The Availability Analysis is a critical element of an Affirmative Action Plan (AAP) since, by its very definition, it determines the “qualified labor pool” from which the contractor will be selecting to fill open positions. Those availability results, then, have a direct impact on the outreach efforts towards minorities and women the contractor puts forth, as well as the resources required to invest in such efforts. Given the potential impact of the availability analysis, due diligence should be exercised when determining what external availability source data will be used for the AAP.
This article will examine the external data component of the availability analysis, the implicit biases that may taint the various sources of data, and whether a single, standardized methodology would yield reasonable results for contractors.
Primer: The Availability Analysis in an AAP is designed to “establish a benchmark against which the demographic composition of the contractor’s incumbent workforce can be compared in order to determine whether barriers to equal employment opportunity may exist within particular job groups” (41 CFR 60-2.14(a)). In lay terms, availability represents the percentages of minorities and women who are “available” to work in the various job group categories at the contractor’s establishment. A key part of availability is the external availability factor which represents the contractor’s external hiring to fill positions.
It is important for AAP practitioners to understand the requirements of 41 CFR 60–2.14(d): contractors “must use the most current and discrete statistical information available to derive availability figures.” In 2012, the United States Census Bureau released the 2006–2010 American Community Survey EEO Tabulation (a.k.a. 5-year ACS data). It is the most recent data available of the workforce, represented by 488 occupations, and “serves as the primary external benchmark for comparing the race, ethnicity, and sex composition of an organization’s internal workforce, and the analogous external labor market, within a specified geography and job category.” The ACS data are available for a wide variety of geographic areas (e.g., U.S., States, Counties, Core-Based Statistical Areas).
ACS Data: One main advantage in using the ACS data is that it is the most commonly used data by AAP practitioners and widely accepted by the Office of Federal Contract Compliance Programs (OFCCP). These data allow for customization by geographic area and also allow practitioners to select the occupation codes that are most relevant to the positions defined in the AAP. For example, if you employ Construction Managers, Registered Nurses, or Software Developers, there are specific occupation codes for each of these jobs. This level of customization can help contractors create a reasonably accurate model of their qualified external labor pool.
There are, however, some inherent limitations in using these data. For instance, over time they may seem outdated since new census data are only released every 10+ years. The ACS data were originally supposed to be released more frequently at every five years, but it is still the 2006-2010 ACS data that are being used today.
Another potential drawback with the ACS data is that many census occupations are still very broad and can unintentionally skew the targets for minorities and women. For example, the occupation of 0050 Marketing and Sales Managers includes managers of selling children’s clothes and construction equipment. So one has to wonder if that census data would create targets for minorities or women that are too high (or low) for a retail clothing store or a tractor company.
An equally important issue with the ACS data that is easily overlooked is whether or not the data itself can be “tainted.” The occupational census data are supposed to represent who is in the U.S. workforce. But if there are systemic or institutional barriers preventing certain groups from entering or staying in the workforce – or steering them into certain types of jobs – would the occupational census data accurately reflect who is qualified and available to work in a given job?
For example, if there are societal, cultural, or other barriers that have, historically, limited women from getting science, technology, engineering, and mathematics (STEM) jobs, would the 2010 census occupations for Mathematicians or Engineers be a good measure of how many female mathematicians and engineers there should be today?
It is important, then, to consider the historic case Hazelwood School District v. United States (U.S. Supreme Court 1977) when trying to reconcile these issues of tainted data. This case examined employment discrimination against African American teachers in the Hazelwood School District, a suburb in St. Louis County, Missouri. A key issue in the decision was if there was “a proper comparison (…) between the racial composition of Hazelwood’s teaching staff and the racial composition of the qualified public school teacher population in the relevant labor market” (emphasis added).
In the District Court’s finding of no pattern or practice of discrimination, it was found to be in error for comparing Hazelwood’s 1.8% African American teacher workforce in 1973 to its student population of approximately 2% African American. In the Appellate Court, the Government’s finding of discrimination was found to be in error for selecting St. Louis County and St. Louis City as the relevant labor market (15.4% African American teachers in 1970 census) and disregarding other possible relevant statistics.
The key point here is that the Supreme Court did not make a judgment on what the relevant labor market should have been, only that consideration should have been given to all available statistics. Namely, that 3.7% African American teachers were hired; the representation of African American teachers in St. Louis County and St. Louis city was 15.4%; and the representation of African American teachers in St. Louis County alone was 5.7%.
The Hazelwood case does not make the use of census data inappropriate simply because there is a chance that it may be tainted or used incorrectly. What the case does remind practitioners of, however, is to carefully consider all available and relevant data for use in the availability analysis, and utilize the most appropriate source – or sources – as needed.
