making the novice's visually-guided actions a physical rehearsal of the expert's feedback-free actions. However, these interfaces have not been extensively tested in real use. We carried out studies of the adoption of rehearsal-based expert methods in two dissimilar applications -- a game that directly rewards rapid selections, and a drawing program that has no particular need for urgency. Results showed very different patterns of use for the guidance-free expert method. In the game, participants quickly switched to sustained use of expert selections, whereas few users regularly used the expert method in the drawing program, even after ten weeks and more than 1800 selections. These studies show that rehearsal alone does not guarantee that users will switch to expert methods, and that additional factors affect users' decisions about what methods to use. Our studies also revealed several issues that should be considered by designers of rehearsal-based techniques -- such as perceived risk in making selections without visual guidance, the value of guidance that shows possible options in the UI, and training that reminds users of an expert method and motivates its use.

Proceedings of Graphics Interface 2015, Halifax, Nova Scotia, Canada, Canada, 3 - 5 June 2015, 223-231