Recommendations

During the course of the experiment, certain shortcomings were observed. Also, this experiment was purposefully less-than complete due to the added demands required for a complete study.

The most common complaint received about the Email Clients (all four) centered around the menu handling. The users typically wanted the items to be highlighted when the mouse icon was over them. Also, there was a strong desire for the menus to collapse themselves after a command had completed. In short, because the email client didn't mimic the norm in its command placement and behavior, users were less enthusiastic in their initial impressions.

The reason menus didn't collapse after executing a command was initially based upon the belief that this would give the user information that the command had been executed - adding another level of feedback. Also, if the menu collapsed after the command was initiated, it was thought that the user wouldn't be able to click multiple items while waiting for the command to complete. After further thought on the matter, however, the menu could have been collapsed at the end of the command execution -thereby allowing multiple mouse clicks, as well as supplying the user with the desired menu characteristic.

As for not highlighting command names when the mouse was over them. This wasn't an option only because it was not known how to execute a command based upon mouse position (in Visual Basic). This might have reduced the number of multiple clicks (in all versions) because the borders surrounding commands weren't readily apparent.

Another shortcoming that was observed centered around the actual feedback window. More than one user attempted to close the window while taking the test. This -rather than closing the window- closed the current frame within the window, allowing the underlying frame -of a different feedback type- to appear. While it isn't evident that this affected the study, the program would have been improved if separate windows were created for each feedback type rather than using a single window with multiple frames.

This study was far from an exhaustive study on the effects of different types of feedback. One significant type of feedback, in use today, is the changing of the mouse icon. Many systems have hourglass and/or spinning disk images which can be used to indicate when the system is in a busy state. It is hypothesized that changing the mouse icon while executing a command would produce results similar to the feedback window which contained only text. Adding changing icons to this study would have added another dimension - thereby increasing the number of experiments needed - as well as changing the time-requirements, thus it was not pursued.

Further study is needed to see if there would be significant differences in the results with a change in icon treatment upon issuance of a command. Another aspect that might change the results is the existence of a feedback zone within the main program window -rather than a separate window containing feedback. These two items need to be studied to determine if they cause differences in a user's practices during delays.

Department of Computer Sciences Direct questions and comments to the student editorial team