Geography Reference
In-Depth Information
The GUI layouts appeared in random order to avoid learning effects (however,
the task order was not manipulated). As indicators for the students
performance,
the
number of mouse clicks
(how many clicks does it take to solve a task?) as well as
the
mouse-click position
(do they click in the right places?) and the
required time to
solve the tasks
(how long does it take for them to solve the tasks?) were recorded on
each design. A
maximum limit
of 10 clicks and 20 seconds per task and layout type
was implemented and participants were informed about these limits. After
succeeding in or failing in a task, an immediate response was given by means of
an alert. Hypertext Markup Language (HTML) image maps and jQuery
2
were used
to provide layout interactivity on the client-side. A node.js
3
application stored the
test data in a SQLite
4
database on the server-side.
After having dealt with the tasks, the students were asked to rate the five GUI
layouts in terms of attractiveness and usability for a second time. In addition, they
had to choose their favorite layout. By asking them to rate the attractiveness of the
design before and after executing the tasks, we were able to determine the influence
of the task solving process on the GUI attractiveness rating.
The final part (Part D) was optional and not part of the core study, three topics on
specific atlas functions were inspected: (a) the placement and behavior of an
information panel, (b) the labeling of tool buttons, and (c) the use and behavior of
tool panels. These topics were aimed at revealing the preferences of digital natives
on the accessibility of atlas tools and thematic information. The study closed with a
text field for qualitative remarks and suggestions on how to improve the atlas
concept (and the survey). The entire study was conducted in German.
'
Results
We carried out a statistical analysis based on collected data, i.e., the responses to
questions, mouse click positions and recorded task completion times. Performance
(effectiveness and efficiency) and attractiveness measures are derived from these
data and their mean scores were compared between the five GUI layouts. Individual
differences were identified by pairwise significance tests at a 95 % confidence level
(i.e., p-value
0.05). To explore the clicking activity per task and layout, maps of
the first and other clicks as well as density surface maps were created with the
Heatmap plugin
5
of Quantum GIS. We calculated success rates per task using SQL
statements in the database, and used these as reference values for comparison.
Most tasks were completed successfully using Layout 4 (Table
1
); that is, in
93 % of cases neither the time limit of 20 s nor the limit of 10 clicks was reached.
<
