Table 1.

Data for demonstration

Risk level at baseline	Exposure before baseline	Exposure at baseline
		non-user		user
		person-years	cases	Person-years	cases
Randomization in place (corresponding to Estimation 1)
I1	non-user	10000	100	10000	200
	user	15000	150	15000	300
	Subtotal	25000	250	25000	500
I2	non-user	7500	225	7500	450
	user	7500	225	7500	450
	Subtotal	15000	450	15000	900
I3	non-user	2500	125	2500	250
	user	7500	375	7500	750
	Subtotal	10000	500	10000	1000
Non-randomized, Correct stratification of confounding effects (corresponding to Estimation 2, 3)
I1	non-user	6000	60	14000	280
	user	6000	60	24000	480
	Subtotal	12000	120	38000	760
I2	non-user	7500	225	7500	450
	user	7500	225	7500	450
	Subtotal	15000	450	15000	900
I3	non-user	4000	200	1000	100
	user	12000	600	3000	300
	Subtotal	16000	800	4000	400
Non-randomized, Incorrect stratification of confounding effects (corresponding to Estimation 4)
I1	non-user	6000	60	14000	280
	user	6000	60	24000	480
	Subtotal	12000	120	38000	760
I2 and I3	non-user	11500	425	8500	550
	user	19500	825	10500	750
	Subtotal	31000	1250	19000	1300