A Way to Detect Bias
2 days ago
- Bias in selection can be detected without knowing the applicant pool, based solely on the performance of selected candidates.
- The technique requires a random sample of selected applicants, subsequent performance measurement, and roughly equal ability distribution among groups.
- Bias means that disadvantaged applicants must be more qualified to be selected, so they will outperform others if performance is measured validly.
- This approach was demonstrated by First Round Capital, which found that startups with female founders outperformed those without by 63%.
- The technique is rarely used, but increasing availability of data on selected candidates may lead to more such analyses in the future.