Accuracy Metrics in Mobile Text Entry

M. Koivisto (Finland) and A. Urbaczewski (USA)


Text entry, mobile Internet, and quality of metrics


There are several metrics utilized to ascertain the relative merits of human computer interaction. In the field of text entry, the Minimum String Distance (MSD) and the Keystroke Classification (KC) metrics are both used to measure text entry usability. This paper examines text entry in mobile devices to see which metric is a better measure of text entry performance. Eighty-seven subjects performed three text entry tasks, each one utilizing a different text input method. Data were collected to calculate both KC and MSD metrics. Discriminant analysis was then used with each metric to classify the 257 cases to their input method. In measuring uncorrected errors, both MSD and the non-corrected error rate (NCER) component of KC were equally weak in classifying the case to its group. In all error and speed conditions, MSD was a much better classifier of the device utilized. The data show that this may be due to KSPC being a better measure of efficiency than accuracy. For accuracy metrics the time of the measurement was an important factor. Metrics measuring the errors in original string was superior to metrics measuring the errors after error correction. Further research is required to more fully understand these phenomena.

Important Links:

Go Back