Economic Self-Help Group Programs for Improving Women s Empowerment: A Systematic Review Carinne Brody, Thomas De Hoop, Martina Vojtkova, Ruby Warnock, Megan Dunbar, Padmini Murthy, Shari L. Dworkin October 2015 Copyright 2014 American Institutes for Research. All rights reserved.
Women s Economic Self-Help Groups Female participants come together Receive Economic Intervention Collective Finance: Saving or Credit. Enterprise or Livelihoods Intervention: Business Training. 2
Summary Results Positive effects on women s economic, social, and political empowerment Poorest of the poor often do not participate No evidence for adverse effects on intimate partner violence No evidence for positive effects on psychological empowerment 3
Overview Presentation Evaluation Questions Theory of Change Systematic review: Methodology Quantitative Analysis Qualitative Analysis Synthesis Conclusion 4
Evaluation Questions 1. What is the impact of women s economic self-help groups on individual empowerment in low and middle-income countries? 2. What are the perspectives of female participants on factors determining their participation in, and benefits from, economic self-help groups? 5
Theory of Change 6
Measurement of empowerment? Economic: E.g. Women s bargaining power over financial decision-making, women s land ownership Social E.g. Freedom of movement, women s bargaining power over family size decision making. Political E.g. Political participation. Psychological E.g. Feelings of autonomy, confidence, self-sufficiency. 7
Systematic Review Systematic critical appraisal of literature Synthesis of literature (Waddington et al., 2012) Impact of self-help groups Perspectives of female self-help group members 8
Inclusion Randomized Controlled Trials, Quasi- Experimental Evaluations, and Multivariate Regression 9
Innovation: Inclusion Qualitative Studies Systematic reviews that only include quantitative evidence can be considered too rigid, failing to address other important questions, such as why an intervention does not work (Snilstveit, Oliver, & Vojtkova, 2012) In-Depth Interviews Ethnography Focus-Groups 10
Number of Included Studies 23 Quantitative Studies 11 Qualitative Studies 11
Risk of Bias Assessment Quantitative Selection bias 14% 24% 62% Performance bias 24% 10% 67% Low risk of bias Outcome and analysis reporting bias 29% 24% 48% Medium risk of bias High risk of bias Other biases 38% 38% 24% 0% 20% 40% 60% 80% 100% 12
Risk of Bias Assessment Quantitative 1. Selection bias: It is not clear whether the randomization was successful. The randomization was based on a relatively small sample of 4 treatment and 4 control villages. The authors do not show a table demonstrating balance across observable and unobservable characteristics. 2. Performance bias: The new members and old members appear to be selected from the same locations, suggesting that bias resulting from spillovers is an important concern. 3. Outcome Reporting Bias: There are serious inconsistencies in the reporting. The results reported in the text do not match those reported in the tables. 4. Other biases: The study uses recall data over a 4 year recall period for the DID estimation component. This may result in bias. 13
Risk of Bias Assessment Qualitative Table 5.2 Summary of quality appraisal of included qualitative studies Screening Question: Is there a clear statement of study aims? 100% Screening Question: Is qualitative methodology appropriate? 100% Appropriate research deign 0% Appropriate recruitment strategy 89% 11% 0% Appropriate data collection method 89% Consideration of researcher-participant relationship 22% 44% Consideration of ethical issues 0% Rigorous data analysis 44% Clear statement of findings 100% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Yes Can't tell No 14
