Abstract
Economists are increasingly interested in how to conduct experiments with teenagers. This paper evaluates whether different methodological factors impact the answers of teenagers to standard experimental tasks on measuring time preferences, risk preferences and cognitive abilities, among others. Results show: (i) the recruitment process matters. When the school includes the experiment as an institutional activity, the dropout rate is observed to be significantly reduced; (ii) hypothetical payments elicit similar results to monetary payments; (iii) adding visual elements to the experiment's interface improves the quality of answers; and iv) although the type of device has no effect on the results, who administers the experiment does have an effect. We conclude by giving three suggestions to researchers interested in conducting experiments with teenagers: first, run the experiment as a school-programmed activity; second, the use of actual payments is not necessary, which increases the cost and complicates the recruitment; and third, integrate visual components to the task.