A defining feature of the United States Federal statistics system is the decentralization of statistical survey activities among multiple agencies. Consequently, there is no overall approach to questionnaire development, evaluation, and testing (QDET) for establishment surveys among government agencies. Besides being subject to legal restrictions on data sharing, these agencies have different data collection goals and, to some degree, different target sub-populations within the overall population of establishments. As a result, methods for developing and testing establishment survey questionnaires have evolved somewhat independently.
The purpose of this paper is to identify and compare methods for developing and testing establishment survey questionnaires across four U.S. Federal agencies: the Bureau of Labor Statistics, the Census Bureau, the Energy Information Administration, and the General Accounting Office. In the paper, we raise common QDET issues for the establishment setting and explain notable differences among agencies.