Risk of Bias Assessment Qualitative 1. Recruitment of participants: The authors did not explain how the participants were selected. 2. Data collection tools: The authors did not describe their interview guides and data format. 3. Data saturation: There was no mention of data saturation as reason for stopping for recruitment. 4. Researcher-participant relationship: The authors did not report information about the researcher-participant relationship. 5. Data-analysis: The authors did not describe the analysis of the data in sufficient detail. 15
Effects on Economic Empowerment: RCTs Study % ID ES (95% CI) Weight Banerjee et al., 2014 India 0.01 (-0.04, 0.05) 35.51 Desai and Joshi, 2012, India 0.28 (0.12, 0.45) 30.19 Sherman et al., 2010, India 0.30 (-0.11, 0.70) 16.82 Kim et al., 2009 + Pronyk et al., 2006, South Africa 0.45 (0.06, 0.84) 17.48 Overall (I-squared = 81.3%, p = 0.001) NOTE: Weights are from random effects analysis 0.22 (-0.01, 0.44) 0.22 (-0.01, 0.44) 100.00 -.842 0.842 Impact SHGs on Economic Empowerment Based on RCTs 16
Effects on Family-Size Decision-Making Power: RCTs Study % ID ES (95% CI) Weight Desai and Tarozzi, 2013, Ethiopia -0.23 (-0.96, 0.50) 21.00 Kim et al., 2009 + Pronyk et al., 2006, South Africa 0.44 (-0.51, 1.39) 14.08 Desai and Joshi, 2012, India 0.45 (0.29, 0.62) 64.92 Overall (I-squared = 37.6%, p = 0.202) 0.31 (-0.09, 0.70) 100.00 NOTE: Weights are from random effects analysis 0.31 (-0.09, 0.70) -1.39 0 1.39 Impact SHGs on Social Empowerment Based on RCTs 17
Effects on Economic Empowerment: RCTs + Medium-Risk Selection-Bias Studies Study % ID ES (95% CI) Weight Banerjee et al., 2014 India 0.01 (-0.04, 0.05) 20.32 De Hoop et al., 2014 India 0.03 (-0.21, 0.27) 12.15 Pitt et al., 2006, Bangladesh 0.12 (0.03, 0.21) 18.81 Deininger and Liu, 2013 India 0.28 (0.20, 0.36) 19.34 Desai and Joshi, 2012, India 0.28 (0.12, 0.45) 15.45 Sherman et al., 2010, India 0.30 (-0.11, 0.70) 6.80 Kim et al., 2009 + Pronyk et al., 2006, South Africa 0.45 (0.06, 0.84) 7.14 Overall (I-squared = 86.8%, p = 0.000) 0.18 (0.05, 0.31) 100.00 NOTE: Weights are from random effects analysis 0.18 (0.05, 0.31) -.842 0.842 Impact SHGs on Economic Empowerment Based on RCTs and Medium Risk of Bias Quasi-Experimental Studies 18
Effects on Mobility: RCTs + Medium-Risk Selection-Bias Studies Study % ID ES (95% CI) Weight De Hoop et al., 2014 India 0.04 (-0.20, 0.27) 17.74 Pitt et al., 2006, Bangladesh 0.29 (0.19, 0.38) 39.81 Deininger and Liu, 2013 India 0.15 (0.07, 0.22) 42.45 Overall (I-squared = 70.8%, p = 0.033) 0.18 (0.06, 0.31) 100.00 NOTE: Weights are from random effects analysis 0.18 (0.06, 0.31) -.378 0.378 Impact SHGs on Mobility Based on Quasi-Experimental Studies 19
Mechanisms: Qualitative Economic Empowerment Women report feeling empowered after handling financial transactions Social Empowerment Women reported feeling mutual support within their groups Intimate Partner Violence Initial increase in violence but eventually they gained respect from their husbands 20
Limited participation of the poorest Too financially constrained for participation Caste or class discrimination 21
Revised Theory of Change: Some Key Assumptions Introduction Women s Self-Help Group Women become members of self-help group No caste and class discrimination Husband and motherin-law approve participation Knowledge about and time for participation 22
Revised Theory of Change 23
Potential Moderators: Training Microfinance did not show positive effect on empowerment in other SR So is training the secret? What type of training? Health education? Entrepreneurial skills? Women s rights? 24
Key Lessons Self-Help - Groups Scaling of Self-Help have positive Groups depends effects on on cost-effectiveness and other outcome economic, measures. social, and political empowerment No evidence - Self-Help for adverse Groups may effects be improved on intimate by partner stimulating the poorest to participate. violence - The findings are applicable to South Asia Poorest of but the extrapolation poor do to not the often rest the participate world is in selfhelp group hard. programs 25
