Table 2.

Examples Illustrating Effects of the Conditional Independence Assumption on Estimates and on Fit of Model 2 When Exposure Has No Effect on Risk^[Note]

	When ConditionalIndependence Holds		When Child'sGenotype ModifiesProbability of Exposurea
ParentalMating Typeand No. ofCopies ofVariantAllele inOffspring	E = 1	E = 0	E = 1	E = 0
	Approximate Expected Count
(1,2):
2	25	30	49	6
1	25	30	31	24
(1,1):
2	15	25	34	6
1	30	49	41	38
0	15	25	15	25
(0,1):
1	49	126	79	96
0	49	126	49	126
	Estimated R1
	1.00	1.00	1.55	0.76
	Estimated R2
	1.00	1.00	2.47	0.22
	P Value for Test of Genotype-Exposure Interactionb
	1.00		10−9
	P Value for Test of Fitc
	1.00		.99

Note.— Illustrative data are approximate expected counts if a random sample of 1,000 case-parent triads were taken from the structured population described in the text.

^aDescription of dependence in text.

^b2-df LRT of H₀:η₁=η₂=0 in model 2.

^c4-df LRT of fit to model 2.