1983 Paper

A new operating console for air-traffic controllers was under development in the research department of Pressey Inc., the giant Japanese electronics consortium.

The conventional style of operating console included a small monochrome display screen on which appeared symbols indicating aircraft locations, together with upper case alphanumeric messages about aircraft identity, speed, and altitude. Monitoring events on the display, the controller's job was to direct air traffic by issuing vocal commands to pilots over a radio link.

In an attempt to improve the quality of information available to air-traffic controllers, the console was re-designed to include a large colour display screen on which symbols and alphanumeric characters could be presented in various colours to contrast with the background screen colour; in addition, the display was altered so that the alphanumeric characters could appear in both upper and lower case.

Pressey's Human Factors department was called in to evaluate the new display. They made use of an information sheet issued weekly by the civil aviation authority on controller's performance (with the conventional monochrome upper-case display), which gave for each controller the mean response time to an event on the display, and the number of "near misses" (occasions aircraft were allowed to come within 500 metres of each other), averaged over six-hour shifts.

Performance with the new colour display was estimated for ten volunteer air-traffic controllers who each attended for a one-hour testing session in an air-traffic control simulator, using a console with the new display. For each participant measures were obtained of the time to respond to events on the display, and of the number of "near misses" that occurred. Data from ten randomly selected individuals were extracted from the aviation authority's figures to compare with the data gathered from the volunteers in the simulator. Average performance is shown below.

Mean response time (msecs) Mean "near misses"
Conventional monochrome 527 0.05
upper case display [min=352,max=1342]
New colour upper & lower 470 0.3
case display [min=324,max=1570]

A t-test on mean response times revealed a significant difference between the two types of display (t = 2.45, df = 18, p < 0.05), but the difference between "near misses" was not significant (t = 1.87, df = 18, n.s.).

The scientists in Human Factors reported that both use of colour and provision of upper and lower case characters were desirable improvements in the display, since they each resulted in faster responses without a concomitant increase in serious errors.

The design for the new operating console was handed over to the Production department, who complained that the new display screen would be so expensive that the console would be too costly for their customers. They suggested using a smaller colour screen, and preserving the larger size of symbols and alphanumeric characters. Human Factors were given the task of determining whether the smaller display would have a significant effect on performance.

A two-factor experiment was designed to compare performance with the large and small displays, and to determine the symbol and character size giving better performance. No air-traffic controller volunteers were available as participants, so the scientists selected as participants ten secretaries chosen randomly from the typing pool, reasoning that unskilled users were acceptable in this case since they wished simply to compare performance with different displays. Each participant received 15 minutes pre-training, and then four daily one-hour sessions in the simulator. The order of experimental conditions was: large display, small symbols; small display, large symbols; small display, small symbols; large display, large symbols. As before, response times and "near misses" were recorded for each participant. Mean response times (in msecs.) averaged over all ten participants are tabulated below:

Display screen size
Large Small
Symbol and Large 508 590
character size Small 623 546

The mean response times were entered into a 2-way within-subjects analysis of variance. There was no significant difference in performance between conditions using the different display screen size (F(1,9) = 1.52, n.s.), but the difference due to symbol and character size was significant (F(1,9) = 7.90, p < 0.05), with large symbols and characters giving faster mean responses. The interaction between symbol size and display size was also significant (F(1,9) = 8.74, p < 0.05).

Since there was no effect of display size and on average the fastest responses were obtained with large symbols, Human Factors recommended that production of the console should proceed as suggested, using a small display screen having large symbols and characters.

[Back to Criticise-the-Experiment List] [Back to Main Page]