Table 1.

Estimates of $c$ and $q$

	Anchor sensitivity c	Phenotype prevalence q
True value	0.5 (—)	0.1 (—)
Ideal learning	—	0.10 (0.0003)
Naive logit	—	0.05 (0.002)
EN algorithm	0.37 (0.04)	0.14 (0.01)
ML method	0.50 (0.021)	0.10 (0.004)

Values are mean (empirical standard error) over 1000 iterations.

EN: Elkan and Noto.