|
Low-fi Testing
First paper prototype
This prototype was the paper manifestation of the desired
feature set, and an amalgamation of our sketches. The user
first accessed a login dialog, and after authentication came
to the main window. This consisted of an exploration panel
in the center, for the chart display, with surrounding panels
used to:
- Select surveys and saved datasets (sets of surveys). Clicking
the "Manage Datasets" button on this panel brought the user
to the dialogs used to create and edit datasets.
- Navigate through the categories in the selected survey
or dataset. By clicking on a category, all the charts on
that survey page would be displayed. Our intention was that
by making this navigation mechanism so similar to the one
in the Reporting Tool, users would find it familiar and
easy to use.
- Save charts. Checking a box on the chart placed a reference
to the chart in a Library panel for later display or export.
Clicking on the Export button showed the dialog used to
save the charts and/or underlying data as specified by the
user. The main workflow for comparing desired charts against
one another was to first navigate to each of those charts
through the categories, saving each one into the library.
Then use the library to display just those charts in the
exploration panel.
- Make queries. Like the Filtering function in the reporting
tool, a query could be created that when applied, would
redraw the visible charts so they only showed the data that
met the specified criteria.
- Access advanced features. Below the chart display window
was a toolbar used to determine the type of chart in the
display, display statistics in a Stats panel, and display
comments.
Second paper prototype
The first paper prototype was very flexible and feature-rich,
but difficult for our users to navigate during testing. Our
users gave us a lot of really valuable feedback, and we found
the process to be so instructive that we immediately began
work on creating and testing a second paper prototype.
It was too cumbersome for them to have to navigate through
the categories, save the charts they wanted into a library,
and then show the charts from the library in the exploration
panel. It was clear that the library should not be a necessary
intermediate step. We realized that we were needlessly duplicating
too much of the existing reporting tool workflow, and that
users really wanted to lay out charts for comparison in a
crosstab format.
So we made headers for the rows and columns and put the controls
directly on the headers. We considered the hierarchy of information
in the database: Surveys contain Categories, which contain
Questions, which contain Responses (at that time the terminology
for "Responses" was "Answers"). We determined that only certain
combinations of these levels were meaningful when creating
a chart in one of the cells that represented an intersection
between a column and row. Users would be prevented from making
"illegal" combinations of the levels in the hierarchy.
To reflect this hierarchy, the column and row headers each
had a tabbed panel containing three tabs: Dataset, Question,
and Answer (there was no tab for Category, because although
it is essential for navigating to the desired question, comparison
against a Category is not meaningful). These tabs allowed
the user to change levels on the column and row. Rather than
having the Category, Question, and Answer controls on a side
panel, we placed them directly onto the column and row headers
so that they could serve as labels as well. To display charts,
the user selected information from the column and row tabs,
and charts appeared at the intersecting cells. A panel of
radio buttons at the bottom determined whether the grid of
charts displayed charts, data, or both.
We removed the advanced chart combination functionality in
this version. We wanted to get the crosstab workflow hammered
out first, and would add the combination charts later, time
permitting. Also, we substantially reworked the dataset and
export dialogs.
This time our users found the interface much easier to navigate,
and were able to complete the tasks rapidly and with a much
better success rate.
|
