Abstract
An experiment was conducted to evaluate user performance under four different menu item arrangements: alphabetic, probability of selection (most popular choices are positioned near the beginning of the list), random, and positionally constant (consistent assignment of individual items to screen positions). During the initial stages of practice, the rule-based approaches produced faster mean search times, but after moderate amounts of practice, the positionally constant arrangement appeared to be most efficient. People seem to remember quite easily the location of items on a display, indicating that positional constancy can be an important factor in increasing the efficiency of the search of computer menus and other displays.