Thomas de Hoop 202-403-6803 tdehoop@air.org 1000 Thomas Jefferson Street NW Washington, DC 20007 General Information: 202-403-5000 TTY: 887-334-3499 www.air.org 26
Effects on Political Empowerment: RCTs + Medium-Risk Selection-Bias Studies Study % ID ES (95% CI) Weight Pitt et al., 2006, Bangladesh 0.11 (0.02, 0.20) 57.59 Desai and Joshi, 2012, India 0.29 (0.12, 0.46) 42.41 Overall (I-squared = 71.2%, p = 0.062) 0.19 (0.01, 0.36) 100.00 NOTE: Weights are from random effects analysis 0.19 (0.01, 0.36) -.456 0.456 Impact SHGs on Political Empowerment Based on RCTs and Medium Risk of Bias Quasi-Experimental Studies 27
Effects on Psychological Empowerment: RCTs + Medium-Risk Selection-Bias Study % ID ES (95% CI) Weight De Hoop et al., 2014 India 0.00 (-0.24, 0.24) 95.16 Kim et al., 2009 + Pronyk et al., 2006, South Africa 0.50 (-0.55, 1.56) 4.84 Overall (I-squared = 0.0%, p = 0.363) 0.02 (-0.21, 0.26) 100.00 NOTE: Weights are from random effects analysis 0.02 (-0.21, 0.26) -1.56 0 1.56 Impact SHGs on Psychological Empowerment RCTs and Medium Risk of Bias Quasi-Experimental Studies 28
Effects on Intimate Partner Violence: RCTs + Medium-Risk Selection-Bias Studies Study % ID ES (95% CI) Weight Kim et al., 2009 + Pronyk et al., 2006, South Africa 0.04 (-0.13, 0.21) 59.79 De Hoop et al., 2014 India 0.11 (-0.09, 0.32) 40.21 Overall (I-squared = 0.0%, p = 0.600) 0.07 (-0.06, 0.20) 100.00 NOTE: Weights are from random effects analysis 0.07 (-0.06, 0.20) -.321 0.321 Impact SHGs on Domestic Violence Based on RCTs and Medium Risk of Bias Quasi-Experimental Studies 29
Effects on Economic Empowerment with Training: RCTs + Medium-Risk Selection-Bias Study % ID ES (95% CI) Weight De Hoop et al., 2014 India 0.03 (-0.21, 0.27) 12.08 Deininger and Liu, 2013 India 0.28 (0.20, 0.36) 56.53 Desai and Joshi, 2012, India 0.28 (0.12, 0.45) 22.01 Sherman et al., 2010, India 0.30 (-0.11, 0.70) 4.53 Kim et al., 2009 + Pronyk et al., 2006, South Africa 0.45 (0.06, 0.84) 4.86 Overall (I-squared = 16.7%, p = 0.308) 0.26 (0.17, 0.35) 100.00 NOTE: Weights are from random effects analysis 0.26 (0.17, 0.35) -.842 0.842 Impact SHGs on Economic Empowerment RCTs and Medium Risk of Bias Quasi-Experimental Studies with Training 30
Effects on Economic Empowerment without Training: RCTs + Medium-Risk Selection-Bias Study % ID ES (95% CI) Weight Banerjee et al., 2014 India 0.01 (-0.04, 0.05) 56.82 Pitt et al., 2006, Bangladesh 0.12 (0.03, 0.21) 43.18 Overall (I-squared = 77.9%, p = 0.034) 0.06 (-0.05, 0.16) 100.00 NOTE: Weights are from random effects analysis 0.06 (-0.05, 0.16) -.211 0.211 Impact SHGs on Economic Empowerment RCTs and Medium Risk of Bias Quasi-Experimental Studies without Training 31
Effects on Economic Empowerment High-Risk Selection-Bias Study % ID ES (95% CI) Weight Osmani, 2007, Bangladesh 0.37 (-0.10, 0.83) 29.31 Nessa et al., 2012, Bangladesh 0.65 (0.41, 0.89) 53.34 Swendeman et al., 2009, India 1.15 (0.47, 1.83) 17.36 Overall (I-squared = 42.1%, p = 0.178) NOTE: Weights are from random effects analysis 0.65 (0.33, 0.98) 100.00 0.65 (0.33, 0.98) -1.83 0 1.83 Impact SHGs on Economic Empowerment Based on High Risk of Bias Studies 32
Effects on Social Empowerment High-Risk Selection-Bias Study % ID ES (95% CI) Weight Rosenberg et al., 2011, Haiti Steel et al., 1998, Bangladesh Nessa et al., 2012, Bangladesh Swendeman et al., 2009, India Overall (I-squared = 10.3%, p = 0.342) NOTE: Weights are from random effects analysis 0.22 (-0.24, 0.69) 15.63 0.32 (0.16, 0.49) 71.10 0.79 (0.26, 1.32) 12.09 0.88 (-0.89, 2.65) 1.18 0.37 (0.18, 0.56) 100.00 0.37 (0.18, 0.56) -2.65 0 2.65 Impact SHGs on Social Empowerment Based on High Risk of Bias Studies 33
Publication Bias: Economic Empowerment 0.1 Studies 1% 5% 10% Standard error.2.3.4-1 -.5 0.5 1 Effect estimate 34
Publication Bias: Social Empowerment 0.2 Studies 1% 5% 10% Standard error.4.6.8 1-2 -1 0 1 2 Effect estimate 35