Beyond the ACS data are other external sources that may prove beneficial when used under the right circumstances:
The Civilian Labor Force (CLF): CLF data is published by the U.S. Bureau of Labor Statistics (BLS) and represents the subset of Americans who are employed and unemployed, are at least 16 years old, are not serving in the military, and are not institutionalized. An advantage over using ACS data is that the CLF data are available on a monthly basis, far more frequently than decennial census data. Additionally, since the CLF includes both employed and unemployed individuals who are “available” to work (i.e., 16 years or older, non-institutionalized), it is less likely to be tainted by socioeconomic or other factors. The CLF data is very broad, which makes it a reasonable availability comparison for unskilled labor occupations.
While CLF data can work well for unskilled occupations, it is less than desirable for companies with highly skilled workers to use CLF data. After all, it may not be altogether accurate to use data that includes unemployed 16-year-olds for positions requiring a Master’s degree and five years of work experience. Another potential drawback to the CLF data is that there are limited data available for American Indians and Alaska Natives, Native Hawaiians and Other Pacific Islanders, and people who are of Two or More Races due to the relatively small sample sizes of those groups. So any AAPs where individual minority group data are sought would probably preclude the use of CLF data.
Industry-Specific Data: ACS Industry-specific data is based on the same EEO tabulation as described above and can be gathered by census occupation and geographic area. What makes this source different, however, is that the occupational census data can be gathered by type of industry (e.g., Manufacturing, Retail Trade, Wholesale Trade, Administrative, etc.) which can allow for further customization in the availability analysis. For example, if the job requirements for Production Managers in the Automotive Manufacturing Industry differ from the Computer/Electronic Manufacturing industry – and there are marked demographic representations between the two industries – this type of source data may help to legitimately account for those differences. And these industry-specific figures could provide more relevant and accurate availability data since workers from non-relevant industries could be excluded from the contractor’s occupational census data.
At the same time, it is important to be aware that industry data may be too restrictive and not applicable for all jobs in a given company. For example, does a software company only hire Sales Managers who have worked at other software companies, or those who have worked as a manager in any industry? If the answer is the latter, then using industry-specific data for the sales positions may not reflect reality. And because the industry data can be quite restrictive, there may be some occupations, geographies, or industries that do not yield viable data for analysis purposes, because the sample sizes are too small. And finally, since this source is similar to ACS data, the same potential issues of tainted data apply as well.
Educational Attainment Data: Another type of widely used data is that of educational attainment. And for this type, perhaps the most commonly used sources are the Non-Partisan and Objective Research (NORC) organization at the University of Chicago and the National Center for Education Statistics (NCES). A third source derived from surveys conducted by NCES is the Integrated Postsecondary Education Data System (IPEDS). These data are often utilized in availability for positions where particular levels of educational attainment are a major job requirement, such as for faculty job groups in colleges and universities.
The ACS census occupation data for “Postsecondary Teachers” would represent professors as well as instructors at colleges, junior colleges, trade schools, and other educational institutions. As a result, using that source to represent professors at four-year universities, for example, may be significantly too broad and not accurately represent all professors who are available in the U.S. workforce. Making use of NORC or NCES data, on the other hand, could provide a more realistic representation for professors. Not only could the level of educational attainment be controlled (e.g., PhD) but so too can the fields of study (e.g., Humanities, Engineering, Life Sciences, etc.). An added benefit to using these sources is that they are released more frequently than decennial census data.
The issue of potential tainted data, however, does affect this type of data also. In much the same way that certain groups can be seen to have been historically disadvantaged from entering the workforce or certain occupations, similar barriers can be seen in the educational space. Having certain groups possibly predisposed to having a lower rate of postsecondary educational attainment may call into question the calculation of the “qualified” labor pool based on these data, so great care should be taken with their use. For example, strictly using NORC data on PhD attainment for professor faculty positions could be appropriate since that degree level is a common job requirement. On the other hand, using the same source data for an engineering position that requires a PhD or equivalent may be questionable since it may not represent those in the workforce with sufficient work experience to meet the job requirements.
Educational Attainment/Industry Specific: Similar to education-related data from NORC and NCES, data for medical and dental fields may be found through the Association of American Medical Colleges (AAMC) and the American Dental Education Association (ADEA). AAMC offers data on medical school enrollees and graduates, as well as faculty by degree type, and by department. Dental school enrollees and graduates, in addition to data on faculty for dentists and allied health may be found through ADEA. As such, these sources may be used on faculty job groups that cover medical or dental schools, and the data would reflect actual faculty representation as reported by U.S. medical and dental schools annually.
Much like the NORC or NCES data, there are some limitations to the AAMC and ADEA data. It is still possible for tainted data to be an issue, especially since there are fewer opportunities when acceptance into a medical or dental school program can be even more difficult than getting an undergraduate education.
With so many factors to weigh when selecting which external data sources to use, consider the following to help you attain reasonable – and defensible – availability results:
Of course, if you are uncertain about which data is the right data for your AAP, connect yourself with a consultant. If you don’t have a consultant, feel free to reach out to Biddle at email@example.com.