Table 1

Statistical table

FigureComparisonType of testStatisticConfidence interval or power
a 1B Rest vs working memory best iGC classification accuracy valueBinomial testClopper–Pearson confidence intervals96.3–98.0%
bRest vs working memory best dGC classification accuracy value 90.5–93.2%
cRest vs task maximum classification accuracy (each bar) vs chancePermutation testp valuep < 0.001
d 1F N-way task classification maximum iGC accuracy valueBinomial testClopper–Pearson confidence intervals73.3–75.4%
eN-way task classification maximum dGC accuracy value 46.4–48.7%
fN-way task classification maximum accuracy values (each bar) vs chancePermutation testp valuep < 0.001
g 1G Subtask classification maximum accuracies (each bar) vs chancePermutation testp valuep < 0.05
h 1H (left)Rest vs task dGC classification accuracies with 2×, and 3× sampling rate (vs 1×)Wilcoxon one-tailed signed-rankp value2×: p = 0.02;
3×: p = 0.06
i Rest vs task iGC classification accuracies with 2×, and 3× sampling rate (vs 1×) 2×: p = 0.01;
3×: p = 0.01
j 2C ZCA, GEV classification accuracy values with dGC and iGC vs chancePermutation testp valuep < 0.001
k 2D Rest vs task unweighted dGC classification accuracy value (each bar) vs chancePermutation testp valuep < 0.001
l 3A Each dGC, iGC, PC matrix connectivity value bound with black square vs corresponding null distributionPhase scramblingp valuep < 0.05
m 3C Each dGC, iGC, PC matrix connectivity value bound with black square vs corresponding null distribution of phase-scrambled surrogatesBenjamini–Hochberg correctionp valuep < 0.05
n 5C Each prediction corr value with filled circle in stem plotBenjamini–Yekutieli correctionp valuep < 0.05
o 5D Correlation coefficients between observed and predicted composite scores, for the same subject vs across different subjectsKolmogorov–Smirnov testp valuep < 0.001