- 4-
Technical Note
Scope of the survey
The Occupational Employment Statistics (OES) survey is a semiannual
mail survey measuring occupational employment and wage rates for wage
and salary workers in nonfarm establishments in the United States.
Guam, Puerto Rico, and the Virgin Islands also are surveyed, but their
data are not included in this release. OES estimates are constructed
from a sample of about 1.2 million establishments. Forms are mailed
to approximately 200,000 establishments in May and November of each
year for a 3-year period. The nationwide response rate for the May
2007 survey was 77.9 percent based on establishments and 73.5 percent
based on employment. The survey included establishments sampled in
the May 2007, November 2006, May 2006, November 2005, May 2005, and
November 2004 semiannual panels.
The occupational coding system
The OES survey uses the Office of Management and Budget's (OMB)
occupational classification system, the Standard Occupational Classi-
fication (SOC) system. The SOC system is the first OMB-required occu-
pational classification system for federal agencies. The OES survey
categorizes workers in 1 of 801 detailed occupations. Together, these
detailed occupations make up 23 major occupational groups, one of which--
military specific occupations--is not included in the OES survey. The
major groups are as follows:
Management occupations
Business and financial operations occupations
Computer and mathematical science occupations
Architecture and engineering occupations
Life, physical, and social science occupations
Community and social services occupations
Legal occupations
Education, training, and library occupations
Arts, design, entertainment, sports, and media occupations
Healthcare practitioner and technical occupations
Healthcare support occupations
Protective service occupations
Food preparation and serving related occupations
Building and grounds cleaning and maintenance occupations
Personal care and service occupations
Sales and related occupations
Office and administrative support occupations
Farming, fishing, and forestry occupations
Construction and extraction occupations
Installation, maintenance, and repair occupations
Production occupations
Transportation and material moving occupations
Military specific occupations (not surveyed in OES)
For more information about the SOC system, please see the Bureau of
Labor Statistics (BLS) Web site at http://www.bls.gov/soc/.
- 5 -
The industry coding system
The OES survey uses the North American Industry Classification Sys-
tem (NAICS). For more information about NAICS, see the BLS Web site at
http://www.bls.gov/bls/naics.htm.
The OES survey includes establishments in NAICS sectors 11 (logging
and agricultural support activities only), 21, 22, 23, 31-33, 42, 44-45,
48-49, 51, 52, 53, 54, 55, 56, 61, 62, 71, 72, 81 (except private house-
holds), state government, and local government. The U.S. Postal Service
and the executive branch of the federal government also are included. An
establishment is defined as an economic unit that processes goods or pro-
vides services, such as a factory, mine, or store. The establishment is
generally at a single physical location and is engaged primarily in one
type of economic activity.
The OES survey covers all full- and part-time wage and salary workers
in nonfarm industries. The survey does not include the self-employed,
owners and partners in unincorporated firms, household workers, or unpaid
family workers.
Survey sample
BLS funds the survey and provides the procedures and technical support,
while the State Workforce Agencies (SWAs) collect most of the data. BLS
produces cross-industry and industry-specific estimates for the nation,
states, metropolitan statistical areas (MSAs), metropolitan divisions, and
nonmetropolitan areas. Industry-specific estimates are produced at the
NAICS sector, 3-digit, 4-digit, and selected 5-digit industry levels. BLS
releases all cross-industry and national estimates; the SWAs release indus-
try-specific estimates at the state and MSA levels.
State Unemployment Insurance (UI) files provide the universe from which
the OES survey draws its sample. Employment benchmarks are obtained from
reports submitted by employers to the UI program. Supplemental sources are
used for rail transportation (NAICS 4821) and Guam because they do not re-
port to the UI program. The OES survey sample is stratified by metropoli-
tan and nonmetropolitan areas and industry. Samples selected in panels
prior to May 2005 were stratified using MSA definitions based on the 1990
Metropolitan Statistical Area standards. Beginning with the May 2005 panel,
the sample was stratified using new MSA definitions based on the 2000 Met-
ropolitan Statistical Area standards.
An annual census is taken of the executive branch of the federal govern-
ment, the U.S. Postal Service, state government, and Hawaii's local govern-
ment. In order to provide the most occupational coverage, larger employers
are more likely to be selected than smaller employers. The unweighted em-
ployment of sampled establishments make up approximately 65 percent of total
national employment.
Concepts
Occupational employment is the estimate of total wage and salary employ-
ment in an occupation across the industries surveyed. The OES survey defines
employment as the number of workers who can be classified as full- or part-
time employees, including workers on paid vacations or other types of paid
leave; workers on unpaid short-term absences; salaried officers, executives,
and staff members of incorporated firms; employees temporarily assigned to
other units; and employees for whom the reporting unit is their permanent
duty station regardless of whether that unit prepares their paycheck.
