Skip to main content
. Author manuscript; available in PMC: 2021 Mar 1.
Published in final edited form as: Behav Genet. 2020 Feb 10;50(2):127–138. doi: 10.1007/s10519-020-09993-9

Table III:

Proportion of confidence intervals out of 192 simulated intervals (each replicated 10,000 times) with expected coverage or non-liberal coverage.a,b

Proportion
with
expected
coverage		 Proportion
with non
liberal
coverage
Robust CI 0.13 0.08
Typical CI 0.01 0.10
Typical Adjusted by 2 0.19 0.72
Standard DEA Bootstrap standard CI 0.16 0.11
Univariate DEA WGS Bootstrap standard CI 0.30 0.26
Standard DEB .5 Bootstrap standard CI 0.15 0.87
Univariate DEB .5 WGS Bootstrap standard CI 0.05 1.00
a

DEA: Double entry after bootstrapping; DEB: Double entry before bootstrapping; .5: bootstrap resample size was half the (double entered) sample size; 1: bootstrap resample size was equal to the size of the (double entered) sample size; WGS: within group sampling was used for the univariate bootstrap; UGS: ungrouped sampling, or sampling without regard to class membership was used for the univariate bootstrap.

b

Test for proper cases was two-tailed, test for non-liberal was one tailed. This resulted in some intervals faring poorer in the non-liberal Type I error case than in the expected Type I error test.