? What if Zahra flipped a coin?
The Dilemma: Options 1–7
You have seen the pattern: every criterion is identity-neutral on its face, yet every one produces measurable adverse impact — because the workforce structure it operates on was never neutral.
The Alternative: Option 8 — Random Selection
Option 8 is the simplest possible rule: select employees for layoff at random. No criterion. No algorithm. No manager discretion. Just a random number generator and a financial target.
In expectation, random selection produces a layoff pool that is demographically identical to the workforce. Every group's share of the downsized pool equals its share of the total workforce. Adverse impact: zero across every dimension.
It ignores merit and strategic value. A random process treats a top engineer and a redundant role as interchangeable. No board would approve a method that discards human capital indiscriminately.
It feels arbitrary and unjust. Employees would experience a lottery of livelihoods — being selected not because of anything they did or failed to do, but because a number generator chose them. The psychological harm of that arbitrariness may be worse than a reason you disagree with.
It destroys incentive alignment. If performance, loyalty, and skill don't protect you from layoff, why invest in any of them? Random selection signals that the organization doesn't value what it claims to value.
It damages recruitment and reputation. "We laid off 2,000 people at random" is not a sentence any CEO wants in the press release. Future talent will think twice.
These objections are real. Random selection is not a recommendation — it is a benchmark. If your preferred criterion produces measurable, disproportionate harm that a coin flip avoids entirely, the burden is on you to explain why that harm is justified by the business value your criterion preserves.
This Page Compares All Eight Options on Three Dimensions
① Headcount — how many employees lose their jobs to reach the $200M target
② Financial efficiency — savings per employee separated (cost per head)
③ Total disparity — the cumulative demographic distortion inflicted across all identity groups
You have rejected the only approach that distributes layoffs evenly. What is your business rationale for the criterion you chose instead — and does that rationale justify the specific, disproportionate harm it inflicts on identifiable groups of employees? If it does not rest on a defensible business or moral foundation, it is not a strategy. It is a preference with casualties.
All Eight Options at a Glance
| # | Option | Employees Cut | Savings | $/Employee | Max Single-Group Harm | Total Disparity Load |
|---|
Total Disparity Load = the sum of the absolute value of adverse impact (in percentage points) across all seven demographic groups (White employees, Black employees, Hispanic employees, Asian employees, Other employees, Women, and Men). A higher number means the pain of layoffs is distributed more unevenly. Random selection scores 0.0 because, in expectation, no group is over- or under-represented. This measure indicates magnitude, not legality — every option shown here, including the ones with the highest totals, clears the EEOC's 4/5ths rule (see the main simulator for that test). This page asks how unevenly the harm is spread, not whether it is lawful. For the distinction between statistical and practical significance, see the optional module in the main simulator.
How Unevenly Is the Pain Distributed?
Each bar = sum of |adverse impact| across all identity groups. Taller bars = more uneven distribution. Option 8 is the zero baseline.
Maximum Single-Group Harm
The single largest adverse impact on any one group per option. The label identifies who bears the most disproportionate harm.
Adverse Impact by Racialized Identity — All 8 Options
Adverse Impact by Gender — All 8 Options
Headcount and Cost per Separation
Each point = one option. X-axis = employees cut. Y-axis = total disparity load. Ideal position: bottom-left. Option 8 (diamond) sits at the bottom with zero disparity.
Can You Beat the Coin?
Option 3 (LIFO) comes closest to random — its total disparity load is 11.3, the lowest of any criterion-based option. But even LIFO produces +2.9 pp adverse impact on women. Every other option inflicts dramatically more demographic distortion, ranging from 17.4 (AIOE) to 68.3 (Legacy Manufacturing).
Random selection is not practical — organizations have legitimate reasons to prefer strategic criteria. But legitimacy requires justification. If your criterion concentrates +14.7 pp of adverse impact on Hispanic employees (Option 5) when a coin flip produces 0.0, you owe those employees — and a judge — a business or moral rationale for why that concentration was necessary.
The coin doesn't care who you are. That's the point. And that's the problem.
← Back to the Adverse Impact Simulator | Design Your Own Plan →