Reproducibility package for Measuring Agency Through Psychological Constructs In Lower-Income Settings

Psychological constructs related to agency---such as the ability to set goals or feel in control---are increasingly recognized as important determinants of economic outcomes and well-being. Yet measures of these constructs either don't exist or almost exclusively demonstrate evidence of reliability and validity in Western, Educated, Industrial, Rich, and Democratic (WEIRD) contexts. This paper introduces four newly developed scales to assess goal-setting, locus of control, generalized livelihoods self-efficacy, and agricultural self-efficacy, tested through nationally representative and specialized surveys in Kenya, Malawi, Tanzania, Uganda, Benin, and Côte d'Ivoire. Reliability was evaluated using Cronbach's alpha, while validity was assessed through respondent understanding, expert opinions, correlations with development outcomes and demographics, and factor analysis. All scales demonstrate strong evidence of reliability and validity, though measurement properties are weaker for locus of control. We find no evidence of significant item order effects, while 5-point Likert response scales perform better than 3-point response scales. All four psychological constructs are associated with subjective well-being as well as socioeconomic outcomes, with generalized self-efficacy having the strongest relationship. Perhaps because of their lower average levels of agency, nearly all constructs matter more for women’s outcomes. We recommend these validated scales for use in impact evaluations and endline surveys assessing policy interventions targeting psychological well-being, economic livelihoods, and gender empowerment.