Table 3.
A summary of the percent reduction in re-identification risk for each subject when shifting from the worst case to the Twitter and voter registration scenarios
| External Identified Dataset |
Number of Subjects
(×1000) |
Risk Reduction |
||||
|---|---|---|---|---|---|---|
| Journalist |
Prosecutor |
Marketer |
||||
| Q1 | Q1 | Q1 | Median | Q3 | ||
| 1 | 100% | 100% | 99.48% | 99.70% | 99.81% | |
| 4 | 100% | 100% | 99.82% | 99.92% | 99.95% | |
| 7 | 100% | 100% | 99.88% | 99.95% | 99.97% | |
| 20 | 100% | 100% | 99.94% | 99.98% | 99.99% | |
| 40 | 100% | 100% | 99.95% | 99.98% | 99.99% | |
| Voter Registration | 1 | 100% | 100% | 67.47% | 86.78% | 93.86% |
| 4 | 100% | 100% | 88.16% | 94.91% | 97.87% | |
| 7 | 100% | 100% | 92.22% | 96.86% | 98.67% | |
| 20 | 100% | 100% | 95.83% | 98.38% | 99.32% | |
| 40 | 100% | 100% | 96.63% | 98.95% | 99.57% | |
Q: quartile.