Table 3

Type I error rate for the test for the mean effect size when no evidence of bias was present

Publication biasNτ2I2 (%)AllEggerRankHarbordPetersSchwarzerArc-EggerArc-rankTrim
None100.02170.060.060.060.060.060.060.060.060.06
None100.12560.100.100.100.090.100.100.100.100.10
None100.90900.100.100.100.100.100.100.100.100.10
None200.02170.070.070.070.070.070.070.070.070.07
None200.12560.080.080.080.080.080.080.080.080.08
None200.90900.080.080.080.080.080.080.080.080.08
None500.02170.060.060.060.060.060.060.060.060.06
None500.12560.060.060.060.060.060.060.060.060.06
None500.90900.060.060.060.060.060.060.060.060.06
None1000.02170.070.060.060.070.070.070.070.070.07
None1000.12560.060.060.060.060.060.060.060.060.06
None1000.90900.060.060.060.060.060.060.060.060.06
RR=4100.02170.110.110.110.110.110.110.110.110.11
RR=4100.12560.220.220.220.220.220.220.220.220.22
RR=4100.90900.410.410.420.410.410.420.410.410.41
RR=4200.02170.140.140.140.140.140.140.140.140.14
RR=4200.12560.290.280.290.280.280.290.280.280.29
RR=4200.90900.610.610.610.610.600.610.610.610.61
RR=4500.02170.210.210.210.210.210.210.210.210.21
RR=4500.12560.540.530.540.530.530.540.530.530.54
RR=4500.90900.910.910.910.910.910.910.910.910.91
RR=41000.02170.350.350.350.340.350.350.350.350.35
RR=41000.12560.810.800.810.800.800.810.800.800.81
RR=41000.90901.001.001.001.001.001.001.001.001.00
RR=10100.02170.250.250.250.250.250.250.250.250.25
RR=10100.12560.540.530.540.540.530.540.530.540.54
RR=10100.90900.790.790.790.790.790.790.790.790.79
RR=10200.02170.370.370.370.370.370.370.370.370.37
RR=10200.12560.770.770.770.770.770.770.770.770.77
RR=10200.90900.960.960.960.960.960.960.960.960.96
RR=10500.02170.710.710.710.710.710.710.710.700.71
RR=10500.12560.980.980.980.980.980.980.980.980.98
RR=10500.90901.001.001.001.001.001.001.001.001.00
RR=101000.02170.940.940.940.940.940.940.940.940.94
RR=101000.12561.001.001.001.001.001.001.001.001.00
RR=101000.90901.001.001.001.001.001.001.001.001.00
  • The column ‘All’ shows the type I error rates for the test for the mean effect size based on all samples. The remaining columns show the type I error rates based on meta-analyses, in which no publication bias was detected by the test in the column heading.