Experimental evaluation of dialogue types for data entry

Abstract An experiment evaluating four basic types of data entry dialogue with regard to speed showed a clear interaction between type of dialogue and type of data. A form-based dialogue was fastest for many ordered data, while a command dialogue was faster for unordered data. There was no significant difference between the number of errors at each dialogue type. The lowest performance times across data types were reached with an extended command dialogue with a built-in facility for command-less, ordered input. This dialogue type gave low performance times for all data types. A study of the subjects' preference regarding interaction mode in the command dialogues showed that most subjects preferred an interaction mode which optimized speed at the cost of less feedback and less opportunity for error control. This can be expressed in terms of dialogue gears: a majority of the subjects chose a high gear of interaction. This did not, however, result in a higher error rate for those subjects.