Page 398 - Special Topic Session (STS) - Volume 2
P. 398
STS507 Katherine Jenny T. et al.
estimates not considered in the initial design. Again, these applications would
be constrained by reliability and confidentiality mandates. In any case, the
appropriate solution is rarely obvious; research is required. Of course, time and
resource constraints can be prohibitive, so the research problem may be
simplified to allow for a transparent and repeatable solution, thus delaying the
unaddressed details or unforeseen nuances until the implementation stage,
leaving little time or resources for additional research.
Efficiency frequently dictates the research and implementation processes.
In the Economic Directorate of the U.S. Census Bureau, it is a common practice
to establish a “dedicated” research team with a fixed duration comprising
representatives from the relevant job series with differing experience levels,
perhaps utilizing matrix management. Team responsibilities encompass
defining and scoping the research problem, obtaining data, designing and
conducting the research, writing and testing programs needed to carry out
the research, documenting the findings, and presenting the “data-driven”
recommendation. Assuming that the recommendation is endorsed, a
subsequent implementation team is established. This team’s composition can
differ greatly from the research team, as expert staff are required – and
production programmers must be included in the discussions – although
some overlap in membership between the research and implementation
teams is desirable. As with the research team, the implementation team usually
operates under a fixed deadline. Team responsibilities include writing
specifications that implement the research recommendation while addressing
the issues that were deliberately unaccounted for in the research. Logistical
issues such as coding, testing, and validation are likewise included. An
education component is not unusual, as the implementation team members
may be unfamiliar with the methods under consideration.
The Economic Census (EC) is the U.S. Government's official five-year
measure of American business and the economy. The 2017 EC leadership team
endorsed a number of innovative updates, each introducing a new set of
methodological and production implementation challenges. This paper
focuses on the introduction of the North American Product Classification
System (NAPCS) (see https://www.census.gov/eos/www/napcs/more.html).
The EC collects a core set of data items from each establishment called general
statistics items: examples include total receipts or value of shipments, annual
payroll, and number of employees in the first quarter. In addition, the EC
collects information on the revenue obtained from the sales of good and
services (hereafter referred to as “products”). Although data on over 8,000
different NAPCS products are requested, evidence from prior economic
censuses indicates that many products are rarely reported. Legitimate zero
values are expected for the many products in an industry, at both the
individual establishment and total industry levels. Respondents can report
387 | I S I W S C 2 0 1 9