My sister says this disease has a name.
But I'm always in a hurry when my project requires blue beads, I need a bunch of them, and I want them now. I have no time on a cool morning to paw through 18 hangers to find the Eileen Fisher thin jersey. (My sister also says that owning 18 black tops is a disease…she's wrong!)
It's just a fact that if one has a lot of beads – or tens of thousands of applications, a crushing number of requisitions and resumes that your department heads forwarded to you because they don't have the gumption to tell their brothers-in-law or next-door neighbor's caddy that they have to apply like everyone else (they do, don't they?) – one must sort. The only question is: when.
It's pretty clear that it's easier to find blue beads when you need them if you have a box for just blue beads. It's clear that it's easier to find all the job seekers whose resumes were never read because there were no openings in the jobs they sought, if you've put just those in a "box" separate from the box of those applicants who were rejected on the basis of their comparative experience for Requisition 1805-Eng178.
We all have to sort. The problem with sorting selection data under pressure is twofold: (1) you will make mistakes and (2) obviously, you won't have made use of this data to manage your compliance, in the language of Executive Order 11246, as amended:
"...to take affirmative action to ensure that applicants are employed, and that employees are treated during employment, without regard to their race, color, religion, sex, sexual orientation, gender identity or national origin."
You may "know" that you are maintaining records and you may "know" that you've trained all recruiters and all selecting officials to disposition each and every job seeker in accordance with your organization's policies. But I'll bet you my secret recipe for Gravad Lax that there are plenty of nights you lose sleep over whether they are doing it and about what your data will show. Perhaps you have neither "ensurance" that the organization is not discriminating nor "insurance" to use in defense of a claim for back pay!
You will have guessed...I'm in favor of sorting in the first instance. This is easier than you think…it simply requires starting out with more "boxes" in which to put things after you've made a choice. And, unlike in the last century, today's "boxes" are virtual and once in the "box" we can easily retrieve exactly the box we want when we want it – even two years later (or twenty) when the OFCCP demands it. Perhaps it's worth a reminder: the whole point of keeping records is so that you can retrieve them later. The government makes you keep them so it can find them. You need to know what's in them first! Good, specific and customized Disposition Codes are "boxes" worth having!
This is how your organization can come to love disposition codes:
- Recognize that, it is in your organization's best interest to "put at risk" to a challenge of discrimination/demand for back pay only those potential claimants who were actually rejected by the employer and for the same reason. Only well-crafted disposition codes will permit you to retrieve accurate data needed for analysis and submission to the OFCCP.
- Recognize that, in addition to selection process policies – and direction to recruiters – your organization will need appropriate data management techniques and specific disposition codes that will eliminate such a superfluity of job seekers/applicants that your analyses will trigger the "Law of Big Numbers."
- Determine that you prefer (vastly) to defend only those selection criteria (including "tests" in the lexicon of The Uniform Guidelines on Employee Selection Procedures) that are doing that which you intend (and currently only hope and pray?) which is to truly select the "best qualified" person.
- Recognize that, if asked, many people who use disposition codes would rather just use two ("hired" and "not hired") but if given the facts they too can come to want to use codes where:
- They've had input into their design
- They are specific enough that they don't have to waste time agonizing over which one to use
- They know that their rigor in consistently applying the most specific code available helps the organization and that failure to do so hurts the organization and thus will impact their own performance appraisals.
- I provide some "Rules" and some "Tips" and some Exemplars for a variety of organizations and selection processes to help you on your way.
- Recognize that, it is in your organization's best interest to "put at risk" to a challenge of discrimination/demand for back pay only those potential claimants who were actually rejected by the employer and for the same reason.
Well-crafted disposition codes will not produce "false positives" in discrimination analyses.
This is not trickery nor an attempt to "hide" anything. Rather, specific disposition codes that are consistently applied permit both the enforcement agency and the contractor to focus on whether there is "adverse impact" and, if so, why. That is, properly designed disposition codes will enable the employer to identify the selection criterion that is causing a disproportionate percentage of X group to be rejected.
Disposition codes that describe with reasonable specificity the particular reason for the job seeker not moving forward in the selection process, nor ultimately being selected can be expected to save the non-discriminating contractor untold grief in defending poorly conducted analyses that mix not just apples and oranges but Apples and Androids.
