Current and Alternative Sources of Data on the Science and Engineering Workforce
Appendix E. Characteristics of Establishment-Based Data Collections
The section "Establishment-Based Sample Design" discusses the possible use of an establishment-based data collection for SESTAT and possible sampling frames for establishments. This appendix provides more detail on the sampling frames developed at the Census Bureau and at the Bureau of Labor Statistics. It also includes a description of the Census Bureau's National Employer Survey.
Available Establishment Lists
As described, Dun's Market Identifiers represent the only nationally comprehensive list of establishments available to surveyors outside the federal government. Two other comprehensive lists exist at the Census Bureau and at the Bureau of Labor Statistics. The data can be accessed only by those agencies because of federal confidentiality laws.
The Census Bureau's Standard Statistical Establishment List (SSEL) is assembled from data provided by the Internal Revenue Service (IRS) and the Social Security Administration (SSA). The Census Bureau conducts the Company Organization Survey annually, which helps establish the necessary connections between the IRS and SSA data files.
The Bureau of Labor Statistics (BLS) conducts the Occupational Employment Statistics (OES) Survey. The OES includes annual samples of approximately 400,000 establishments, taking 3 years to fully collect data from a total sample of 1.2 million establishments. The 1997 OES sample frame did not include establishments with fewer than five employees. (Sampling of establishments with one to four employees began in 1998.) According to BusinessUSA, about 70% of companies in the services sector (SIC codes 70–87) and 63% of companies in the computer services sector (SIC 737) have fewer than five employees. In calculating OES estimates, establishments with five to nine employees were weighted more heavily in an effort to compensate for the undercoverage. To the extent that establishments with fewer than five employees differ from those with five to nine employees, the problem of coverage bias may remain.
OES data collection is conducted primarily by mail and is managed by state employment security agencies using samples and procedures provided by BLS. For very large establishments, site visits are used to ensure that the data for the company are collected. Officials from the sampled establishments respond to the survey.
National Employer Survey
The National Employer Survey (NES) is one example of many establishment-based surveys that are conducted in the United States. It is carried out by the Census Bureau, using the Bureau's own SSEL, for the U.S. Departments of Education and Labor. First conducted in 1994 and repeated in 1997, the NES was a telephone survey of employers of more than 4,000 private establishments. In the 2000 survey, employer-respondents received a batch of questionnaires to distribute randomly to their employees. Since the 2000 survey subsampled the sample of employers from the 1997 survey, it is not fully representative cross-sectionally for 2000, but it should be able to provide some longitudinal information based on a 1997 representative cross-section. Public-sector employers, not-for-profit institutions, establishments with less than 20 employees, and corporate headquarters were excluded from the sample.
The jointly funded Educational Quality of the Workforce (EQW) project based at the University of Pennsylvania determined the NES survey design and content and analyzed its resulting data. EQW is primarily interested in studying school-to-career paths and workplace training among workers, so survey items tend to focus on such topics. According to EQW, "…the EQW-NES provides an important baseline for understanding how employers recruit workers, how they organize work, which educational credentials and experiences they use in screening job applicants, and what role education and training play in providing a skilled workforce." The EQW-NES differs from other national surveys because it focuses on the interaction of establishment practice, organizational structure, and workforce proficiency; documents how employers satisfy their need for skilled employees (in particular cataloging employer attitudes toward schools as likely suppliers of skilled employees); and measures the outcomes of both formal and informal training.
As noted above, the 1997 NES sample size was more than 4,000 establishments. The 1997 survey did not sample employees. The 2000 NES, which did sample employees, was half the size of the 1997 survey. Another administration of the NES was planned for 2003. At its current sample size, however, the NES is too small to meet NSF's data needs.
 The survey includes private establishments classified in the agricultural services; mining; construction; manufacturing; transportation; public utilities; wholesale and retail trade; and finance, insurance, real estate, and services industries. All state and local government establishments as well as private and government establishments in selected Standard Industry Classification (SIC) codes are included. See http://www.bls.gov/oes/oes_emp.htm#scope for a complete list of the NAICS codes corresponding to the industries that are included in the survey. Note that only nonfarm employment is included and all self-employed individuals are excluded from this survey.