Evaluation of a Driver Interface: Effects of
Control Type (Knob Versus Buttons) and
Menu Structure (Depth Versus Breadth)

Daniel Manes and Paul Green

November, 1997

In an initial experiment, four drivers were verbally cued to find items in a menu system while driving a simulator. Three factors were examined: (1) menu structures (deep with three levels of four items each and, broad with two levels of eight items each), (2) controls (a cursor and a number pad), and (3) control/display locations (both high, both low, and display high/control low). The cursor control led to faster task completion times than the number pad, and a full 32 trials per block offered no benefit over just 16, simplifying the design of the main experiment.

The main experiment was similar to the initial study, except 24 drivers participated, a knob control was used instead of the number pad, and 16 rather than 32 trials per block were used. Menu selection times (approximately nine seconds) were not significantly different for the two menu structures or the three control/display locations. However, the knob was significantly faster than the cursor (six percent), and there were 13 percent more lane excursions for the broad menus than the deep.

A GOMS analysis showed that downward scrolls required 0.4 seconds each, except for initial selections where thinking and reading processes lengthened times. Correlations between GOMS and actual selection times ranged from 0.5 to 0.8.

These data, suggest deep menus should be used to minimize lane excursions (a measure of driver distraction) and knobs should be utilized to minimize selection times. Mounting the display high and the control low is also advised.

Graphical Abstract (.pdf) | UMTRI-97-42 Full Report (.pdf)


Close This Window