The Occupational Requirements Survey (ORS) publishes job-related information on physical demands; environmental conditions; education, training, and experience; as well as cognitive and mental requirements. The Occupational Employment and Wage Statistics (OEWS) survey produces employment and wage estimates for approximately 830 occupations. The ORS and OEWS programs survey business establishments (employers). BLS encourages users to consider the employment estimates from OEWS as the most reliable and representative occupational employment data produced by BLS.
ORS can be used to identify the percentage of workers that have specific job requirements. Combining this with OEWS data, users can derive estimates of occupational employment with specific requirements. Understanding the number of jobs with specific requirements can be useful to job seekers, vocational and rehabilitation experts, human resource professionals, hiring managers, safety and medical professionals, and the disability community. The estimates from these two programs can be used to understand the nature of work, benchmark job requirements, develop recruitment plans, and assess job risks. By combining data from these programs, the BLS can provide detailed occupational employment data such as the estimated number of jobs in an occupation with a specific job requirement, employment for jobs with a specific set of requirements, and a list of occupations meeting a selection of requirement thresholds. Users should consider each survey’s design, methodology, and reliability before combining estimates.
To estimate the occupational employment with specific job requirements, multiply the percentage of workers that have selected job requirements from ORS by the occupational employment from OEWS. Some job requirements may be grouped together and are considered additive, such as levels of strength (e.g., light or medium) or specific vocational preparation (SVP). Additive job requirements can be added together before calculating the occupational employment.
For example, to calculate the number of general office clerks required to have specific vocational preparation of a month or less:
SVP levels are additive, so 31.1 percent of general office clerks are required to have a month or less of specific vocational preparation.
There are 2,578,180 general office clerks in the national economy according to OEWS. Therefore, 2,578,180 multiplied by 31.1 percent is about 801,814 general office clerks in the national economy where the minimum specific vocational preparation required is a month or less.
Using ORS percentage estimates and OEWS employment data, estimates of the employment for multiple job requirements can be calculated. This is helpful for determining the lower and upper bounds of employment in a detailed occupation with multiple requirements. It is at the data user’s discretion to determine which bound best fits their purposes. When applying this method, users can combine stand-alone requirements as well as additive requirements by assuming job requirements exist similarly across different subsets of the population.
Consider an example using the following ORS requirement percentages for general office clerks:
Sedentary and light strength levels are additive, with 96.5 percent of general office clerks required to perform work with a sedentary or light strength level. Quiet noise intensity is used as a stand-alone estimate in this example.
There are 2,578,180 general office clerks in the national economy according to OEWS.
Assuming these job requirements exist similarly in different subsets of the population, this example can be refined further by multiplying the two job requirement percentages to the estimated employment from the previous step. This example assumes that a quiet noise intensity requirement exists uniformly across worker strength levels
Therefore, an estimated 776,032 general office clerks in the national economy experience a quiet noise intensity level, and of these, an estimated 748,871 have a strength level not exceeding light.
Extending the example above further, 26.0 percent of general office clerks do not have a supervisor present.
This requirement represents the lowest percentage of the requirements selected, meaning at the maximum, no more than 670,327 general office clerks meet these combined requirements. Again, this number can be refined further by assuming these job requirements exist similarly in different subsets of the population, and each requirement’s percentage can be multiplied together by the occupational employment.
Therefore, an estimated 670,327 general office clerks have no supervisor present, and of these an estimated 194,706 require sedentary or light strength and have a quiet noise intensity level.
Data from ORS can be used to identify occupations that meet user-determined thresholds for multiple job requirements. Once occupations meeting a combination of requirements are identified, OEWS can be used to find the total employment for those occupations. This is helpful for job seekers who would like to see the types and numbers of jobs available for their given set of skills and limitations.
This approach provides the total employment in occupations that meet requirement thresholds; it does not provide the number of jobs with specific requirements. Additionally, in cases where the threshold outlined covers more than one estimate from an additive group, it may be necessary to add estimates to identify occupations. This would create a broader estimate (such as “sedentary or light strength level”), rather than listing each requirement separately.
For example, consider the following occupational requirement thresholds:
General office clerks and insurance sales agents are two occupations currently published by ORS that meet this combination of thresholds. There are 2,578,180 general office clerks and 422,600 insurance sales agents in the national economy according to OEWS.
When considering the standard errors of combined estimates, one approach is to create a simple confidence interval around each of the estimates. Another more rigorous approach is to use a pooled standard error calculation to produce a confidence interval for the combined estimates. More information on standard errors and confidence intervals is available on the ORS standard errors page.
Using the example above, there are 2,578,180 general office clerks and the relative standard error is 0.4 percent. For SVP 2, the estimate is 28.4 percent, and the standard error is 3.2.
For the OEWS estimate, first convert the relative standard error to a standard error.
2,578,180 * 0.4 percent = 10,313
Next, calculate the OEWS confidence interval.
2,578,180 +/- (1.645 * 10,313) = 2,578,180 +/- 16,964 = [2,561,216, 2,595,144]
For the ORS estimate, calculate the confidence interval.
