Carnegie-Mellon researchers have published a report on statistical uncertainty in Census Bureau data | Direct Employers Association


WIR asked: “How do you handle external availability calculations in AAPs when you start to feel that they are NOT accurately reporting true external availability?

Judy: “I’ve never used the external candidate stream instead of census data. The flow of external candidates has MUCH more problems than census data and these problems are only getting worse with the high turnover of technical support staff. What I will do on occasion is use an internal candidate stream. Let’s say that there is a training program for machinists and that the catchment area is that of machine operators. Machine operators are 80% women, but very few of them would like to be machinists. The client keeps track of who applies internally for the intern program, and of course 90% of them are men. So yes – we’ll be using three years of internal data instead of using everyone in the power pool. The only time I use the external candidate stream is as a first line of defense when a client misses their target. If the target is to hire 20% minority engineers and they miss the target, we will talk about the broad recruitment that has been done and the fact that, despite our good faith efforts, only 10% of candidates belonged to minorities. Then we will promise to try harder next year.

“I have been using new census data since February 2021, when it arrived at the data warehouse in Excel. The new occupancy index will be published in 2028, and the next set of EEO tables will not be published until 2031. So it will not disappear or be replaced. I promised my new friends at the Census Bureau that I would give them a list of codes not to combine next time. These are great people who just didn’t understand how data is actually used for AAPs. They are smart and kind, answered all my questions and eager to get feedback on how to improve the tables. Whenever you access the portal, you can click on the “Was this helpful” button and give them your feedback. I encourage everyone to do so. In the meantime, 95% of the data is excellent and usable. So don’t throw the baby out with the bathwater. Find other codes, another source, or use the bottom of the MOE for the 5% unusable data. »

As an Affirmative Action practitioner, I don’t look at the margin of error unless the percentage of women or minorities seems, on the face of it, unreasonable for the catchment area I’m using. Then I look at the total number of people in the population. If it’s really small, I’ll look for a different code with more people. The larger the population, the better the data. If that doesn’t work, I’ll find another source or use the bottom of the margin of error. Fortunately, regulations do not require federal contractors or subcontractors to set goals for minority subgroups. I hope the OFCCP will not go down this path. Not only would this be problematic with EOMs since the numbers will be smaller for each race/ethnicity, but there is no data for two or more races in the EEO tables.

“Also, we’ve known all along that census data isn’t perfect. That’s why we get 20% off! We have four options for determining underutilization, the make any difference rule that I hope no one uses, the very popular 80% rule, the also very popular whole person rule, and the gap rule -2 types which is popular with large companies although it is too often used for small groups of jobs. (I never recommend 2 standard deviations to my clients because their headcounts are too small, and it makes my clients’ eyes roll in their heads. Also, I like to use one ruler for all my workgroups.) ”

Stan: “With the 2010 census data, we had situations where the census data showed no one in particular occupational categories in the reasonable catchment area for some clients, while at the same time the clients were employing people in these categories. If I remember correctly, in those cases we would add a percentage to reflect those people.

“I continued to prepare ‘consolidated’ PAAs long after the implementation of the functional PAAs, and I extended the ‘campus’ approach of the OFCCP beyond higher education institutions to other providers and service providers, so my attitude towards availability estimates is also “relaxed.” For most of our clients, goals were like corks floating in the ocean, disappearing a year into a trough, rising on a wave to become visible another year, then disappearing again.”

“And if you have a small establishment, with a job group of about ten managers, you set a target (or not) based on the entire population of the job group, but you don’t risk certainly not to turn the twelve in one year. Maybe one, maybe two. What is a minority goal of 12% compared to this number? Similarly, if you have a job group with a mix of three or more census occupational classifications, you perform the weighted availability calculation, but you are unlikely to hire in more than one of these classifications. You hire an accountant, but all those HR pros drive the numbers up (or down). »

“Companies like OutSolve (and Gaucher Associates) strive to prepare AAPs as efficiently as possible, in order to keep costs low. To the extent that researching and determining if, when, and where to address margins of error increases the time it takes to prepare a PAA, it means additional cost. »

“Since the OFCCP has thus far paid little or no attention to the construction of availability estimates, there is little pressure to change these estimates to a contractor’s advantage ( or to satisfy customers who want to have goals). And where the OFCCP focuses on an entrepreneur’s goals and efforts to achieve those goals through outreach and recruitment, it’s not so much about having a goal (say for a group of jobs consisting of airline pilots, given that the 2020 occupational classification includes both pilots and air hostesses), as being able to stay ahead of the OFCCP in terms of suitable recruitment sources. Contractors don’t want the agency to come up with sources they don’t already use, or have considered and rejected for good reason.

“We had an audit where the regional operations manager was so desperate to find fault, that he identified a group of service jobs containing a title of medical technician, which included perhaps six people out of a total of 200+ in the job group, and insisted that he more properly belongs in a technician job group. Only this particular title was described in the EEO-1 guide as a service position (like “’Sanitation Engineer’” for garbage collector). We also encountered compliance officers who were confused about how to construct a weighted average of census data. »

“The agency’s compliance officers are being whipped from above, and the request is about wage and hiring discrimination ‘scalping’, so that’s where the focus is. And that’s is where AAP preparers spend most of their time, trying to make sure these things don’t happen.

Michael: “There is no ‘true external availability’. When I have tentatively completed the availability analysis, I subject the contractor incumbents and proposed availability percentages to Fisher’s exact test. If the differences are not statistically significant and I don’t have to declare a goal, I don’t adjust anything.

“But if some are statistically significant, I go back to all the assumptions I made and wonder if the census data is responsible for a false positive. This includes whether I chose best matches or whether I should consider other census titles that might better estimate minority and female availability.

“I look at the hiring and applying percentages for the most recent year, and multiple years, if I have the data. If the contractor is doing a good job of recruiting minorities and women in other groups of jobs, I’m inclined to consider merging the hiring or applying percentages, whichever is greater, into the census percentage for the outlier job group.

“I am aware of other considerations. The OFCCP doesn’t do a lot of audits, so the chances of any of my clients’ PAAs being audited are slim. Stating a goal is not a red flag during an audit, but it does require good faith efforts over the year to achieve it. If the entrepreneur is doing “everything” they can think of to hire more minorities or women in the job group and after several years the minority or women hiring percentages do not change, it may be time to re-evaluate if availability and goal percent are just too high and the census may give us a false positive.

Following Judy and Stan’s comments about pilots and flight attendants, Michael made the following observations:

“In the 2000 EEO reports, 7.1% of airline pilots in the United States were minorities and 4.0% were women. For flight attendants, 27.2% were minorities and 78.7 % were women.

In the 2006-2010 ACS reports, 9.4% of airline pilots in the United States were minorities and 4.7% were women. For air hostesses, 27.8% belonged to minorities and 79.6% were women.

In the 2014-2018 ACS reports, airline pilots and flight attendants were grouped under the same census title: “Air Transport Workers”, with 22.8% minority and 34.1% women.”

“Airlines most likely keep their pilots in a separate job group from their flight attendants. The ACS gives airlines only one set of data to use for both job groups. If I was preparing an AAP for an airline, I would not use the ACS EEO reports for either group.

Previous Why Ukraine's Donbass Region Matters to Putin: QuickTake
Next Amtrak cancels some long-distance trips as threat of freight strike looms