Recent evaluations have begun to use qualitative data in a manner that helps improve the quality and relevance of studies through the inferences that are drawn from them, and their applicability to policy makers and programme implementers. This paper reviews this work and identifies good practices to integrate qualitative methods into quantitative impact evaluations (IEs) and systematic reviews (SRs). Using recent literature on the characteristics of such practices, we developed two tools to assess the methodological rigour and mixed methods integration of 40 IEs and 7 SRs, drawing upon previous approaches. Our findings are that successful mixed methods quantitative impact evaluations: (1) provide a clear rationale for integration of methods; (2) deploy multidisciplinary teams; (3) provide adequate documentation; and (4) acknowledge limitations to the generalisability of qualitative and quantitative findings. Successful integration tended to improve mixed methods impact evaluations by collecting better data to inform the study design and findings, which helped contextualise quantitative findings. Our main observation on the integration of mixed methods in the systematic reviews is that mixed methods systematic reviews bringing together literatures that answer different questions can go beyond the 'sum of their parts' to provide holistic answers about development effectiveness. The findings of this study inform several recommendations to improve the conduct and reporting of mixed methods impact evaluations and systematic reviews.

The belief that contraceptive use causes infertility has been documented across sub-Saharan Africa, but its quantitative association with actual contraceptive use has not been examined. We collected and analyzed sociocentric network data covering  percent of the population in two villages in rural Kenya. We asked respondents to nominate people from their village (their network), and then we matched their network (alters) to the individual participant (ego) to understand how their beliefs and behaviors differ. We asked about contraceptive use and level of agreement with a statement about contraceptive use causing infertility. We calculated the average nominated network contraceptive use score and the average nominated network belief score. Holding the individual belief that contraceptive use causes infertility was associated with lower odds of using contraceptive (AOR = ., p = < .); however, when one's own nominated network connections held this belief, the odds of using contraceptive were even lower (AOR = ., p <.). Our findings show that this belief is associated with lower odds of contraceptive use and highlights the role that other people in one's network play in reinforcing it. Sexual and reproductive health programs should address this misperception at the individual and social network level.

No abstract