SOLVING GEOSPATIAL TASKS IN 2D AND 3D VISUALIZATIONS OF MAPS - Information Visualisation: Techniques, Usability and Evaluation - page 241

Experiment II: Combinatorial geospatial assessment
It can be concluded from previous research that motion parallax is an
important factor for exploiting three-dimensional visualizations. Motion
parallax and dynamic occlusion require the visualized content to be three-
dimensional geometries with sufficient spatial complexity, such as
investigated in [15][16] and [17].
Visualizations of maps such as described in experiment I, or
visualizations of other tilted planar structures as, for example, studied in
[12][18] and [19], present fairly simple two-dimensional content,
represented in a 3D context. The absence of complex self-occluding 3D
structures in those presentations might explain why 3D visualizations have
so far not shown any significant effects.
The second experiment described here emanates from the experiment
described in the previous section, in [19], and is inspired by the work of
Bleisch et al. [18]. The objective is to investigate whether more complex
geospatial tasks can benefit from 3D representations. In the same context,
it is also analysed whether user performance is related to other factors than
visualization.
Task and stimuli
For this experiment, a task was chosen that required a combinatorial
assessment of two quantities - the height of bars as well as their distance
in 2D map space. The objective was to judge relative height differences
among a number of bars located in geographic positions, and to evaluate
them in relation to the distances between the bars.
Out of a number of simultaneously visible bars, participants had to
decide on the pair of bars with the largest difference in height relative to
their distance in map space. In a 2D visualization this task can be
considered as the evaluation of the lengths of imaginary 2D line segments
by mentally judging their proportions. The same is true for a 3D
presentation of bars; however, in 3D there exists an alternate approach that
implies immediate assessment of the slopes of imaginary lines connecting
the bar ends. There exist in 3D visualizations, quite obviously, some
ambiguities as to the true slope angles of perceived lines, which in the
strong 3D visualization, to some extent, can be resolved by interactively
examining the 3D bar diagram from different viewing angles. Visual
conditions tested in this experiment were 2D and strong 3D according to
the definitions above.