Anthropometrics: Important Aspects of Users’ Bodies - Foundations for Designing User-Centered Systems - page 61

keys but they may also be describing a keyboard that requires too much or too little
pressure to operate the keys. You may also hear users talk about a soggy feel to the
key action, which means that they cannot clearly tell whether a particular key has
been pressed or not, based on the key's motion.
Typing is still the most common way of entering data into computer-based
systems (including mobile devices). It is thus useful to briefly consider typing in
more detail and to identify some of the regularities of behavior in this area. Card
et al. ( 1980 , 1983 ) provide a good overview of the general issues of human typing,
whilst Salthouse ( 1986 ) focuses on transcription typing.
Users' typing speeds range from a few words per minute (wpm) to over
100 wpm, although there is often a trade-off between speed and accuracy. It is
sometimes said, for example, that expert typists achieve speeds of 90 wpm with
90% accuracy. The differences in speed are primarily due to practice—those that
type more tend to get faster at it, using two hands rather than single-finger or hunt-
and-peck typing styles.
Typical typing speeds tend to occupy relatively narrow ranges for different
categories of typist: novices can generally type at least 10 wpm; good journeyman
programmers are closer to 30-60 wpm; and users whose jobs includes large
amounts of typing are more likely to achieve rates above 60 wpm. We can
extrapolate from these numbers to find keystroke times of about 750 ms per
keystroke for slow typists ranging down to 125 ms per keystroke for fast typists.
These numbers are useful for predicting reaction times and for working out how
fast the computer has to respond.
Many users can type some keys faster than others. For example, the ''n'' key
(average typing time 221 ms) takes longer to type than the space bar (155 ms), as
shown in Fig. 3.6 (Card et al. 1983 , p. 62). This result suggests that when you
design interfaces you should consider associating keys that are faster to type with
actions that have to be performed frequently. There are exceptions to this rule,
however, which include situations where you deliberately want to slow down the
user so that they have time to consider the effect of their actions (such as in safety
critical systems), and where memorability is important.
Keyboard users make errors. Salthouse's ( 1986 ) review of transcription typists
reports error rates in the range 1-3.2%, noting four types of observable errors: (1)
substitution of letters (''work'' for ''word''), (2) intrusions (''worrd'' for ''word''),
(3) omissions (''wod'' for ''word''), and (4) transpositions (''wodr'' for ''word'').
These categories are not exclusive, and sometimes may be mistaken for each other.
Only about half (40-70%) of these errors are caught by the typist and can account
for 35% of expert typists' time (Landauer 1987b , p. 151).
The Typist model (John 1996 ) summarizes many of the behavior regularities of
typing noted above. Typist can be used to explain how typing occurs and how
errors arise. Its structure and behavior can also be examined to identify where
interfaces for typing could be improved, and where changes would not make a
difference. In transcription typing, for example, providing the typists with an
additional paragraph of material to be typed is not useful because transcription
typists do not look that far ahead.