Summary of Statistical Analysis Results
| Project | Spearman ρ | P > |ρ| | Slope | tratio | P > |t| |
| A | 0.9646 | <0.0001 | 0.818 | 75.1 | <0.0001 |
| B | 0.9890 | <0.0001 | 0.874 | 98.2 | <0.0001 |
| C | 0.9846 | <0.0001 | 0.766 | 71.6 | <0.0001 |
| D[i] | 0.8692 | <0.0001 | 0.855 | 68.3 | <0.0001 |
| E | 0.9956 | <0.0001 | 0.884 | 144.3 | <0.0001 |
| F | 0.9968 | <0.0001 | 0.865 | 151.6 | <0.0001 |
| All | 0.9964 | <0.0001 | 0.845 | 174.5 | <0.0001 |
[i] In project D, the Spearman correlation coefficient ρ was artificially low as only very few bases (10) bases had a quality score of 5, and none of these bases contained an actual error (expected: 3.16 errors). Exclusion of this quality score gave a Spearman correlation coefficient of 0.9786 (P < 0.0001). The frequencies in the slope calculations were weighed by the number of bases at any given quality score and, thus, were not sensitive to such small sample distortions (see Methods).