Survey quality measures
Availability of data
The Occupational Employment Statistics Survey (OES) is conducted by the Bureau of Labor Statistics in order to obtain information about employment and wages in industry within the United States. SRS cosponsors this survey in order to obtain more detailed information about the employment and wages of scientists, engineers, and support technicians than would otherwise be collected and reported by BLS.
The survey samples approximately 400,000 establishments each year. In general, an establishment represents a single physical location and engages in one type of economic activity.
c. Key variables
- Hourly wages
- Annual wages
- Standard Industrial Classification (e.g., food and tobacco, textiles, and apparel)
- Standard Occupational Classification
2. Survey design
[Note: Much of the material below has been excerpted or reproduced verbatim from U.S. Department of Labor, Bureau of Labor Statistics: Occupational Employment and Wages, 2000 (Bulletin 2549, April 2002, "Appendix B. Survey Methods and Reliability of the 2000 Occupational Employment Estimates," pp. 131-138). Users are encouraged to consult this Appendix for more complete explanations. Until supplanted by later survey cycles, the 2000 Appendix B can be accessed at http://stats.bls.gov/oes/2000/oestec2000.htm. Thereafter, users should consult the OES Home Page at http://stats.bls.gov/oes/home.htm.]
a. Target population and sample frame
The population for the survey consists of nonfarm establishments located in the United States or its territories. The sample frame consists of units reporting to State Employment Security Agencies.
b. Sample design
The sample frame is stratified on the basis of area, industry, and size of establishment. The survey is conducted over a three-year cycle, with one-third of the sample units included in the survey each year. With some exceptions, units sampled in one year are not included in the sample the following two years. Establishments with 250 or more employees are included with virtual certainty over the three-year cycle of the survey. For establishments with between 4 and 250 employees, each size class is sampled with probability proportionate to the total employment within the size class. Prior to 1996 the survey was conducted in three annual waves: manufacturing industries, nonmanufacturing industries other than trade or regulated industries, and nonmanufacturing trade or regulated industries. The sample size for the survey of manufacturing establishments was approximately 150,000. For nonmanufacturing establishments it was approximately 300,000 and for trade and regulated industry establishments it was approximately 275,000. Starting with the 1997 survey it is anticipated that 1.2 million economy-wide establishments have been surveyed over each three-year survey cycle.
c. Data collection techniques
Survey forms are mailed to personnel offices at the establishments sampled or in a few cases to company headquarters. The instruments are tailored to each industry, so that detailed information is not requested for occupations unlikely to be found within the industry. For example, establishments in the communications industry are not specifically requested to provide information about biologists. (A census is taken of Federal Government establishments each year. Data representing Federal Government employment and wages are obtained from the Office of Personnel Management at the end of the survey process.)
Two additional mailings are sent to nonresponding establishments at approximately three-week intervals. Telephone or personal visit followups are made for those nonrespondents considered critical to the survey because of their size. Establishments that remain nonresponding are accounted for by a two-step imputation process. (See OES web site, cited above, for a detailed explanation of the imputation process.)
Establishments are asked about their employment for a given reference week, which varies from industry to industry (e.g., the weeks of October 12, November 12, or December 12). The lag between the reference date and the mailing date is determined by the individual States administering the survey and is generally about 2 months.
3. Survey quality measures
a. Sampling variability
The relatively large sample size for this survey means that estimates based on the whole population are subject to fairly low sampling error. However, estimates for small subgroups are subject to considerable sampling error. Estimates of sampling error for all cells are included in the Detailed Statistical Tables published by SRS.
There are possible coverage error problems, since the establishment frame is likely to be incomplete because of the births and deaths of establishments.
(1) Unit response --All States achieve at least a 75 percent unit response rate, with most States approximating 80 percent. Statistical imputation procedures are used to adjust for nonresponse within stratification cells. This procedure does not necessarily adjust for all nonresponse bias, since respondents and nonrespondents may differ within cells.
(2) Item response --Item nonresponse data are not published by the Bureau of Labor Statistics. A variation of the mean imputation procedure is used to impute for missing wage and employment data.
Both industry and occupation are difficult to measure and, therefore, subject to considerable measurement error.
Because questionnaires are tailored to individual industries, respondents are not specifically asked about employment in occupations that are fairly unusual in an industry. (Approximately 120 different industry questionnaires are administered to the 80 industries covered in the survey.) This may result in an underreporting of some occupations in each industry and an overreporting of residual categories, such as "all other professionals."
4. Trend data
Information from this survey was first available in 1977. Because the survey was conducted prior to 1996 in three annual waves (manufacturing industries, nonmanufacturing industries other than trade or regulated industries, and nonmanufacturing trade or regulated industries), information on any particular type of industry prior to 1996 was available only triennially. Starting with the 1996 survey cycle, information on all nonfarm industries was available annually.
Individuals interested in doing trend analyses are warned that there have been changes made to occupational and industrial classifications over time. For example, the classification of industries used in this survey was revised in 1987. More information about changes affecting trend data is included in the technical notes in the publications from this survey. Additional information may also be obtained from the contact person listed below.
5. Availability of data
The data from this survey are published by SRS in Detailed Statistical Tables in the series Scientists, Engineers, and Technicians in the United States. The most recent report in this series (NSF 02-313) is for 1998.
Information from the OES is also included in the Science and Engineering Indicators series.
b. Electronic access
Data from the OES are available in electronic format here and from the Bureau of Labor's OES web site at http://stats.bls.gov/oes/home.htm.
c. Contact for more information
Additional information about NSF's use of this survey can be obtained by contacting
Science and Engineering Indicators Program
Division of Science Resources Statistics
National Science Foundation
4201 Wilson Boulevard, Suite 965
Arlington, VA 22230
Phone: (703) 292-7813