An Assessment of Third‐Party Reporting of Close Ties to Measure Sensitive Behaviors: The Confidante Method to Measure Abortion Incidence in Ethiopia and Uganda

Abstract

Indirect estimation techniques are important tools for measuring sensitive and stigmatized behaviors. This includes third-party reporting methods, which have become increasingly common in the field of abortion measurement, where direct survey approaches notoriously lead to underreporting. This paper provides the first in-depth assessment of one of the most widely used of these techniques in the field of abortion measurement: the confidante method. We outline six key assumptions behind the confidante method and describe how violations of these assumptions can bias resulting estimates. Using data from modules added to the performance monitoring for action surveys in Uganda and Ethiopia in , we compute one-year abortion incidence estimates using the confidante method. We also perform a validation check, using the method to estimate intrauterine device /implant use. Our results revealed implementation problems in both settings. Several of the method's foundational assumptions were violated, and efforts to adjust for these violations either failed or only partially addressed the resulting bias. Our validation check also failed, resulting in a gross overestimate of intrauterine device/implant use. These results have implications more broadly for the potential biases that can be introduced in using third-party reporting of close ties to measure other sensitive or stigmatized behaviors.

Induced abortion incidence and safety in Burkina Faso in 2020: Results from a population-based survey using direct and social network-based estimation approaches

“…However, abortion rates among respondents who reported having 0 versus 1 or more friends were not significantly different (4.9 versus 4.2) and is accounted for in the adjusted estimates. Ultimately, the magnitude of differences between the friend and adjusted surrogate samples were small (less than 2 percentage points, except for residence) and showed similar patterns of contraceptive use, a validation approach suggested by others [ 26 ]. The adjustment of abortion incidence accounting for surrogate sample selection bias was minimal.…”

“…While we anticipate the self-reported abortion rates are substantial underestimates, friend estimates are not inherently more accurate. The best friend methodology relies on several assumptions [ 26 ]. In the absence of a validated data source on abortion, it is not possible to evaluate the extent to which the friend incidence estimates are accurate.…”

“…While we anticipate the self-reported abortion rates are substantial underestimates, friend estimates are not inherently more accurate. The best friend methodology relies on several assumptions [26]. In the…”

“…We used design-based F-tests to evaluate whether respondent and adjusted friend characteristics were statistically significantly different. To adjust for transmission bias, we multiplied friend abortion rates by the inverse probability of respondents sharing information about their own abortion with their closest friend [ 18 , 21 , 25 , 26 ]. Given we observed higher sharing for pregnancy removal than period regulation, we accounted for the distribution of the abortion types and respective sharing in calculating the transmission bias adjustment factor.…”

“…These methods broadly assume that 1) women have close friends who are similar in the aggregate with regard to socioeconomic and reproductive characteristics (i.e., no selection bias in the surrogate sample of female friends), 2) women share their abortion experience with their close friends (i.e., no transmission bias); and 3) reporting of abortion is improved when talking about friends versus self (i.e., reduced social desirability bias). Studies using social network-based methods to estimate abortion incidence have generally produced higher estimates of abortion incidence, however, results have been mixed in relation to the extent of assumption violation and the success of subsequent adjustments for observed biases [20,[22][23][24][25][26]. While social network-based method assumption violations may bias estimates of abortion incidence, much of the current global estimates rely heavily on AICM estimates, which may be biased in other ways as they rely on untestable key informant assessments of the extent of clandestine abortions that don't lead to hospitalization to adjust facility-based postabortion care rates.…”