Fighting Poverty: International Attitudes Toward Poverty Reduction Programs

All respondents first watch a two-minute neutral explainer video introducing the four core program types (unconditional cash transfers, conditional cash transfers, in-kind transfers, public works) plus one randomly assigned “new” program (asset transfer or personalized tax refund), followed by a knowledge-check quiz. The explainer presents no information on program effectiveness and is common to all arms (it is not part of the treatment contrast).

Respondents are then randomized to one of three conditions:
• Treatment 1 — “Conditions faced by the Poor” video. Accurate information on the living conditions, vulnerability, and limited opportunities faced by poor households. Designed to shift perceptions of poverty.
• Treatment 2 — “Effectiveness” video. Corrects three empirically incorrect beliefs: that cash transfer recipients waste money on harmful goods; that cash transfers reduce labor supply; and that consumption tax exemptions primarily benefit poor households.
• Control. No second video after the common neutral explainer.

2026-06-05

2028-03-30

Standardized indices, pooled across countries with country fixed effects:
• Overall support — qualitative index. Five-point Likert items on general support and on each core program.
• Overall support — quantitative index. Preferred total spending across the seven hypothetical budget allocations.
• Relative support for cash transfers. Support for UCTs alone and for UCTs+CCTs combined.
• Tax exemptions index. Support for consumption tax exemptions relative to targeted social assistance.

See questionnaire and pre-analysis plan

Randomized cross-country survey experiment, where respondents are allocated to one of three conditions:
• Treatment 1 — “Conditions faced by the Poor” video. Accurate information on the living conditions, vulnerability, and limited opportunities faced by poor households. Designed to shift perceptions of poverty.
• Treatment 2 — “Effectiveness” video. Corrects three empirically incorrect beliefs: that cash transfer recipients waste money on harmful goods; that cash transfers reduce labor supply; and that consumption tax exemptions primarily benefit poor households.
• Control. No second video after the common neutral explainer.

Not available

Randomization performed by the survey platform using a computer pseudo-random number generator. All randomized elements are independent and assign options with equal probability within their respective sets; treatment arms use a fixed 30/30/40 split.

Individual

No

N/A

16,000 respondents (target 2,000 per country × 8 countries).

Per the 40/30/30 allocation: ~6,400 control; ~4,800 “Conditions faced by the Poor”; ~4,800 “Effectiveness” (9,600 in the two treatment arms combined). Within each country: ~800 control, ~600 per treatment arm.

On the standardized primary indices, pooled across countries with country fixed effects (α = 0.05 two-sided, power 0.80, control-group SD = 1): • Pooled treatment vs. control (N = 9,600 vs 6,400): ~0.045 SD • Single treatment arm vs. control (N = 4,800 vs 6,400): ~0.054 SD • Conditions vs. Effectiveness (N = 4,800 vs 4,800): ~0.057 SD • Pooled treatment × median-split interaction: ~0.090 SD • Single-arm × median-split interaction: ~0.107 SD For a single binary outcome (5-point Likert collapsed to support/strongly support, assuming 50% control agreement): ~6.4 pp pooled treatment and ~7.6 pp single arm within a single country. These MDEs are conservative — they ignore variance reduction from country fixed effects and covariates, and treat indices as having unit residual SD.