EIA has evolved a unique approach to improving information quality in
a decentralized statistical agency. The idea is to give the program of
fices full line responsibility for the quality of their products, but
to rely on a quality assurance group to oversee their efforts. The qua
lity assurance group assures that these efforts are effective through
a systematic process of data and model evaluation, applied research to
address technical problems, and administrative procedures and standar
ds to promote improved practice. Together the program offices and the
quality assurance group keep the agency's leadership fully apprised of
quality concerns and actively involved in solving quality problems. A
fter reviewing the overall approach, this article describes the kinds
of evaluations that are done and the mix of administrative procedures
and standards used to accomplish changes. The article concludes with a
discussion of the kinds of problems that are amenable to this approac
h.