Table 1

Summary of methodological issues

Scientific/study design issueStatistical concernSolution(s)
Current versus total durationCensoring status▶Current duration—ignore censoring
▶Total duration—infer censoring from questionnaire or follow-up
Fixed window samplingLeft truncation (length-biased sampling)▶Design study with longer sampling window
▶Restrict analysis to partnerships initiated within the sampling window, or
▶Use Kaplan–Meier estimate modified to account for left truncation
▶Randomly sample partnerships within a long window
Fixed number of partners samplingSample of partnerships biased towards individuals with fewer partners; possible length-biased sample▶Weight partnerships by total number of partners in lifetime
▶Randomly sample partnerships within a long window
Multiple partnerships per individualCorrelated data▶Use bootstrap or robust variance to compute CIs