Using Amazon Mechanical Turk for Consumer Behaviour Research

Abstract

Jade Charette-Côté was a graduate student in a North American business school who recently started working on her dissertation. Charette-Côté's objective was to replicate the main finding of previous research upon which her dissertation was based using an experimental study. Following her supervisor's suggestion, she prepared an experimental study using the online survey software Qualtrics and ran the experiment on Amazon Mechanical Turk (MTurk). However, inferential statistical tests revealed that she was unable to replicate the expected effect. Disappointed with the results, Charette-Côté took a careful look at the data and realized some unexpected patterns, which may have led to the failed replication attempt. Indeed, these patterns made her question the quality of the data, but she was not sure what the main issues were and therefore did not know how to address them in future experiments. Charette-Côté's supervisor asked her to find best practices for collecting data using MTurk so that she could identify the issues that needed to be fixed and learn how to fix them before running a follow-up survey. She wondered how she could redesign the survey to improve reliability. Where should she start?

How to cite

Related cases