Table 1.

Model selection. (Base model comprises of oestradiol, education, race/ethnicity, body mass index, smoking habits, age of menarche, overall health and age of first interview. MS, marital status; SF, sexual frequency; MP, lives with a male partner; M, lives with a male; TM, total number of males in household. K, number of parameters; AIC, Akaike information criterion; w_i, model probability. Best fitting model(s) in italics. (n = 2936, n events = 1324).)

		all models			hypothesis 1 (models a–d)			hypothesis 2 (models a, e–g)
model	K	AIC	ΔAIC	wi	AIC	ΔAIC	wi	AIC	ΔAIC	wi
a, base	18	17 399.57	11.40	0.00	17 399.57	11.40	0.00	17 399.57	0.00	0.35
b, base + MS	20	17 403.05	14.88	0.00	17 403.05	14.88	0.00	—	—	—
c, base + SF	20	17 389.35	1.18	0.35	17 389.35	1.18	0.36	—	—	—
d, base + SF + MS	22	17 388.17	0.00	0.64	17 388.17	0.00	0.64	—	—	—
e, base + MP	19	17 401.37	13.21	0.00	—	—	—	17 401.37	1.81	0.14
f, base + M	19	17 399.88	11.71	0.00	—	—	—	17 399.88	0.32	0.30
g, base + TM	19	17 400.53	12.36	0.00	—	—	—	17 400.53	0.96	0.22