Case studies and evaluations

Design

Stepped-wedge cluster design with difference-in-differences estimation

Sample

48 clusters · 96,400 residents · 24-month follow-up

Principal findings

• Acute watery diarrhoea incidence in children under five fell 55.1% (95% CI ±6.2).
• Household water collection time fell from 148 to 41 minutes per day.
• Community maintenance entities reached operational cost recovery at month 18.

Design

Randomised encouragement design with administrative tax-registry linkage

Sample

5,240 participants · 1,880 enterprises tracked

Principal findings

• 68% of incubated enterprises still trading at 24 months versus 41% in the comparison arm.
• Average 3.1 additional jobs created per surviving enterprise.
• Loan portfolio repayment rate of 91% across four cohorts.

Design

Interrupted time-series across 318 reporting facilities

Sample

318 facilities · 61 districts · 6 countries

Principal findings

• Median detection-to-notification interval fell from 9.1 to 3.4 days.
• Weekly reporting completeness rose from 41% to 88%.
• Cost per person under surveillance of USD 0.42 per year.

Design

Propensity-score matched cohort with savings-group administrative data

Sample

3,900 women · 260 savings groups

Principal findings

• Nutrition and hygiene expenditure rose 27% in participating households.
• Women held 44% of enterprise leadership positions by endline.
• Group savings mobilisation multiple of 4.6x initial capitalisation.

Design

Quasi-experimental with anthropometric panel surveys

Sample

41,700 children screened · 18 districts

Principal findings

• Stunting prevalence in programme districts fell 26.2% (95% CI ±2.7).
• Referral completion within seven days improved from 38% to 79%.
• Screening cost of USD 1.90 per child per round.

Design

Comparative statutory audit with enforcement-pipeline tracing

Sample

37 statutory instruments · 21 authorities

Principal findings

• 19 national or sub-national policies amended or informed by audit findings.
• Enforcement action rates rose in 12 of 21 audited jurisdictions.
• Audit cost of USD 0.19 per capita in the audited jurisdiction.

Design

Implementation study with data-quality auditing

Sample

14 registries · 4.1m records harmonised

Principal findings

• Duplicate record rate reduced from 11.4% to 1.8% after deterministic matching.
• Median query response time for district dashboards under 900 ms.
• All exchanges operate under ministry data ownership and audit-trail retention.

Design

Panel study with composite vulnerability index stratification

Sample

22,300 households · 9 districts

Principal findings

• Household income recovery to pre-shock levels achieved 11 months earlier than comparison districts.
• Negative coping strategy index reduced by 34%.
• Water-point functionality sustained above 92% through the dry season.