Table 2 RVE meta-regression results across all outcomes by publication, implementation and effect characteristics

From: Social safety nets, women’s economic achievements and agency in 45 countries: a systematic review and meta-analysis

 

Basic controls

Extended controls

Coefficient

SE

I2

Coefficient

SE

I2

Panel A: study level

Region (omitted = sub-Saharan Africa)

 South Asia

0.033

(0.029)

91.78%

0.046

(0.029)

91.95%

 Latin America and the Caribbean

−0.063*

(0.026)

−0.043

(0.046)

 Middle East and North Africa

−0.009

(0.041)

0.017

(0.047)

 East Asia and Pacific

−0.070

(0.037)

−0.045

(0.044)

 Europe and Central Asia

0.519***

(0.023)

0.526***

(0.043)

Income group (omitted = lower-middle)

 Low income

0.038

(0.029)

91.63%

0.044

(0.034)

 Upper-middle income

−0.020

(0.025)

0.008

(0.041)

Fragile setting

−0.010

(0.037)

91.89%

−0.019

(0.041)

Urban setting

−0.033

(0.026)

91.89%

−0.014

(0.030)

Panel B: intervention arm level

Intervention type (omitted = UCT)

 CCT

−0.055

(0.032)

91.79%

−0.075*

(0.031)

92.64%

 Food, voucher or in-kind transfers

−0.046

(0.049)

−0.045

(0.055)

 Asset transfers

0.002

(0.031)

−0.014

(0.039)

 Public work programmes

0.011

(0.049)

0.056

(0.050)

 Social care services

−0.009

(0.029)

−0.009

(0.042)

Implementer (omitted = NGO, UN or other)

 Government

−0.008

(0.027)

91.88%

−0.026

(0.033)

 Researchers

−0.002

(0.035)

−0.025

(0.041)

Scale of implementation (omitted = mid-level)

 Pilot

−0.058*

(0.026)

91.79%

−0.079*

(0.030)

 At scale

−0.028

(0.028)

−0.028

(0.034)

Value of economic benefit (omitted = first tercile)

 Second tercile

−0.018

(0.025)

91.88%

−0.004

(0.028)

 Third tercile

−0.008

(0.031)

−0.005

(0.035)

Targeting approach

 Poverty targeted

−0.001

(0.026)

91.93%

−0.010

(0.034)

 Gender targeted

0.012

(0.024)

0.036

(0.040)

Plus components (omitted = no Plus)

 Gender-neutral

0.022

(0.050)

92.12%

0.049

(0.053)

 Gender-sensitive

0.026

(0.043)

0.037

(0.053)

 Economic

0.051

(0.032)

0.059

(0.044)

 Health or protection

0.005

(0.045)

0.007

(0.048)

 Training or information

−0.008

(0.037)

−0.010

(0.043)

 Other

−0.011

(0.041)

−0.021

(0.041)

Female sample age (omitted = ≤24 years)

 Age: 25–39 years

0.002

(0.031)

91.96%

−0.019

(0.039)

 Age: 40+ years

0.042

(0.041)

0.053

(0.051)

Panel C: effect level

Outcome category (omitted = productive work intensity)

 Labour force participation

0.045

(0.023)

91.26%

0.044

(0.025)

91.40%

 Care work participation

0.227

(0.162)

0.229

(0.158)

 Care work intensity

−0.049

(0.029)

−0.050

(0.029)

 Savings

0.138**

(0.037)

0.134**

(0.037)

 Debt or loans

0.021

(0.052)

0.027

(0.054)

 Assets

0.125*

(0.041)

0.127*

(0.041)

 Expenditure

0.110

(0.054)

0.107

(0.055)

 Decision-making

0.031

(0.034)

0.025

(0.034)

 Autonomy and self-efficacy

0.028

(0.034)

0.025

(0.032)

 Voice

0.113*

(0.046)

0.115

(0.048)

Duration of the intervention (omitted = <12 months)

 12+ months

0.016

(0.028)

91.88%

0.019

(0.026)

Time post-intervention at survey (omitted = <12 months)

 Over 12+ months

−0.011

(0.023)

91.88%

−0.001

(0.022)

  1. All estimates are from RVE regressions with n = 1,307 effects and 93 studies. CCT, conditional cash transfer; SE, standard error; UCT, unconditional cash transfer. *P < 0.05, **P < 0.01, ***P < 0.001. I2 estimates are entered once per model and apply to the remaining estimates in each category; double quote symbol (") indicates ‘same as above’. Basic control estimates are obtained from separate regressions by background characteristic, controlling for publication type, year-of-publication splines, an indicator for low- or medium-quality assurance, and whether the publication reports an individual (rather than cluster) RCT. Extended control estimates additionally control for all other covariates at the same level (study, intervention or effect).