Inconsistent choices among adolescents in El Salvador

Abstract

This paper examines inconsistencies in the decision-making of a sample of 2,248 adolescents from El Salvador when completing two classic experiments: temporal discounting and risk preferences. Inconsistency in responses is a significant issue when collecting experimental data because it implies a loss of the sample, as these data come from subjects who do not respond to the task as they should and could indicate, for instance, a lack of understanding of the task. To mitigate this problem, we reduced the number of decisions, designed tasks with a strong visual component, and adapted them to the context with the assistance of a local pedagogical team. Despite these adaptations, we first observe participants' significant difficulties in avoiding errors such as multiple switching. Secondly, we investigate whether developmental factors (age), cognitive skills, and task repetition improve consistency. Lastly, we explore whether inconsistency in decision-making somehow shapes their immediate and long-term educational expectations.