In fact, there are two powerful incentives for a contractor to be able to retrieve very specific information/data on the impact of its selection criteria: 1) to perform authentic "adverse impact analyses" that actually have the capability of informing the contractor if any of its selection criterion need review or modification and 2) to prevent the OFCCP – and others – from pressing claims based on data that are not discrete. A certain motivation on the part of enforcement agencies and plaintiff's counsel has to do with "The Law of Big Numbers."
The Law of Big Numbers
This "law" is not actually a law or regulation nor is it within the scope of either this article or the writer's expertise. But sufficient for our purposes here, and very importantly, the more applicants and the more selections that are analyzed, the greater the likelihood that disparities in selection rates among and between groups will be statistically significant. If a so-called "adverse impact analysis" examines 2,000 applicants for 100 Technician I openings, it is much more likely that there will be a "disproportionate impact" on some group than if the analysis was – accurately – of 200 applicants and 5 selections for Job A and 132 applicants and 15 selections for Job B and 47 applicants and 1 selection for Job C, etc.
If, indeed, there were 2,000 applicants and 100 selections all for the same job and the single "test" of, for example, Bachelor's degree in Engineering, then we might actually know whether there was a statistically significant rejection of women or Native Americans or Hispanics, etc. based on them not having that credential. If there was and the job was for Engineer, the employer could easily validate that requirement. If the job wasn't for Engineer, the employer probably could NOT rebut any presumption of discrimination and the OFCCP would hand the CEO the pen to write the check for back pay.
But if the applicants were for "particular positions" ranging from accountants to librarians to engineers to attorneys, with likely different mixes of minority and female applicants for each of those different positions, what will we have learned from the purported analysis?
Nothing.
So why does the agency do these gross analyses? Because employers allow them to do so and because employers often don't know why those 1,800 applicants weren't selected, much less why the 200 were! Unless they can sort their data into not only the "particular" job but the particular selection criterion they can perform no better analysis themselves.
If the employer has been so foolish as to "consider" (or even accept, for that matter) thousands and thousands of resumes just because it can, their analyses even by job title will likely trigger this "Law of Big Numbers."
The OFCCP's regulatory "loophole" permits contractors to eliminate from its adverse impact analyses persons who don't meet "basic qualifications" (which is not the same as "minimum qualifications" in an employer's parlance, I assure you!) At the same time, the OFCCP will insist that contractors "roll up" or aggregate data of applicants that were not tested by the same "test" nor even for the same "particular position" for which the contractor must have considered them.
Compliance Guidance: Contractors should always refuse to aggregate things that are not alike! You can't keep the OFCCP from doing it with your data – for the sole purpose of reaching their apparently desired result: an allegation of discrimination. You can, however, refuse to do it for them. Ellen Shong & Associates did that/does that routinely. When called upon to defend an aggregation of dissimilar events/persons/jobs (e.g. purported "adverse impact analyses") that was done by Job Group or by Department) one simply does so by dis-aggregating to – at most – the job title, and citing a little regulatory law to the agency.
Even that depends on how discrete the employer's job titles are. If Research Assistant can be a person in the library who collates research materials or a person doing stem cell research in a lab, then you have problems. You have problems in Job Group design, selection analyses and – most especially – in pay analysis. Fix this.
The agency wants "basic qualifications" tied to "the particular job" for which the employer "considered" the job seeker. It is even more important that discrimination analyses control for the same selection process (decision makers) or at least the same selection device.
Making reference to the specific requirements of The Uniform Guidelines on Employee Selection Procedures may be instructive. In pertinent part, they direct that [employers] "… maintain and have available for each job information on adverse impact of the selection process for that job" and that "Where a total selection process for a job has an adverse impact, the user should maintain and have available records or other information showing which components have an adverse impact. "41 CFR §60-3.15 (Emphasis mine. The Guidelines are, thereafter, almost completely occupied with the requirement of "validation studies" of such components having adverse impact. Over and over in these regulations the law directs that to be "validated" requires – in the simplest of terms – relatedness of the criterion to performance of "the job." 41 CFR §60-3.5.
Having knowledge of causation permits the contractor – even on its own initiative! – to modify the qualification/selection criterion that is disproportionately excluding, for example, Hispanics. It also permits the agency to challenge that causative criterion…specifically. Finally, it permits the employer to demonstrate that though the criterion has adverse impact, it is related to performance of the job.
