Skip to main content
. 2023 Oct 13;12(10):10. doi: 10.1167/tvst.12.10.10

Table 4.

Significance of All Pairwise Differences Between Strategies in Terms of Mean Absolute Error (MAE), Mean Sensitivity Absolute Error (MS-AE), and Total Number of Presentations Per Test

P
Reliability Comparison MAE MS-AE Total Presentations
Reliable S-ZEST, with spatial correlations S-ZEST, without spatial correlations <0.001 0.052 <0.001
S-ZEST, with spatial correlations ZEST, with spatial correlations <0.001 0.894 <0.001
S-ZEST, with spatial correlations ZEST, without spatial correlations <0.001 0.006 <0.001
S-ZEST, without spatial correlations ZEST, with spatial correlations <0.001 0.250 <0.001
S-ZEST, without spatial correlations ZEST, without spatial correlations <0.001 0.882 <0.001
ZEST, with spatial correlations ZEST, without spatial correlations <0.001 0.048 <0.001
High FPs S-ZEST, with spatial correlations S-ZEST, without spatial correlations <0.001 0.077 <0.001
S-ZEST, with spatial correlations ZEST, with spatial correlations <0.001 <0.001 <0.001
S-ZEST, with spatial correlations ZEST, without spatial correlations <0.001 <0.001 <0.001
S-ZEST, without spatial correlations ZEST, with spatial correlations <0.001 <0.001 <0.001
S-ZEST, without spatial correlations ZEST, without spatial correlations <0.001 <0.001 <0.001
ZEST, with spatial correlations ZEST, without spatial correlations <0.001 <0.001 0.0280
High FNs S-ZEST, with spatial correlations S-ZEST, without spatial correlations <0.001 0.07 <0.001
S-ZEST, with spatial correlations ZEST, with spatial correlations <0.001 0.066 <0.001
S-ZEST, with spatial correlations ZEST, without spatial correlations <0.001 <0.001 <0.001
S-ZEST, without spatial correlations ZEST, with spatial correlations <0.001 0.999 <0.001
S-ZEST, without spatial correlations ZEST, without spatial correlations <0.001 0.127 <0.001
ZEST, with spatial correlations ZEST, without spatial correlations <0.001 0.133 <0.001

These comparisons were calculated by using the average result of the simulation for each eye, so that the amount of simulations did not affect the significance of the P values. All comparisons were corrected for multiple discoveries using the Bonferroni–Holm method. The strategy on the left is generally the better performing of the two, with the exception of the MS-AE for the S-ZEST with spatial correlations in the presence of high FPs.