Fig. 6.

Probability of each group being the best (y-axis) against the extent of inconsistency, $\left|{\overline{\omega }}_{\mathit{BC}}\right|$ (x-axis), within the obtained range for each data set. a when I² is 0% and the standard error is 1, b when I² is 40% and the standard error is 1, c when I² is 70% and the standard error is 1, d when I² is 0% and the standard error is 2, and e when I² is 40% and the standard error is 2, f when I² is 70% and the standard error is 2, g when I² is 0% and the standard error is $2\sqrt{2}$, h when I² is 40% and the standard error is $2\sqrt{2}$, and i when I² is 70% and the standard error is $2\sqrt{2}$. The black dotted, gray solid, and black solid lines indicate the probability corresponding to groups a, b, and c, respectively. A vertical line marks the point at which some lines cross, and the percentages in the dark gray and dim gray boxes represent the proportions of $\left|{\overline{\omega }}_{\mathit{BC}}\right|$ that resulted in a consistent ranking and an inconsistent ranking of treatments relative to the original ranking, respectively