Compliance Guidance: You should understand that if you use the same test for multiple jobs, you really should aggregate applicants across those jobs in order to determine whether that test has "adverse impact." (I have never, personally, heard of the agency demanding this, but it's the right thing to do and absolutely the corollary to the contractor's argument about not aggregating dissimilar criterion.) Caveat: One such test that is so commonly used it may very well trigger the "Law of Big Numbers" is "High School Diploma or GED." Employers seem to have decided that since we know fewer minorities graduate from high school than whites, all they have to do is provide the alternate qualification of "GED" to avoid a finding of adverse impact. First, I think that's highly doubtful if ever investigated. Second, how would you begin to show that such a qualification was "job related"? What does possession of this credential, in and of itself, predict about an individual's knowledge, skill or ability to perform any job? If you think it demonstrates sufficient reading and writing skills, you haven't seen as many applications as I have – including those with the name of the high school misspelled. If you think it proves the person can "stick to it", you don't know any 17-year-old drop outs who hold two jobs to help support their family. If you think you can demonstrate that a person can't do an unskilled job without this "qualification," you definitely don't remember the Supreme Court's first adverse impact decision, Griggs v. Duke Power – nearly 50 years ago. - Recognize that, in addition to selection process policies – and direction to recruiters – your organization will need appropriate data management techniques and specific disposition codes that will eliminate such a superfluity of job seekers/applicants that your analyses will trigger the "Law of Big Numbers."
Among other things, to avoid shooting yourself in the foot with bad big data, you need to distinguish – carefully – between persons who don't – or should not – progress through the selection process for reasons having nothing to do with your consideration of their qualifications for a particular position.
For example – and this ought to be a law! – don't "consider" any expression of interest unless the job seeker specifies interest in a particular job. The OFCCP regulation can't require this of a job seeker, but you certainly may do so.
Long before there was a definition of "Internet Applicant," it was always intended by discrimination laws that it was the employer's rejection that was potentially unlawful. Discrimination is the denial of an opportunity on the basis of race or sex or national origin, for example; that's actionable. No denial, no discrimination. No opportunity, no discrimination. Neither disparate treatment based on prohibited factors nor neutral selection criteria that disparately impacted women or minorities? No discrimination. There are all sorts of reasons why a job seeker might not be hired that have nothing whatsoever to do with the employer's assessment of his qualifications.
Many of those reasons have to do with job seekers themselves. Not their qualifications to perform the job but such things as their availability to do so; their willingness to do so for the shift you need them for and at the money you're willing to pay; whether the employer would break the law by hiring them (this ranges from minimum age to legal right to work in the U.S. to DOT or DOD or AEC rules. Or the job seeker can simply change his mind or perhaps decide she doesn't really like you. No job seeker in these situations – or many others – should ever be in any data or analyses that you submit to the OFCCP.
So, you need disposition codes for each of these and more.
Compliance Guidance: In addition, I strongly urge every organization to establish policies and protocols that strategically limit its exposure by limiting the number of applicants that it "considers" to that number which is truly sufficient for you to fill jobs with well qualified people.
In order to not trigger the "Law of Big Numbers," the employer must not only be able to retrieve data on specific jobs and the specific reasons for rejecting people, and resist the OFCCP's efforts to "roll up" different jobs with different selection criteria. It must also manage the volume of applications/resumes it will consider for jobs over the course of an entire year.
Pages and pages have been written on the definition of Internet Applicant and, especially, on the definition of "basic qualifications," so I won't include that full discussion here. However, I refer you to 41 CFR 60-1.3 (1) and (4) and also to the agency's website for FAQs.
However, one thing about the definition of Internet Applicant (or any "applicant" for that matter) is noteworthy. It is what the employer does that assigns the status of "applicant" to a job seeker: evaluate the job seeker's qualifications. If the employer never makes a substantive determination about Joe's qualifications for a particular position then Joe is NOT an applicant.
The definition of Internet Applicant (or any applicant) does not require a contractor or employer to evaluate the qualifications of everyone who expresses interest in employment. The definition simply says that if the contractor "considers" a person for a particular position – and the individual meets the other elements of the definition -- s/he is an "Internet Applicant" for which the contractor must solicit race/ethnicity and sex information. The regulation directs contractors who wish to limit the number of job seekers it will "consider" (and you do want to do this) to use appropriate "data management" techniques to do so. They boil down to: you can't "select" who you're going to "consider."
