The causal impact of user-centered counseling on women's reproductive health outcomes: evidence from Malawi
OA Version
Citation
Abstract
Nearly half of all pregnancies in the U.S. —and comparable numbers worldwide—are unintended, leading to adverse health outcomes for mothers and children, financial constraints and family instability, diminished educational and developmental outcomes for children, and worse labor market outcomes for women. Effective family planning counseling can mitigate these consequences by correcting misinformation and misconceptions around contraceptive methods, and improve overall contraceptive decision-making for women. User-centered counseling (UCC) refines the standard counseling approach by first eliciting women’s contraceptive preferences, then tailoring the session to discuss a subset of methods aligned with those preferences. We conduct a randomized controlled trial in Malawi involving over 700 women, where we evaluate the effects of UCC alongside partner invitation (PI), where women are encouraged to involve their partners in the counseling process. In the first chapter, we study the impacts of both interventions on several outcomes around contraceptive method use dynamics and fertility. We find that women encouraged to involve their partners tended to move from injectables to implants, suggesting a partner-driven preference. However, UCC appears to neutralize this effect. Contraceptive consistency, proxied by the maximum number of consecutive months women use same method or number of months women use any method, was significantly influenced by UCC, fostering more regular use of contraceptives, although it was somewhat mitigated when combined with PI. In terms of fertility outcomes, we observe that UCC seems to reduce pregnancies and births, but when combined with PI, the effect is the opposite. Finally, neither UCC nor PI significantly alter the likelihood of discontinuation. In the second chapter, we explore how women’s preferences over contraceptive methods evolve and whether method uses are concordant with these preferences or not. We first document that preferences vary significantly in both the short and long term. We then observe that both user-centered counseling and partner invitation interventions have minimal impact on aligning women’s contraceptive preferences with their actual usage, both in the short and long term. In the third chapter, we apply two methods to examine the heterogeneous impacts of UCC on a range of family planning outcomes. Firstly, we utilize the Sorted Effects method, which involves estimating and ordering partial effects based on covariates, graphically representing this variation, and also comparing the mean characteristics of the most with the least affected women. Secondly, we adopt the Generic Machine Learning method, suitable for our data’s high-dimensional nature, which helps estimate features (Best Linear Predictor (BLP) and Group Average Treatment Effects (GATES)) of the conditional average treatment effects without relying on predetermined covariates. Nonetheless, our experiment’s power to uncover meaningful heterogeneities in the impact of UCC is limited.
Description
2024