Table 2. Recall error variation by participant characteristics.

	First quartile of recall error			Last quartile of recall error
	RRR1	95% CI		RRR	95% CI
Sex
male	1.00			1.00
female	0.80	0.54	1.17	0.32	0.20	0.52
Race
white	1.00			1.00
black	1.44	0.99	2.11	2.98	1.98	4.47
other	0.74	0.31	1.81	1.07	0.39	2.91
Age at Wave IV (in years)
1st quartile [25.2, 28.5)	0.88	0.50	1.54	0.96	0.54	1.72
2nd quartile [28.5, 29.5)	0.82	0.50	1.32	1.15	0.61	2.15
3rd quartile [29.5, 30.3)	1.00			1.00
4th quartile [30.3, 31.5]	1.10	0.71	1.71	1.35	0.75	2.42
Age at first sex (in years)
1st quartile [0, 13.6)2	0.57	0.30	1.11	10.92	6.17	19.33
2nd quartile [13.6, 14.9)	0.85	0.53	1.37	2.22	1.29	3.81
3rd quartile [14.9, 16)	1.00			1.00
4th quartile [16], [18]	1.68	1.07	2.65	0.37	0.14	0.99
Health
excellent, very good or good	1.00			1.00
fair or poor	1.03	0.62	1.71	0.92	0.50	1.68
Highest education achieved
didn't finish high school or lower	1.00			1.00
high school graduate	0.55	0.30	1.00	0.96	0.51	1.78
with some college education or above	0.26	0.14	0.48	0.57	0.31	1.05
Health insurance in the last 12 months
without cover	1.00			1.00
with cover but less than 12 months	1.04	0.59	1.84	0.82	0.45	1.48
with cover for 12 months	1.49	0.93	2.39	0.73	0.44	1.20
Sexuality
100% heterosexual	1.00			1.00
other	0.89	0.54	1.47	1.01	0.54	1.90
Total number of sex partners over lifetime
1 to 5	0.85	0.48	1.51	1.56	0.80	3.05
6 to 9	1.00			1.00
10 to 19	1.18	0.70	2.00	0.64	0.34	1.20
20 or more	1.67	0.99	2.80	0.49	0.26	0.92

¹Relative-risk ratio

2[ ] and ( ) were used as notations for closed and opened intervals, respectively.

^*−0.92 and 1 were the 25 and 75 percentile, respectively.

^*Recall error falling within the 25−75 percentile were used as the base-level in the multinomial logistic regression (in a similar manner to the control group in a logistic regression)