Collaborative
Research: Quality Graphics for Federal Statistical Summaries
Supported by the National Science Foundation
through their Digital Government program.
The
Federal government distributes a vast quantity of statistical summaries
in printed and electronic form. The full wealth of information that
might be derived from these summaries is not being realized because
limited attention is paid to disseminating summaries in understandable
forms.
The
objective of this research is to develop and assess quality graphics
for federal statistical summaries. The development and assessment
process will consider perceptual and cognitive factors in reading,
interacting with, and interpreting statistical graphs, maps, and metadata
representations. The purposes of the quality graphics include exploration
by agency users evaluating data quality and looking for emergent trends,
decision making by public policy makers, and communication of statistical
summaries to the public.
The
proposed research addresses four topic areas: converting tables to
graphs and maps, representing metadata, interacting with both graphs
and maps, and conveying multivariate spatial and temporal relationships.
The research features use of Web-based “middleware” components to
provide rapid development of graphics for usability testing.
The
featured middleware is a Java Graphics Component Library (GPL). A
long history of research has recently culminated in a rigorous graph
algebra that is the foundation for the library. The library also benefits
from the collective intellectual effort being poured into Java. Principles
of human perception and cognition will be used to guide the construction
of statistical graphics, and usability tests will help to refine those
principles in the context of federal statistical summaries.
The
research team has been carefully selected to bring together outstanding
and compatible investigators. 
This
group provides expertise in statistical graphics, cartography, human
perception and cognition, and software development. The team is composed
of university, industrial, and federal agency researchers that have
the collective expertise needed to conduct the research. For example,
one industrial partner, AT&T, has addressed scalability issues and
has been dealing with five gigabytes of data a day, every day. The
collaborators from agencies endorsing this research (representing
BLS, NCI, NCHS, EIA, NASS, CENSUS, EPA, and BTS) provide the domain
expertise and vision to develop software products that will be useful
to the agencies.
Research
products include a better understanding of principles of cognition
and perception as applied to static and interactive statistical graphs
and maps, improved middleware components, and evaluated demonstration
software than can lead to production software for web-distribution
of statistical summaries. The graphics software will embed elements
fostering statistical literacy. Beyond the usual papers and conference
presentations, the milestones include annual multi-agency presentations
in Washington, D.C., to promote transfer of methodology.
The
research promotes communication advances sought by the President’s
committee on information technology. The results will impact the public
through better communication, understanding, and policy in the areas
of health, environment, energy, agriculture, employment and transportation.
