Introduction:
More sharply than in the past, the future of official statistics surveys is framed by data quality–cost tradeoffs dictated by current and anticipated budget pressures. In a larger sense, the issue may be decision quality–cost tradeoffs (Karr, 2012), because society may deem the resultant decisions— not the data—to be the end product of official statistics. In either case, whether official statistics agencies will be participants or bystanders as events unfold remains to be seen. The most pressing short-term need is to ensure that quality–cost tradeoffs be informed by scientific knowledge and reasoning. Efforts to do this, we believe, are hindered by a fundamental gap: currently, survey science is not to any meaningful extent a laboratory science. The World’s Simplest Survey Microsimulator (WSSM) is a step toward filling this gap.
To explore three issues—need, utility and feasibility—surrounding simulation models for Federal surveys, in April 2011 the National Institute of Statistical Sciences (NISS) sponsored an interdisciplinary Workshop on Microsimulation Models for Surveys. Details and supporting papers for the workshop are available.1 Cox (2012), and Cox (2013) delve deeper into these issues and offer a case for development of a simulation laboratory for Federal surveys. Karr et al. (2012) presents a prototype design for WSSM.
tr181.pdf