Table 2.

Key findings

Author(s) (Date)	Context	Study design	Sample size	Social network collection methods	Social network analysis methods	Variables	Key Findings in Relation to Mechanisms	p- value
Billari, Philipov & Testa (2009) [21]	Bulgaria	Cross-sectional	Men and women: 10,003, ages 18-34.	Name generator	Logistic regression models	Intentions to have a first and second child; attitudes, norms and perceived behavioral control related to fertility behavior.	Normative pressure, or the “perception of social influence” affected reproductive behavior, changing from being in favor of high childbirth rates to pro-contraception.	Norms; P = 0.00
Bove, Vala-Haynes, & Valeggia (2012) [5]	Mali	Cross-sectional	324 women, ages 15-80	Number of individuals a respondent identified.	Logistic and linear regression models	Pregnancy histories, women’s knowledge of contraception, and illness symptoms in the past three months and the treatment (if sought, financing and sources of social support).	Social influence in larger social networks resulted in increased pregnancy in the previous two years, associated with a larger social network.	Larger social network associated with increased odds of pregnancy during the previous 2 years, p < .01
Dynes, Stephenson, Rubardt & Bartel (2012) [4]	Ethiopia and Kenya	Cross-sectional	Ethiopia: 520 women; 300 men	Random generator.	Logistic regression model	Perceptions of current norms and community norms on current contraceptive use	Perceptions of social norms influenced reproductive behavior, including son preference and contraceptive use.	difference between women’s perception of the community ideal number of sons and their own actual number of sons is negatively associated with contraceptive use (Ethiopia OR 0.74, 95% CI 0.61–0.89; Kenya OR 0.77, 95% CI 0.66–0.89). a higher score on the family planning perception of other’s approval index was significantly associated with current contraceptive use among men and women in Kenya (OR 2.03, 95% CI 1.35–3.05 and OR 1.36, 95% CI 1.06–1.74, respectively); this association, however, was not present among samples in Ethiopia.
			Kenya: 655 women; 310 men
Edmonds, Hruschka, Bernard, & Sibley (2012) [19]	Bangladesh	Cross-sectional	246 women, 18-49 years.	Network generator and network characteristics.	Logistic regression models	Place of delivery, whether home or facility	The collective advice of others, or social influence, whether correctly perceived or not, affected birth decisions of women.	Skilled Birth Attendant Endorsement by network p = .000
Gayen & Raeside (2007) [18]	Bangladesh	Cross-sectional	694 women who had at least one child,	Name generator.	Logistic regression models	Experience of neonatal death and choice of assistance for delivery	Social influence impacts choice in type of assistance while giving birth.	Degree centrality in relation to unqualified assistance P = 0.00; degree centrality in relation to professional assistance p = .01.
Gayen & Raeside (2010) [20]	Bangladesh	Cross-sectional	694 women currently married of reproductive age	Name generator.	Logistic regression models	Current use of contraception	Both social learning and social influence impacted family planning decisions.	Network members’ approval of contraception, p < 0.05. Network members’ encouragement to use contraception, p < 0.05. Discussion frequency on contraception with network members, p < 0.05.
Kincaid (2000) [14]	Bangladesh	Longitudinal	860 married women, age 14-49.	Random generator.	Logistic regression model; Conditional (static-score) multiple regression analysis	Modern contraceptive use	A social network approach, specifically group discussions in key opinion leader’s homes, allowed for increased social influence to accelerate rate of change concerning contraceptive use.	Social network approach change in ideation p < 0.001, change in contraceptive use, p < 0.001.
Kohler, Behrman & Watkins (2001) [24]	Kenya	Longitudinal	694 women currently married	Name generator	Logistic regression; Measures of network density.	Family planning use	More heterogeneous groups with high amounts of activity were dominated by social learning and more homogenous groups are dominated by social influence.	In low-density Owich, Kawadhgone and Wakula South, the % users influence on family planning is p < 0.01; In high-density Obisa, density influences family planning p < 0.01.
Lindstrom & Munoz-Franco (2005) [23]	Guatemala	Cross-sectional	2871 women, age 18-35.	Random generator.	Multilevel logistic regression model	Contraceptive knowledge	Social learning is integral in areas where networks increased in heterogeneity. Key actors also influenced this learning.	Migration experience, family migration networks, and community urban out-migrant networks were statistically significant at precdicting the number of modern contraceptive methods known, p < 0.05.
