Table 4.

Pairwise TOST, where we test the null-hypothesis that for the target overlap score for each row-method, t_row, and the target overlap score for each column-method, t_column, $\frac{t_{\text{row}}}{t_{\text{column}}}<0.98$, or $\frac{t_{\text{row}}}{t_{\text{column}}}>1.02$.

Dataset: LPBA40
	ART	SyN	LO	LP	LPC
LO			N/A	✓	✓
LP			✓	N/A	✓
LPC			✓	✓	N/A
Dataset: CUMC12
	ART	SyN	LO	LP	LPC
LO			N/A		✓
LP		✓		N/A
LPC			✓		N/A

Dataset: IBSR18
	ART	SyN	LO	LP	LPC
LO		✓	N/A		✓
LP	✓			N/A
LPC		✓	✓		N/A
Dataset: MGH10
	ART	SyN	LO	LP	LPC
LO		✓	N/A		✓
LP				N/A	✓
LPC	✓		✓	✓	N/A

Rejecting the null-hypothesis indicates that the row-method and column-method are statistically equivalent. Equivalence is marked as ✓s in the table. We use Bonferroni correction to safe-guard against spurious results due to multiple comparisons by dividing the significance level α by 204 (the total number of statistical tests). The significance level for rejection of the null-hypothesis is α = 0.05/204.