- 6 -
The OES survey form sent to establishments with more than 10 workers con-
tains between 50 and 225 SOC occupations selected on the basis of the sampled
establishment's industry classification. To reduce paperwork and respondent
burden, no survey form contains every SOC occupation. Thus, data for specific
occupations are collected primarily from establishments in industries that are
the predominant employers of workers in those occupations. Each survey form is
structured, however, to allow a respondent to provide detailed occupational in-
formation for each worker at the establishment; that is, unlisted occupations
can be added to the survey form. Employers with 10 or fewer workers are sent a
form with no occupations listed, and are instructed to fill in the occupations
for their workers.
Wages for the OES survey are straight-time, gross pay, exclusive of premium
pay. Base rate, cost-of-living allowances, guaranteed pay, hazardous-duty pay,
incentive pay including commissions and production bonuses, tips, and on-call
pay are included. Excluded are back pay, jury duty pay, overtime pay, sever-
ance pay, shift differentials, nonproduction bonuses, employer cost for supple-
mentary benefits, and tuition reimbursements.
The OES survey collects wage data in 12 intervals. Employers report the
number of employees in an occupation for each wage range. The wage intervals
used for the May 2007 survey are as follows:
May 2007, November 2006, May 2006, and November 2005
panels
------------------------------------------------------
|
| Wages
Interval |-----------------------------------------
| Hourly | Annual
------------|-------------------|---------------------
Range A | Under $7.50 | Under $15,600
Range B | $7.50 to $9.49 | $15,600 to $19,759
Range C | $9.50 to $11.99 | $19,760 to $24,959
Range D | $12.00 to $15.24 | $24,960 to $31,719
Range E | $15.25 to $19.24 | $31,720 to $40,039
Range F | $19.25 to $24.49 | $40,040 to $50,959
Range G | $24.50 to $30.99 | $50,960 to $64,479
Range H | $31.00 to $39.24 | $64,480 to $81,639
Range I | $39.25 to $49.74 | $81,640 to $103,479
Range J | $49.75 to $63.24 | $103,480 to $131,559
Range K | $63.25 to $79.99 | $131,560 to $166,399
Range L | $80.00 and over | $166,400 and over
------------------------------------------------------
May 2005 and November 2004 panels
------------------------------------------------------
|
| Wages
Interval |-----------------------------------------
| Hourly | Annual
------------|-------------------|---------------------
Range A | Under $6.75 | Under $14,040
Range B | $6.75 to $8.49 | $14,040 to $17,679
Range C | $8.50 to $10.74 | $17,680 to $22,359
Range D | $10.75 to $13.49 | $22,360 to $28,079
Range E | $13.50 to $16.99 | $28,080 to $35,359
Range F | $17.00 to $21.49 | $35,360 to $44,719
Range G | $21.50 to $27.24 | $44,720 to $56,679
Range H | $27.25 to $34.49 | $56,680 to $71,759
Range I | $34.50 to $43.74 | $71,760 to $90,999
Range J | $43.75 to $55.49 | $91,000 to $115,439
Range K | $55.50 to $69.99 | $115,440 to $145,599
Range L | $70.00 and over | $145,600 and over
------------------------------------------------------
- 7 -
Mean hourly wage. The mean hourly wage rate for an occupation is the
total wages that all workers in the occupation earn in an hour divided by
the total employment of the occupation. To calculate the mean hourly wage
of each occupation, total weighted hourly wages are summed across all in-
tervals and divided by the occupation's weighted survey employment. The
mean wage for each interval is based on occupational wage data collected
by the BLS Office of Compensation and Working Conditions for the National
Compensation Survey (NCS).
Beginning with the November 2005 panel the lower boundary of the highest
wage interval was increased from $70.00 to $80.00. The mean hourly wage
value for the highest wage interval was computed separately for NCS data
from 2005 for $80.00 and over, and from 2004 and 2003 for $70.00 and over.
The mean wage rate from 2006 was used for the $80.00 and over interval for
the May 2007, November 2006, May 2006, and November 2005 panels. The aver-
age of the 2004 and 2003 mean wage rates was used for the $70.00 and over
interval for the May 2005 and November 2004 panels.
Percentile wage. The p-th percentile wage rate for an occupation is the
wage where p percent of all workers earn that amount or less and where (100-p)
percent of all workers earn that amount or more. This statistic is calculated
by uniformly distributing the workers inside each wage interval, ranking the
workers from lowest paid to highest paid, and calculating the product of the
total employment for the occupation and the desired percentile to determine
the worker that earns the p-th percentile wage rate.