So, do use high tech data management techniques when using large electronic data bases; always meeting the OFCCP standards for data management.
But you may also use "low tech" data management techniques such as deadlines or "first in/first considered."
Be assured that it is completely lawful for the employer to put in place different protocols for different jobs depending entirely on its need for applicants. If it's very difficult to find a qualified person to fill a particular job, then by all means accept as many resumes as you can attract. Or, establish "interest registers" and encourage people with particular skills to provide their contact information so that you can actively invite them to apply when you do have an opening. Or keep that contact info yourself for the "second choice" applicant that wasn't hired. Or the one that withdrew just before you made a job offer because she accepted a job elsewhere. You can always reach out to such individuals and encourage them to apply when you have other openings while at the same time advertising such opportunities in your usual nondiscriminatory and affirmative way. - Determine that you prefer (vastly) to defend only those selection criteria (including "tests" in the lexicon of The Uniform Guidelines on Employee Selection Procedures) that are doing that which you intend (and currently only hope and pray?) which is to truly select the "best qualified person."
You know that if a selection criterion causes adverse impact and you can show that it is "job related" or that it is "validated" you can continue to use it and not be found to have discriminated against women or minorities. But did you also know that if you can't show that it's job related that selection criterion hasn't worked to select the best qualified white men? So why would you want to use it?
Well-designed disposition codes and properly conducted adverse impact analyses can be powerful management tools. Is it time to review/re-think some of the criteria you have been using for some/all of your jobs? If you're having trouble attracting applicants for the pay you offer or you're having difficulty retaining them, perhaps you are setting the bar higher than you need to. Not only do you risk artificially inflated qualification requirements adversely impacting minorities or women, you risk discouraging job seekers of all types who might be perfectly capable of doing your job and aware that they don't have the qualifications to earn more than you pay.
Don't put your organization in the position of having to defend an allegation of discrimination unless you're confident that your selection criteria are, in fact, predicting high quality performers.
Use your disposition codes – even those for "Basic Qualifications" – to analyze how many people – of all types – are excluded by them. If these are for jobs you routinely have to fill, why? Are the qualifications helping you pick the right people or are they excluding so many you have too few to choose from?
Use disposition codes strategically to capture information you can use to select "good fit" candidates. - Recognize that, if asked, many people who use disposition codes would rather just use two ("hired" and "not hired") but if given the facts they too can come to want to use codes where:
- They've had input into their design of
- They're specific enough that they don't have to waste time agonizing over which one to use
- They know that their rigor in consistently applying the most specific code available helps the organization and that failure to do so hurts the organization and thus will impact their own performance appraisals.
Certainly, there are business justifications for simpler/less discrete disposition codes and the consequent grosser analyses one can do with them. Those justifications are quickly presented and grasped 1) the cost of doing it right, and 2) the need for speed to fill vacancies. However, in my experience, these justifications are infrequently questioned, much less "problem solved." My intent is to help you to do both.
If how one keeps records matters, then I'm going to try to persuade you that to keep records by more specific "codes" is easier/cheaper/faster/safer/more defensible than it is to be able to only retrieve them from more generic "codes."
As a consultant, my least favorite disposition code – though decidedly most profitable -- was "not qualified." It typically took weeks of the client's time and hour$ of mine to disaggregate the data – and re-review hundreds and hundreds of applications. In many – many – cases it was the recruiting staff who was dragged into the conference room full of boxes to try to reconstruct history. Not even necessarily their own, but more likely that of recruiters who were no longer there.
Would not those recruiters – and all of us – been so much happier if the disposition code was "not qualified/no CDL"...and "not qualified/education" and "not qualified/experience"? I don't think you actually have to conduct a "mock audit" with your recruiters – although that's not necessarily a bad idea – to show them how painfully costly it is to defend an allegation of unlawful discrimination when the employer cannot get its arms around its own data.
Compliance Guidance: Consider a "mock audit" of just two or three jobs that you fill most often and, possibly, one or two jobs that are typically difficult to fill. Pull the data you currently have by disposition code and first audit how consistently the codes are applied. If there's a lot of inconsistency, does the audit team think it's because people chose "the closest" disposition code because there wasn't a better one or do you have a code that's so generic (I used to charge double if the client made use of the disposition code "Other"). What bodies will you find when the OFCCP tells you where to look?