A classroom experiment on the causes and forms of bounded rationality in individual choice

Abstract

Bounded rationality is a key concept with implications across all areas of economics. To help students better understand the nature, causes and forms of bounded rationality in individual choice, the authors present a flexible classroom experiment. Beyond providing students with some first-hand evidence of bounded rationality, the experiment shows how a range of factors can prompt bounded rationality and illustrates how it can manifest itself in the form of different behaviors. The experiment can be conducted with any number of students and takes approximately 30 minutes to conduct, with a further 10 to 20 minutes to reflect and debrief.

Keywords:

• Bounded rationality
• choice
• classroom experiment
• heuristics
• information search

JEL CLASSIFICATION:

• A22
• D90
• D80

Notes

1 For key references, see Simon (1955), Payne, Bettman, and Johnson (1993), Gilovich, Griffin, and Kahneman (2002), and Sims (2003). For useful surveys and popular treatments, see Conlisk (1996), DellaVigna (2009), Kahneman (2011) and Thaler (2015).

2 For examples about such effects, see Meredith and Salant (2013) and Lacetera, Pope, and Sydnor (2012), respectively.

3 For more on the value of using clickers in economics classes, see Salemi (2009).

4 Under a paper-based approach, instructors would provide each student with an answer sheet to record their responses at the start of the experiment, and tailor the experiment’s instructions accordingly. To make the processing of such answer sheets as efficient as possible after the session, instructors could use optical mark recognition (OMR) as commonly used for multiple choice tests and surveys.

5 For more on the use of monetary incentives in experiments, see Thaler (2015, 47–49) and Loewenstein (1999). See Rousu et al. (2015) for monetary incentives in classroom experiments.

6 Strictly speaking, digit biases typically refer to calculations in numeric format, rather than in the written format used within our experiment. It would be interesting to compare how students behave across the different formats in future research.

7 These results also remain qualitatively robust if one calculates the correlations across all correct and incorrect options rather than across only the incorrect options. However, the presented approach has the advantage of not being unduly influenced by the format of the correct options.