Annual wage. Many employees are paid at an hourly rate by their employers
and may work more than or less than 40 hours per week. Annual wage estimates
for most occupations in this release are calculated by multiplying the mean
hourly wage by a "year-round, full-time" figure of 2,080 hours (52 weeks by
40 hours). Thus, annual wage estimates may not represent the actual annual
pay received by the employee if they work more or less than 2,080 hours per
year. Workers in some occupations typically work less than full time, year
round. For these occupations, the OES survey collects and reports either the
annual salary or the hourly wage rate, depending on how the occupation is
typically paid, but not both. For example, teachers, flight attendants, and
pilots may be paid an annual salary, but do not work the usual 2,080 hours per
year. In this case, an annual salary is reported. Other workers, such as
entertainment workers, are paid hourly rates, but generally do not work full
time, year round. For these workers, only an hourly wage is reported.
Hourly versus annual wage reporting. For each occupation, respondents are
asked to report the number of employees paid within specific wage intervals.
The intervals are defined both as hourly rates and the corresponding annual
rates, where the annual rate for an occupation is calculated by multiplying
the hourly wage rate by a typical work year of 2,080 hours. The responding
establishment can reference either the hourly or the annual rate for full-
time workers, but they are instructed to report the hourly rate for part-time
workers.
Estimation methodology
Each OES panel includes approximately 200,000 establishments. The OES sur-
vey is designed to produce estimates using six panels (3 years) of data. The
full six-panel sample of 1.2 million establishments allows the production of
estimates at detailed levels of geography, industry, and occupation.
Wage updating. Significant reductions in sampling errors are obtained by
combining six panels of data, particularly for small geographic areas and occu-
pations. Wages for the current panel need no adjustment. However, wages in the
five previous panels need to be updated to the current panel's reference period.
- 8 -
The OES program uses the BLS Employment Cost Index (ECI) to adjust survey data
from prior panels before combining them with the current panel's data. The wage
updating procedure adjusts each detailed occupation's wage rate, as measured in
the earlier panel, according to the average movement of its broader occupational
division. The procedure assumes that there are no major differences by geography,
industry, or detailed occupation within the occupational division. The wage rates
for the highest wage interval are not updated.
Imputation. About 20 percent of establishments do not respond for a given
panel. A "nearest neighbor" hot deck imputation procedure is used to impute mis-
sing occupational employment totals. A variant of mean imputation is used to im-
pute missing wage distributions. The variant of mean imputation for wage distri-
butions also is applied to establishments that provide reports with occupational
totals but partial or missing wage data.
Weighting and benchmarking. The sample establishments in each panel are
weighted to represent all establishments that were part of the in-scope frame
from which the panel was selected. Based on the sampled establishments, sam-
pling weights are adjusted when six panels are combined. Sampling weights are
further adjusted by the ratio of employment totals (the average of November 2006
and May 2007 employment) from the BLS Quarterly Census of Employment and Wages
to employment totals from the OES survey.
May 2007 OES survey estimates. The May 2007 OES survey estimates are based on
all data collected from establishments in the May 2007, November 2006, May 2006,
November 2005, May 2005, and November 2004 semiannual samples.
Reliability of the estimates. Estimates calculated from a sample survey are
subject to two types of error: sampling and nonsampling. Sampling error occurs
when estimates are calculated from a subset (that is, a sample) of the popula-
tion instead of the full population. When a sample of the population is sur-
veyed, there is a chance that the sample estimate of the characteristic of inter-
est may differ from the population value of that characteristic. Differences
between the sample estimate and the population value will vary depending on the
sample selected. This variability can be estimated by calculating the standard
error (SE) of the sample estimate. If we were to repeat the sampling and estima-
tion process countless times using the same survey design, approximately 90 per-
cent of the intervals created by adding and subtracting 1.645 SEs from the sample
estimate would include the population value. These intervals are called 90-per-
cent confidence intervals. The OES survey, however, usually uses the relative
standard error (RSE) of a sample estimate instead of its SE to measure sampling
error. RSE is defined as the SE of a sample estimate divided by the sample esti-
mate itself. This statistic provides the user with a measure of the relative
precision of the sample estimate. RSEs are calculated for both occupational em-
ployment and mean wage rate estimates. Occupational employment RSEs are calcul-
ated using a subsample, random group replication technique called the jackknife.
Mean wage rate RSEs are calculated using a variance components model that accounts
for both the observed and unobserved components of the wage data. The variances of
the unobserved components are estimated using wage data from the BLS National Com-
pensation Survey. In general, estimates based on many establishments have lower
RSEs than estimates based on few establishments. If the distributional assumptions
of the models are violated, the resulting confidence intervals may not reflect the
prescribed level of confidence.
Nonsampling error occurs for a variety of reasons, none of which are directly
connected to sampling. Examples of nonsampling error include: nonresponse, data
incorrectly reported by the respondent, errors in the administrative data used to
create the sampling frame, mistakes made in entering collected data into the data-
base, and mistakes made in editing and processing the collected data. Every at-
tempt is made to minimize nonsampling error through survey methods such as data
editing, imputation methods, and benchmarking of data to current employment totals.
Bureau of Labor Statistics
