A Preliminary Study of Daily Sample Composition on Amazon Mechanical Turk

10 Pages Posted: 1 Mar 2015 Last revised: 12 Mar 2015

See all articles by Kiran Lakkaraju

Kiran Lakkaraju

Sandia National Laboratories (NM)

Date Written: March 11, 2015

Abstract

Amazon Mechanical Turk (AMT) has become a powerful tool for social scientists due to its inexpensiveness, ease of use, and ability to attract large numbers of workers. While the subject pool is diverse, there are numerous questions regarding the composition of the workers as a function of when the “Human Intelligence Task” (HIT) is posted. Given the “queue” nature of HITs and the disparity in geography of participants, it is natural to wonder whether HIT posting time/day can have an impact on the population that is sampled. We address this question using a panel survey on AMT and show (surprisingly) that except for gender, there is no statistically significant difference in terms of demographics characteristics as a function of HIT posting time.

Keywords: crowdsourcing, amazon mechanical turk

Suggested Citation

Lakkaraju, Kiran, A Preliminary Study of Daily Sample Composition on Amazon Mechanical Turk (March 11, 2015). Available at SSRN: https://ssrn.com/abstract=2560840 or http://dx.doi.org/10.2139/ssrn.2560840

Kiran Lakkaraju (Contact Author)

Sandia National Laboratories (NM) ( email )

POB 5800
Albuquerque, NM 87185
United States

Register to save articles to
your library

Register

Paper statistics

Downloads
52
Abstract Views
593
rank
382,135
PlumX Metrics