Skip to main content

Table 1 Type-I error and power comparison for group comparison tests

From: Highly efficient hypothesis testing methods for regression-type tests with correlated observations and heterogeneous variance structure

 

ρ=0.2

ρ=0.8

 

Type-I error

Power

Type-I error

Power

Normal

PB.t (oracle)

0.037 (0.005)

0.841 (0.010)

0.035 (0.005)

0.919 (0.010)

PB.t (estimation)

0.036 (0.005)

0.839 (0.010)

0.029 (0.005)

0.910 (0.009)

Weighted LMER

0.042 (0.007)

0.826 (0.011)

0.022 (0.003)

0.616 (0.016)

Rfast LMER

0.064 (0.008)

0.555 (0.016)

0.063 (0.008)

0.517 (0.017)

Weighted regression t-test

0.032 (0.005)

0.822 (0.012)

0.002 (0.001)

0.373 (0.011)

Paired t-test

0.049 (0.006)

0.503 (0.016)

0.050 (0.006)

0.475 (0.013)

Welch’s t-test

0.031 (0.004)

0.449 (0.015)

0.001 (0.001)

0.090 (0.010)

Double Exponential

PB.wilcox (oracle)

0.032 (0.007)

0.898 (0.012)

0.030 (0.007)

0.950 (0.007)

PB.wilcox (estimation)

0.046 (0.010)

0.861 (0.016)

0.032 (0.007)

0.918 (0.012)

svyranktest

0.121 (0.010)

0.821 (0.013)

0.100 (0.012)

0.615 (0.024)

Friedman

0.050 (0.009)

0.492 (0.018)

0.050 (0.006)

0.513 (0.014)

Wilcoxon signed rank

0.056 (0.008)

0.569 (0.016)

0.054 (0.005)

0.563 (0.015)

Wilcoxon rank-sum

0.042 (0.009)

0.595 (0.013)

0.002 (0.001)

0.211 (0.015)

  1. At the 5% significance level, mean and standard deviation (in brackets) of the type-I error rate and power over 20 sets of SIM1 data are reported