The process of comparing incumbency to availability, typically referred to as the "utilization analysis," and setting placement rate "goals" is a required component of your federal affirmative action program, as outlined in the OFCCP's regulations (external link). Separately, the Uniform Guidelines on Employee Selection Procedures (external link) requires qualifying federal contractors to monitor their "employment selection decisions," such as hiring, through statistical analyses called Impact Ratio Analyses, or IRAs.

Both are designed to identify where the organization might have a potential discrimination issue. Neither is definitive proof of discrimination, though a "hot" IRA result is often considered to be strong evidence of foul play. But the interplay of these two analyses is important to understand.

The OFCCP's requirement to set "placement goals" can be confusing. In short, federal contractors are required to compare the demographics of their current workforce to the demographics of a theoretical workforce that they could have based on the relevant labor market(s). When looking at those demographics by sex and race/ethnicity, contractors are required to set a "goal" where the difference is "more than would be reasonably expected."

This "goal," however, is not a goal at all in the way most people understand and use that term. It is not something recruiters and hiring managers should consider a "target" that they actively try to hit in the course of performing their duties. Their job is to follow the resulting action plan, often a recruiting plan, and hire the best qualified candidates that come through the hiring mechanism.

The "goal" is actually a benchmark for diagnosing how well that hiring mechanism is working and determining whether and to what extent adjustments need to be made to ensure no one is running into any "artificial barriers" in the process based on their legally protected characteristics. For recruiters and hiring managers, what the goal is--whether it is 30% or 56%--is irrelevant. The fact that there is a goal is what's important. More specifically, the fact that there is an "action plan" to follow is really all they need to know. The numeric "goal" itself only leads to confusion and runs the very real risk that it will be interpreted as a quota and/or be treated as a ceiling for the affected group.

But the fact that an organization has set a placement goal is not in and of itself an indication that illegal discrimination is occurring. It is simply a way to survey a vast employment landscape and zero in on areas where the demographic imbalances are enough that they could indicate that there is an issue. Contractors must then investigate to try to determine the cause of the imbalance and, if an issue is uncovered, correct it.

Often, however, the reason for the demographic imbalance is elusive, so the resulting action plans are typically an exercise in the process of elimination--adjusting recruiting practices, partnering with targeted outreach and recruiting organizations, providing additional training, etc.--to either find the cause or determine that whatever is driving the analysis results is not the organization's fault.

The Uniform Guidelines apply to any employment selection decision process, but most OFCCP audits will focus almost exclusively on hiring activity. Similar to the placement goal process, IRAs are a way to examine the results of your overall hiring process to determine whether or not there are areas where systemic discrimination could be occurring and focusing limited resources on investigations in those areas, rather than trying to tackle the whole beast.

The process is fairly straightforward. Based on the applicant "pool" for a particular employment selection decision, contractors must determine the percentage selection rates of various groups based on their sex and race/ethnicity. Typically contractors will compare the selection rates of men and women, and the selection rates of Whites* and People of Color (POC) in the aggregate.
Traditionally, the OFCCP would only drill-down and perform race-/ethnicity-specific IRAs if the selection rates between Whites and POC is significant, though in recent years the agency has often performed these specific analyses regardless, and BCGi encourages all members to do the same as part of their annual AAP preparation process. In these analyses, contractors identify the race/ethnicity group with the highest selection rate and then compares the rate for that group to the selection rate of every other race/ethnicity group that comprises at least 2% of the overall analysis population.

The difference in selection rates (and there will almost always be a difference among the various groups) is considered "significant" based on one of several statistical tests. The Uniform Guidelines, written long before everyone was walking around with a supercomputer in their pocket, calls for the use of the "80% rule" where the difference is significant if the selection rate for one group is less than 80% of the selection rate of the other.

Since those regulations were written, computer technology has evolved to allow federal contractors to more easily apply a more stringent statistical test using common applications such as Microsoft Excel. The OFCCP will accept the results of more stringent statistical analyses, despite the clear directive in the Uniform Guidelines to use the "80% rule," because courts have recognized their superior utility, and the agency will almost never pursue an allegation of discrimination based on the Uniform Guidelines test.
The gold standard for many years was a standard deviation test where a result grater than 1.97 is generally considered to be "significant" because beyond that point there is a less than 5% chance that the difference being examined is due to random chance. This statistical analysis, however, is only appropriate where the "pool" consists of at least 30 people and there are at least 5 expected selections.

For smaller analysis groups, a standard deviation test results become unreliable, so contractors typically switch to an exact binomial test. This test is just as accurate as a standard deviation, even for small groups. It is actually appropriate and accurate for any size analysis group, but it is processor- and memory-intensive. So in years past, use of an exact binomial was reserved for only those instances where a standard deviation test is inappropriate. These days, however, everyone has more than enough computing power at their disposal to perform an exact binomial test in every instance, which is exactly what BCGi recommends.

*Note that BCGi capitalizes all "official" race/ethnicity categories, whether or not the word in question is typically considered to be a proper noun. This is in line with longstanding EEOC practice and does not imply relative importance.

So how to placement goals and IRAs interact? As noted above, the placement goal process simply identifies areas in the organization where the demographics of the workforce appears to be suspiciously "off." The reason for this, however, is hardly ever clear. But one reason might be discrimination in the selection process, which is where IRAs come in.

If an organization is required to set a placement goal for women in a particular AAP job group, for example, and the hiring IRAs for that same job group indicate a statistically significant difference in selection rates between men and women (to the detriment of women), that is a pretty good clue as to why the demographics seem off, so those IRA results should be investigated first. If that investigation indicates that the selection process is in fact having a discriminatory effect, the organization will want to correct that first and consult with legal counsel about any further action regarding people who might have been harmed.

If the organization determines that the selection process is not discriminatory, they are back to developing an action plan that tries to identify the cause through the process of elimination, most often starting with recruiting practices.

But note that contractors should be analyzing all facets of their hiring process with IRAs regardless of whether or not there is also a placement goal for the corresponding AAP job group(s). These are separate requirements on their own, so while the use of the analysis results can overlap, they are not actually tied to one another.

The other way the two work together, though, is often referred to as "triage." BCGi recommends investigating all "hot" IRA results, and most organizations understand the importance there. But when it comes to placement goals, the threat is not so immediate and often an organization has too many placement goals in too many AAP job groups across too many AAPs for each one to receive the attention it deserves. Where there is an AAP job group with a placement goal and "hot" IRA results, those can be moved to the front of the line. Or, alternatively, they can be set aside while the organization focuses on the IRA results that are potentially most likely driving the placement goals and focus on action plans for the remaining AAP job groups where the IRA results are not the obvious potential answer.