Definition
Selection Rate is the proportion (percentage) of candidates selected or hired out of the total applicants.
- It’s most commonly used in HR, recruiting, and fairness analysis.
- In machine learning, it can also refer to the rate at which a model “selects” a group as positive.
Formula
$\text{Selection Rate} = \frac{\text{Number of Candidates Selected}}{\text{Total Number of Applicants}} \times 100\%$
Examples
1. HR / Hiring
- 200 people apply for a job.
- 20 are hired.
- Selection Rate = 20 ÷ 200 = 10%.
This shows how competitive the hiring process is.
2. Fairness in AI / Adverse Impact Analysis
Suppose you’re checking if a hiring algorithm is fair between two groups:
- Group A: 100 applicants, 30 selected → 30% selection rate.
- Group B: 100 applicants, 15 selected → 15% selection rate.
Under the U.S. EEOC’s “four-fifths (80%) rule”, the selection rate for Group B must be at least 80% of Group A’s rate (i.e., ≥ 24%).
Since 15% < 24%, there’s potential adverse impact against Group B.
3. Machine Learning Context
If a model predicts who gets a loan:
- Out of 1,000 applicants, 250 are approved.
- Selection Rate = 250 ÷ 1,000 = 25%.
This is important in fairness audits to check if approval rates differ too much across demographic groups.
Key Points
- Selection Rate = percentage of people/items chosen.
- Used in hiring, admissions, credit approval, or ML fairness checks.
- In fairness law, compared across groups (adverse impact analysis).