Information Technology Reference
In-Depth Information
interleaving of goals, no interruptions, a single method, error-free, expert perfor-
mance, and so on. Indeed, KLMs can be developed without worrying about goals
and selection rules. These limitations make the KLM easier to use, but they also
highlight the types of tasks that cannot easily be modeled using the KLM and
where it will not give accurate results.
Quite a considerable effort has gone into trying to make KLMs more usable—
particularly by building computer tools to apply them (Beard et al. 1996 ; Nichols
and Ritter 1995 ; Williams 2000) and to extend them to new types of interfaces
such as mobile GPS systems (Pettitt et al. 2007 ). These tools provide a language
for defining KLM and GOMS models and computing the task time based on the
analyses, although sometimes it may be just as easy to use a spreadsheet.
11.7 Considerations When Choosing a TA Method
There are some general issues that you need to be aware of when deciding whether
to carry out a TA to inform system design, as they will influence how successful
the TA is. Superficially it appears quite an easy technique to use, but to carry out a
TA that produces useful results that can inform system design is an acquired skill.
It requires skills in collecting the appropriate data or task specifications, analyzing
that material, and then interpreting it.
It may seem obvious, but to perform a TA, you need to have access to a detailed
enough task description. This can either be written (documentation) or captured
through interviews and observation. If you are using interviews and observation,
you need to make sure that you can get access to the appropriate people in a timely
manner. For new or unimplemented interfaces you can often base your analyses on
existing interfaces and extrapolations. If you are analyzing an unbuilt interface you
should note your assumptions very clearly.
Different types of TA place emphasis on different aspects of task performance.
You need to be aware of this. If you are interested in the work context, for
example, then HTA will help; if you are more interested in the cognitive aspects of
the task, you will be better off using some form of CTA. The different methods are
not mutually exclusive.
In general, TA cannot easily simultaneously represent users with different
levels of knowledge, skills, and abilities. Novice users and the way a task is
learned, for example, are not well represented by most of these techniques. TA
techniques, when applied sympathetically, can help support novice users, but this
has to be done by the analyst. Most methods can deal with some aspects of task
learning, such as how much knowledge has to be learned in order to do the task. To
model the time course of learning you may need to develop a more detailed
cognitive model (e.g., see Paik et al. 2010 ; Ritter and Bibby 2008 ).
Most methods of TA are strongly goal-oriented. These methods cannot gen-
erally be applied to situations that have little structure or only very abstract goals,
such as play, conversation, and team building. All TA should be done with an
Search WWH ::




Custom Search