Downloadable ContentDownload PDF
A comparison of Ctrl, Alt, and Shift-key assignment methods for accelerator-commands
This evaluation focused on accessing word-processing functions by testing the efficiency of predetermined key combinations often called "accelerator-commands," "keyboard-equivalents," or "keyboard-shortcuts." Only the alpha-key component of accelerator-command key combinations has previously been addressed by user interface guidelines (International Business Machines Corporation., 1992; Apple Computer Inc., 1987; Digital Equipment Corporation, 1988). This evaluation focused on two other accelerator-command components: accelerator-keys and the shift-key. The evaluation compared key combination assignment principles using a three by three between-subject design. The accelerator-key assignment principle was the first dimension with three conditions: using one accelerator-key, using multiple accelerator-keys for added mnemonics, and using multiple accelerator-keys for categories. The shift-key assignment principle was the second dimension with three conditions: not using the shiftkey, using the shift-key for added mnemonics, and using the shift-key for related commands. Each cell had different key combinations for the 60 commands. The subjects were to press the correct key combination when presented with a definition of the command. The test ended when the subject pressed the correct key combination 19 of 20 times (95% correct responses) in a single trial. None of the overall measures, trials to 95 percent, percentage correct on first trial, or total errors, were significantly different on any dimension. The principles used to predict the overall results including the Category Coding Principle, the Number Of Accelerator-Keys Principle, the Number of Shift-Keys Principle, the Related Coding Principle, and the Mnemonic Principle have the following flawed assumption in common: that by reducing a targeted type of component error (accelerator-key errors, shift-key errors, or alpha-key errors) the overall number of errors will be reduced. The results of the component error measures (accelerator-key errors, shift-key errors, and alpha-key errors) give application designers the following strategies to help predict the key combinations that are more likely to cause users to have the different kinds of component errors: � Key combinations that are more likely to have accelerator-key errors can be predicted better by using the Category Coding Principle than by using the Number Of Accelerator-Keys Principle. � Key combinations that are more likely to have shift-key errors can be predicted better by using the Number of Shift-Keys Principle than by using the Related Coding Principle. � Key combinations that are more likely to have alpha-key errors can be predicted using the Mnemonic Principle. By using the previous strategies to predict the key combinations that will have the most component errors, the following guidelines could be used by application designers to reduce the frequency of component errors if they can predict the frequency with which individual commands will be used: � To reduce the frequency of accelerator-key errors, 1) assign the most frequently used functions to key combinations using the primary accelerator-key and assign the less frequently used functions to a secondary accelerator-key or 2) use the Category Coding Principle. � To reduce the frequency of shift-key errors, 1) assign the most frequently used functions to key combinations without the shift-key and assign the less frequently used functions to key combinations with the shift-key or 2) use the Category Coding Principle for the shift-key. � As previously stated in the guidelines, to reduce the frequency of alpha-key errors, assign the most frequently used functions to mnemonic key combinations and assign the less frequently used functions to non-mnemonic key combinations.