Table 1 Empirical size of the tests under different number of samples (N) and dimensions \((p=q)\) at 0.05 significance level.
From: A fast kernel independence test for cluster-correlated data
Type | N | \(p=q\) | Pearson | Perm | Davies |
|---|---|---|---|---|---|
Normal | 50 | 50 | 0.049 | 0.050 | 0.000 |
100 | 0.050 | 0.051 | 0.000 | ||
200 | 0.055 | 0.061 | 0.000 | ||
100 | 50 | 0.048 | 0.053 | 0.005 | |
100 | 0.049 | 0.050 | 0.000 | ||
200 | 0.058 | 0.057 | 0.000 | ||
200 | 50 | 0.048 | 0.050 | 0.016 | |
100 | 0.052 | 0.053 | 0.006 | ||
200 | 0.053 | 0.053 | 0.001 | ||
Log-normal | 50 | 50 | 0.055 | 0.056 | 0.024 |
100 | 0.049 | 0.047 | 0.009 | ||
200 | 0.041 | 0.040 | 0.005 | ||
100 | 50 | 0.053 | 0.050 | 0.041 | |
100 | 0.052 | 0.053 | 0.038 | ||
200 | 0.050 | 0.050 | 0.017 | ||
200 | 50 | 0.054 | 0.054 | 0.058 | |
100 | 0.044 | 0.044 | 0.047 | ||
200 | 0.055 | 0.055 | 0.036 |
