A predictive model of menu performance

Abstract

Menus are a primary control in current interfaces, but there has been relatively little theoretical work to model their performance. We propose a model of menu performance that goes beyond previous work by incorporating components for Fitts' Law pointing time, visual search time when novice, Hick-Hyman Law decision time when expert, and for the transition from novice to expert behaviour. The model is able to predict performance for many different menu designs, including adaptive split menus, items with different frequencies and sizes, and multi-level menus. We tested the model by comparing predictions for four menu designs (traditional menus, recency and frequency based split menus, and an adaptive 'morphing' design) with empirical measures. The empirical data matched the predictions extremely well, suggesting that the model can be used to explore a wide range of menu possibilities before implementation.

Improving Window Switching Interfaces

“…The number of distinct windows switched to per day ranged from 3 to 177, with a cross participant mean of 39 (s.d. 19). In their study of Web revisitation, Tauscher and Greenberg [20] define the recurrence rate R as the probability that any URL visit is a repeat of a previous visit, giving R=(total URLs visited -distinct URLs visited)/total URLs visited ×100.…”

“…Zipfian distributions have been observed in many areas of computer use [19], such as frequency of command use and menu use. Although window and application switching activities are among the most frequent in computer use, with one study showing a mean time of 21 seconds between actions [1], we are unaware of any empirical studies of users' patterns of revisitation to windows and applications.…”

“…It is well developed and can be extremely fast: studies have shown that items can be found in time proportional to the logarithm of the size of the set, which can be much faster than the linear search time needed for unorganised sets [19]. This fast performance is enabled by spatial constancy -that is, items remaining in the same location over time.…”

Action graphs and user performance analysis

“…Menu selection (Cockburn, Gutwin, & Greenberg, 2007;St. Amant & Horton, 2007) is a special case where the goal is to make a selection, and where each selection has only one way to make it.…”

“…One might think that reducing the number of actions a user needs to achieve a goal reduces the time required for the task; this is certainly true in applications like menu selection (Cockburn et al, 2007) where each menu selection has a unique sequence of user actions. However, if there is more than one way to do a task it is possible that a faster way can be found for doing a task that nevertheless requires more actions.…”