ABSTRACT
Amazon Mechanical Turk (MTurk) has increasingly attracted the attention of researchers as a convenient method to gather human subjects due to its efficiency, low cost, and ease of use in securing data samples. The current study provides a meta-analytic review across three separate MTurk samples. Participants were 1707 US citizens (774 males and 933 females, aged 18–77 years) recruited from Amazon’s MTurk system. Results indicated that across three MTurk samples, demographic characteristics of workers closely approximated the general US population on gender and race but differed on religious affiliation given the very high number of Atheists and Agnostics (38.3%). Across the three samples, significant mean level differences were observed for 14 of the 15 study scales; however, effect sizes were small (eta2 ranged from .01 to .04). Results suggested that individuals working in the fields of personality and spirituality need to be aware that using an MTurk sample may introduce sample bias in their data.
Acknowledgements
The authors would like to thank Rose Piedmont and Drs Martin F. Sherman and Joanna I. Hayward for their comments on an earlier version of this paper.
Disclosure statement
Ralph L. Piedmont is the author and publisher of the ASPIRES. He receives royalties from its sale.
ORCID
Ralph L. Piedmont http://orcid.org/0000-0002-0482-6761