Background Detailed sexual partnership histories are frequently collected, yet data quality is rarely reported. We present data on internal consistency of partnership reports among youth in Tanzania, and assess the usefulness of specific questions.

Methods In 2007/8, sexual behaviour and biological samples for STIs were collected from 13,814 15–30 year olds in 20 communities, in Mwanza, Tanzania. Using a face-to-face questionnaire, participants were asked: number of sexual partners (ever, last year, last 4 weeks); new partners in the past year; detailed questions about last 3 partners in the past year, including dates of first and last intercourse. We examined response consistency and, using STI and pregnancy results, assessed the validity of the question ‘Have you ever had sex?’

Results Of 1108 self-reported virgins, 12.8% had biological results indicative of sexual activity, with misreporting more common among males and younger participants. Within a time period (e.g. 1 year, 4 weeks), inconsistencies between reported number of partners and number calculated from intercourse dates were more common when partnerships began at the start of the reference period. For example, of the 7.5% of participants with inconsistent reports of number of new partners in the last year, 43% had a partnership with a start date reported as ‘exactly’ 1 year ago. The proportion of partnerships reported as ongoing was higher for more stable relationship categories and where last sex was more recent. Among unmarried participants, for example, last intercourse was reported as within the last 4 months in 91% and 84% of ongoing partnerships among males and females, respectively (Table).

Conclusions Internal consistency was good overall. Inclusion of STI testing, and questions on number of partnerships and timing of intercourse allowed the identification of reporting biases. Reports of relationship status were plausible and ongoing status may be a reasonable proxy for recency of sex.