Figure 7
The percentage of case studies for which any of the three protocols described in §2 are significantly better (more than 5% discrepancy) than any other protocol and of cases in which the choice of protocol makes no difference or at least two protocols perform similarly. |