Madhavan, Adams & Simon (2003) [13]	Mali	Cross-sectional	502 women, aged 15-45,	Random generator	Ordinary least-squares regression; logistic regression	Two fertility-related outcomes – completed fertility and contraceptive use	Homogenous networks facilitated social influence as a mechanism for diffusion; ‘gatekeepers’ generally dictated these societal norms and had more influence than others.	Ever use of contraceptives contraceptives: Presence of mother P < 0.05; % of network who are natal kin, p < 0.05; $ of network who are conjugal kin p < 0.01; % of network who live outside villag, p < 0.001.
Musalia (2003) [25]	Kenya	Cross-sectional	200 to 323 women, younger than 50	Name generator.	Logistic regression analysis.	Educational heterogeneity; membership in voluntary organization; network size; contraception use.	Social influence of kin groups affected spousal discussion of contraceptive use, but as gave way to social learning as new ideas were embraced.	Being a member of a social group: Kakamega, p < 0.05; Murang’a, p < 0.01.
Musalia (2005) [26]	Kenya	Cross-sectional	557 women and 536 men	Name generator.	Logistic regression analysis	Ever use of contraception and current use of contraception.	Social influence both hindered and helped the adoption of reproductive behaviors	Current use of contracetion, ntowrk advices use of family planning, p < 0.01; ever use of contraception, network advices use of family planning, p < 0.01.
Sandberg (2005) [29]	Nepal	Cross-sectional	77 currently married women, younger than 50	Name generator	Logistic-regression	Desiring more children.	Social learning and collective social experiences influenced actor decisions and behaviors.	Desiring more children impacted by network infant mortality, p < 0.05; and any child died in last birth interval, p < 0.01.
Valente, Watkins, Jato, Van Der Straten, & Tsitsol (1997) [27]	Cameroon	Cross-sectional	495 women, under the age of 45	Name generator.	Use logit-regression models.	Whether respondent ever-used a contraceptive, a clinic-based method, and a non-clinic based method.	Social influence and social learning were important within networks, though influence is heightened within associations due to encouragement between members.	Perceived approval of contraction, have used contraception, have encouraged network partners to use all p < 0.0001.
Behrman, Kohler & Watkins (2002) [16]	Kenya	Longitudinal	497 women; 324 men	Name generator.	Logit model	Whether a respondent was currently using contraception (at the time of the survey).	Social learning was the primary means of transmitting information through a network.	At least one family planning uer in the network p < 0.05; Number of remaining family planning users in the network, p < 0.01.
Boulay & Valente (1999) [22]	Kenya	Cross-sectional	2,217 women, aged 15-49; 2,152 men, aged 15-54	Random generator.	Logistic Regression models	Family planning knowledge, attitudes, and practices.	Extended social networks led to high amounts of transmission of family planning information passed through community groups.	Family planning knowledge, approval, use and discussion among members of clubs: know 5 modern methods, p < 0.001, and talked about family planning with anyone p < 0.01, with core network only, p < 0.05, and with core and extended networks, p < 0.001.
Valente & Saba (1998) [17]	Bolivia	Longitudinal	First sample: 2300 youngest men and women present in household; Second sample:800 residents in Potosi.	Name generator.	Regression model with demographic controls.	Family planning awareness; reproductive health knowledge; reproductive health attitudes; family planning intention; interpersonal communication; current use of contraceptives.	Social learning in the form of mass media campaigns were associated with behavior change for individuals who have networks with low amounts of contraceptive use.	Network exposure and current use of contraception (p < 0.01) was associated with family planning awareness p < 0.01, reproductive health knowledge p < 0.01, reproductive health attitude p < 0.01, family planning intention p < 0.01,
Godley (2001) [28]	Thailand	Cross-Sectional	1,563 women aged 18-35 who had been married 10 years or less.	Random generator.	Logistic regression models; multilevel networks	Choice in contraceptive.	The specific social network of extended kin influenced contraceptive choice both through both social learning and social influence.	Method choice without television with p < 0.05, and method choice with television, p < 0.05.