Skip to main content

Table 1 Estimated type I error rates for the three mixed model approaches for binary traits (LEAP, GMMAT and CARAT), the LMM method (GEMMA) and logistic regression with principal component based correction (PCA)

From: Comparison of mixed model based approaches for correcting for population substructure with application to extreme phenotype sampling

Cohort Sample Size (N)

Sub-sample Size (0.2N)

LEAP

GMMAT

CARAT

GEMMA

PCA

5000

1000

0.0405

0.04135

0.102.

0.061

0.0575

10000

2000

0.0417

0.0475

0.089

0.046

0.0605

20000

4000

0.0450

0.0515

0.0945

0.052

0.0555

  1. *Based on m=1999 simulations