Can teaching calculation and estimation strategies improve financial decision-making? The role of emotions and deliberation [Author Accepted Manuscript]

Financial decision-making often requires numerical reasoning. For example, monetary lotteries—a common paradigm in economic psychology—demand evaluation of payoffs and probabilities. To make choices consistent with normative standards such as Expected Value (EV), individuals must perform accurate calculations. However, limited numerical skills often hinder EV-consistent choices and reduce potential earnings. This study tested whether brief interventions could improve financial decision-making and subsequent emotional responses, defined as lower negative emotions and/or higher positive emotions. Participants were randomly assigned to one of three conditions: Control, Calculation, or Calculation & Estimation. In the Control condition, participants made choices in ten monetary lotteries. In the Calculation condition, they received instructions on computing EV prior to the task. In the Calculation & Estimation condition, participants were additionally trained in estimation strategies designed to simplify computations. Although overall group differences did not reach statistical significance (p < .05), planned comparisons showed that participants in the Calculation condition made significantly more EV-consistent choices than those in the Control condition. Exploratory analyses further revealed that deliberation time mediated the link between condition and EV-consistent choices: participants in both intervention groups spent more time deliberating, which in turn contributed to improved financial outcomes. Individual differences also played a role: objective numeracy predicted more EV-consistent choices, both numeracy and math attitudes predicted deliberation time, and trait math anxiety predicted stronger negative emotional reactions. In sum, the findings suggest that teaching EV calculation can promote deliberation and support more rational economic choices, even through brief interventions. This work was financially supported by the National Science Centre, Poland (grant number 2018/31/D/HS6/02899). reviewed acceptedVersion