Latent Defect Estimation
Uses the Lincoln-Petersen capture-recapture method to estimate how many defects remain undiscovered. Two independent groups each review/test the same product — overlapping finds let us estimate the total population of defects.
| Defect Name | Severity | Group A | Group B | |
|---|---|---|---|---|
10
Estimated Total Defects
3
Estimated Undiscovered
5
Total Group A
4
Total Group B
3
Only Group A
2
Only Group B
2
Found by Both
7
Total Found
Defect Breakdown
| Defect | Severity | Group A | Group B | Category |
|---|---|---|---|---|
| Defect 1 | — | ✓ | — | Only A |
| Defect 2 | — | ✓ | — | Only A |
| Defect 3 | — | — | ✓ | Only B |
| Defect 4 | — | ✓ | ✓ | Both |
| Defect 5 | — | ✓ | — | Only A |
| Defect 6 | — | ✓ | ✓ | Both |
| Defect 7 | — | — | ✓ | Only B |
Formula: Estimated Total = ⌈(Group A × Group B) / Both⌉ = ⌈(5 × 4) / 2⌉ = 10
How It Works
Capture-Recapture
Originally used to estimate wildlife populations, the Lincoln-Petersen method works by having two independent groups "capture" defects. The overlap tells us about the total population.
Independent Testing
Both groups must test independently and thoroughly. Ask each group: "Have you tested thoroughly enough that you feel the major majority of issues are found?" The estimate improves with thoroughness.
The Formula
Estimated Total = ⌈(Total A × Total B) / Both⌉. If the groups find few overlapping defects, the estimate rises sharply — lots of territory remains unexplored.
Questions to Ask Both Groups
- Have you tested thoroughly enough that you feel all or the major majority of issues are found?
- How surprised would you be if another group found double the issues you did?
- How surprised would you be if this analysis found there are only a few latent defects?