28.4 +/- (1.645 * 3.2) = 28.4 +/- 5.3 = [23.1, 33.7] percent
Using the lower bounds will provide a smaller number of workers with selected job requirements and using the upper bounds will provide more workers with those requirements.
Lower bounds: 2,561,216 * 23.1 percent ≈ 591,641 general office clerks with SVP 2
Upper bounds: 2,595,144 * 33.7 percent ≈ 874,564 general office clerks with SVP 2
Summary Calculation: (2,578,180 +/- 16,964) * (28.4 +/- 5.3) ≈ [591,641, 874,564]
Users interested in a more rigorous approach to estimating the variance of a composite ORS-OEWS estimate may consider using the variance of a product of two independent random variables.
Continuing with the same example, this formula can be used to estimate the variance of a composite ORS-OEWS estimate as follows. Note that the composite ORS-OEWS estimate is the product of the ORS estimate and the OEWS estimate. The variance of an estimate is equal to the square of its standard error.
= ((0.284)^2 + (0.032)^2 )*((2,578,180)^2 + (10,313)^2 ) - ((0.284)^2*(2,578,180)^2)
Therefore, the 90% confidence interval for this estimate is:
≈ (0.284*2,578,180) +/- (1.645*82,554)
≈ 732,203 +/- 135,802
Using the lower bound will provide a smaller number of workers with selected job requirements (approximately 596,401 general office clerks with SVP 2) and using the upper bound will provide more workers with those requirements (approximately 868,005 general office clerks with SVP 2).
OEWS provides employment estimates at multiple levels of geographic detail—national, state, metropolitan, and nonmetropolitan areas—while the ORS provides estimates representative of the national economy and not by geographic area. Users should keep in mind the level of accuracy required for their specific purpose, particularly when using these methods for state, metropolitan, and nonmetropolitan areas.
Users should be aware that although ORS data is not produced by state, metropolitan, or nonmetropolitan areas, certain requirement estimates may differ by region for a variety of reasons. Users should evaluate regional differences before combining estimates from ORS. The following are a few examples of factors that may cause regional differences in requirements.
The credentials required for some occupations may vary based on state and local laws. Some occupations only require licenses in certain states to operate. Here are a few examples of occupations that have license requirements in some, but not all, states: hairdressers, security guards, electricians, and plumbers. Electricians require a license to perform their work in many states, but in others, like New York, there are no laws requiring credentials, prior work experience, or on-the-job training for electricians.
Based on climate and geographical differences as well as differences in industry composition between specific regions, it is reasonable to expect that the frequency of occupational requirements for outdoor exposure would not be consistent for an occupation when comparing different state and metropolitan areas. For example, a cashier in Florida may be required to spend more time outdoors than a cashier in a northern state.
Union membership rates vary widely across the United States. This will likely correlate to differences in the requirements placed on workers, especially when looking at localities with large differences in union participation.
Examples provided are based on estimates for the 2021 reference period for both ORS and OEWS.
Note that for all combined ORS-OEWS estimates, the reliability of the combined estimate will necessarily be lower than the reliability of each individual estimate. For more information about each survey’s design, see the ORS Handbook of Methods and the OEWS overview.
Except for additive requirements, job requirements are collected separately. The actual amount of overlap between distinct requirements is unknown. For example, all jobs with requirement A could also have requirement B, or no jobs with requirement A may also have requirement B. The lower bound assumes the percentage of jobs with a given requirement is not affected by whether jobs have other requirements. That is, if 30 percent of jobs have requirement A, that 30 percent applies regardless of jobs having requirement B. The upper bound is the estimated number of workers with the least prevalent requirement also have the more prevalent ones. There are a variety of assumptions included in this approach and requirements could even be negatively correlated, e.g., in the extreme case, no jobs with requirement A also have requirement B.
The pooled standard error approach assumes that the ORS and OEWS estimates are independent.
A more rigorous approach, beyond the scope of this factsheet, is to use parametric bootstrapping to estimate the variance of a combined ORS-OEWS estimates. Replicate samples can be generated for job requirements estimates by drawing random estimates from the normal distribution defined by the estimate and its standard error. The same can be done for a corresponding OEWS employment estimate. For each replicate, the randomly sampled ORS estimate and the randomly sampled OEWS estimate can be multiplied to produce a combined ORS-OEWS estimate. The combined estimates for each replicate can in turn be used to produce a variance estimate for the combined ORS-OEWS estimate:
represents the ORS estimate multiplied by the OEWS estimate,
represents the number of bootstrap replicates, and
represents the product of the randomly sampled ORS estimate and the randomly sampled OEWS estimate for each replicate r.
The standard error for the estimate is the square root of the variance.
Other methods for calculating the reliability of estimates are included for reference, see A measurement error model approach to survey data integration: combining information from two surveys and Data Fusion: Identification Problems, Validity, and Multiple Imputation. The authors and funding for the research are not affiliated with the BLS.