Using predictors to partition menu selection times

Abstract

Selection times of drop-down menus are in many ways influenced by cognitive and motor processes of the user and by design variables of the menu. Since the number of these variables is too large, the contribution of individual variables to selection time cannot be assessed by using factorial designs. Multiple regression is introduced to solve this problem. The technique uses selection times as criterions and a set of general menu characteristics as predictors. The non-standardized slopes β report the increase ( or decrease) in selection time which can be assessed for each predictor. In a first experiment, the validity of the technique was demonstrated replicating various well-known effects in a mouse-driven editor. For example, the selection times increased with the number of subordinate menu items or atypical items. Further, due to motor components of the mouse movement, selection times depended on the spatial position of an item within the menu. In a second experiment, mouse selection was replaced by key selection to stress cognitive processes contributing to response times. The technique yielded results that were sensitive to this variation. Limitations of the technique are discussed.