Conclusions

Work To Be Done

Because of the vast amount of data collected in this experiment, further analysis may yield additional results. The following questions should be addressed:

1. Can the task be reduced in difficulty without compromising accuracy in order to shorten the data collection time?
2. Does the dynamic task accurately characterize human performance for an office application?
3. Would changing the number of valid trials required, currently set at five, yield better results?
4. What are the characteristics regarding training and fatigue for these long duration tasks?
5. Is there a more appropriate selection of target characters (i.e. one that would result in fewer invalid trials)?

We hypothesized that we could predict which targets were missed and/or mis-identified. We have defined for each subject a search speed in characters per second. By assigning a value for a space to be less than one character and to a line feed to be greater than one character, we should be able to correlate the time of target location with the position of the target in the text string. McConkie has indicated that through the use of gaze tracking, he could obtain the data on errors more directly.

If this technique is extended to Asiatic cultures, where a literate subject can recognize a more complex character set, some conclusions may be reached regarding search for icons.

Several display manufacturers have indicated a willingness to supply advanced displays utilizing digital micro-mirror devices and plasma techniques in the 1280 x 1024 pixel resolution, 20-inch diagonal class. If a proper size subject pool can be recruited, the use of the test techniques from this study could provide valuable data for these devices.

Advice for System Designers

Today’s new display technologies offer possibilities of novel applications and systems by virtue of higher resolution, larger or smaller display sizes, lighter weight, lower power, smaller footprint, etc. In many cases, these benefits carry a cost in terms of the production cost of the systems. They may also carry a cost or a benefit in terms of human performance.

This study suggests a technique for determining the difference in human performance. Proper task definition is critical, both in the administration of the test, and in the interpretation of the results.

For text search in an office environment, our experiment indicates that there may be a very small performance penalty for the AMLCD in the dynamic case.

Continue
Return to the Title Page for Comparison of AMLCD and CRT Displays