Development and evaluation of direct manipulation lists (poster session)

Abstract

Item selection from a large list is an important task in many interfaces. This research examined a direct manipulation list for browsing and retrieving information. With direct manipulation lists several different retrieval methods can be considered such as scrolling, paging, using string search methods, or using an "elevator". The question asked was which retrieval methods would lead to fast and efficient retrieval for different list lengths and levels of user familiarity with the lists. A 3 x 2 between-subjects factorial design was used with list sizes of 25, 50, and 100 items and users who were either familiar or unfamiliar with the lists. In this experiment, only use of an elevator and paging were studied due to the common use of elevators in many state-of-the-art interfaces and an initial experiment which showed that scrolling was not an efficient method. The results showed that the list size and user familiarity with the list were both significant factors in the retrieval time. A more detailed analysis of retrieval performance revealed that these effects could have been influenced by users having difficulty estimating the position of an item. Moreover, users in the middle of a list tended to use a single method (such as paging), despite the predicted and observed efficiency of using both paging and the elevator. Based upon these results, a new method for direct manipulation is proposed which separates the elevator into regions with specified ranges of items so that the user knows where items are in the list. A task analysis of this method shows promise in ease of use as compared to other methods.