Too Good to Be True: Bots and Bad Data From Mechanical Turk
Margaret Webb, June P. Tangney
Abstract
Psychology is moving increasingly toward digital sources of data, with Amazon’s Mechanical Turk (MTurk) at the forefront of that charge. In 2015, up to an estimated 45% of articles published in the top behavioral and social science journals included at least one study conducted on MTurk. In this article, I summarize my own experience with MTurk and how I deduced that my sample was—at best—only 2.6% valid, by my estimate. I share these results as a warning and call for caution. Recently, I conducted an online study via Amazon’s MTurk, eager and excited to collect my own data for the first time as a doctoral student. What resulted has prompted me to write this as a warning: it is indeed too good to be true. This is a summary of how I determined that, at best, I had gathered valid data from 14 human beings—2.6% of my participant sample ( N = 529).