Have you ever heard of Simpson’s Paradox? It’s a statistical anomaly that may create an inference of discrimination where no discrimination exists. It’s almost always found in large samples and has everything to do with disposition codes according to Douglas B. Brown of Douglas B. Brown & Associates, LLC.
During his session at the DirectEmployers 2015 Annual Meeting & Conference, Brown explained that the OFCCP is aware of Simpson’s Paradox and counts on it to produce a statistically significant outcome against at least one group when dealing with large applicant pools. If a contractor is unable to manage, or analyze its pools, it may be stuck with the OFCCP’s analysis and result in adverse outcomes (inferences of systemic discrimination).
If you’re short on time, check out this brief interview with Brown. Or better yet, watch the video and keep reading!
Currently, most contractors attempt to comply with the OFCCP regulations by mandating the maintenance of records identifying the name, race, sex, application date, position applied for and disposition of each application. This includes the use of disposition codes that identify why the applicant didn’t advance in the selection process. If you’re not familiar with disposition codes, they can be a short note on a manual flow log, or selected from a list developed by the contractors, or chosen from pre-populated options in the ATS software. When properly applied, they should enable contractors to:
- Know when applicants exited the pool.
- Know why applicants exited the pool.
- Analyze their selection process to determine whether there is adverse impact overall as well as at any discrete step in the process.
- Control the size of the applicant pool.
Problems with Disposition Codes
Disposition codes don’t reveal when an applicant fell out of the selection process nor accurately describe why an applicant fell out of the selection process. Even if properly in place, individuals assigning codes in the ATS don’t always properly apply them.
A potential way to address these three issues is by conducting a “step” analysis. Brown recommended that employers should be just as interested as the OFCCP in being able to analyze the results of the various steps in the selection process. Most applicant tracking systems don’t provide adequate options to allow for the identification of when and why an applicant fell out of the selection process. It is for this reason that most organizations need additional disposition codes. In addition, organizations also need to ensure their HR and operations personnel who are responsible for applying codes are doing so properly.
Recommendations for Disposition Codes
The majority of work for codes needs to occur between when the individual becomes an applicant and the time an actual “hire” transpires. Contractors must have the ability to differentiate “when” a selection decision occurs (under the assumption these individuals meet the basic qualifications):
- Reviewed, not phone screened
- Phone screened, not referred
- Referred, not interviewed
- Interviewed, not offered
Another significant practice Brown emphasized is that contractors need to be able to differentiate “why” the applicant does not progress in the selection process, such as:
- Experience – not as much
- Experience – too much
- Skill set – type of experience
- Competency – communications, fit, teamwork – (specify in notes)
- Licensure – certification
- Assessment results
The new codes effectively do away with “other candidate better qualified” and “other candidate better fit.”
Issue With New Codes
Humans are creatures of habit, and as Brown pointed out, HR and operations personnel tend to migrate to a handful of codes they’re most comfortable with; they don’t apply the requisite rigor to make an accurate choice. New codes will only work if the people inputting the codes make accurate choices, much like the old IT adage of “garbage in, garbage out.”
Bottom line is that while new codes mean more to learn and additional work, there’s no question of the value. Implementing the additional codes allows for:
- More precise identification of when and why applicants fall out of the selection process if there is an indicator of adverse impact.
- Earlier and more efficient problem identification and resolution.
- Less probability of a detailed file dive.
- Less costly/disruptive responses to OFCCP information requests.
Special thanks to Douglas B. Brown for providing his expertise on this topic. Members can view his full presentation in our online community.