Database Reference
In-Depth Information
still high, it has been reduced slightly by experimental manipulation. However, the
necessity of long observations may be shortened and results may be more readily
interpretable and specific questions are more likely to be answered.
Laboratory Experiment:
In a laboratory experiment the experimenters fully design
the study. They establish what the setting will be, how the study will be conducted,
what tasks the participants will do, and thus plan the whole study procedure. Then the
experimenter gets people to participate as fully as possible following the rules of the
procedure within the set situation. Carefully done, this can provide for considerable
precision. In addition, non-realistic behaviour that provides the experimenter more
information can be requested such as a 'think aloud' protocol [43]. Behaviour can be
measured, partly because it is reasonably well known when and where the behaviour
of interest may happen. However, realism is largely lost and the degree to which the
experimenter introduces aspects of realism will likely reduce the possible precision.
Experimental Simulation:
With an experimental simulation the experimenter tries to
keep as much of the precision as possible while introducing some realism via simula-
tion. There are examples where this approach is essential such as studying driving
while using a cell phone or under some substance's influence by using a driving simu-
lator. Use of simulation can avoid risky or un-ethical situations. Similarly although
less dramatically, non-existent computer programs can be studied using the 'Wizard
of Oz' approach in which a hidden experimenter simulates a computer program. This
type of study can provide us with considerable information while reducing the dan-
gers and costs of a more realistic experiment.
Judgment Study:
In a judgment study the purpose is to gather a person's response to
a set of stimuli in a situation where the setting is made irrelevant. Much attention is
paid to creating 'neutral conditions'. Ideally, the environment would not affect the
result. Perceptual studies often use this approach. Examples of this type of research
include the series of studies that examine what types of surface textures best support
the perception of 3D shape (e.g. [34, 38]), and the earlier related work about the per-
ception of shape from shading [39]. However, in assessing information visualizations
this idea of setting a study in neutral conditions must be considered carefully, as wit-
nessed by Reilly and Inkpen's [62] study which showed that the necessity for an in-
teractive technique developed to support a person's mental model during transition
from viewing one map to another (subway map to surface map) was dependent on the
distractions in the setting. This transition technique relates to ideas of morphing and
distortion in that aspects of the map remain visible while shifting. These studies in a
more neutral experiment setting showed little benefit, while the same tasks in a noisy,
distracting setting showed considerable benefit.
Sample Survey:
In a sample survey the experimenter is interested in discovering
relationships between a set of variables in a given population. Examples of these
types of questions include: of those people who discover web information visualiza-
tion tools how many return frequently and are their activities social or work related?
Of those people who have information visualization software available at work what
is the frequency of use? Considering the increased examples of information visualiza-
tion results and software on the web, is the general population's awareness of and/or
use of information visualization increasing? In these types of studies proper sampling
Search WWH ::
Custom Search