At the International Conference on Establishment Surveys in 1993, Esbjorn Ohlsson presented methods for coordinating sample selection across surveys and within repeated surveys. The approach assigns a permanent random number (PRN) to a unit as it enters the survey frame. The Bureau of Labor Statistics used these methods while redesigning its Occupational Employment Statistics and Current Employment Statistics surveys and developing a new design for the Job Openings and Labor Turnover Statistics survey.
This paper provides examples of what happens when theoretical concepts meet practical application. We discuss the process of assigning PRNs to records as they are added to the frame. We then focus attention on the methodology that was applied to: 1.) update business establishment surveys to capture a representative portion of these new establishments; 2.) update the samples to compensate for units that have gone out of business; and 3.) refresh a portion of the units for sample rotation purposes. Finally, we provide some solutions that were implemented to reduce solicitation cost and overcome other problems resulting from direct application of the theoretical method.