Unemployed and Turking: Is Amazon Mechanical Turk a Viable Source for Unemployment Research?

Abstract

Organizational research conducted using data obtained from Amazon’s Mechanical Turk (MTurk) has proliferated in recent years. However, there are critical concerns over the efficacy of these data for employment research. The purpose of this study was, alternatively, to assess the efficacy of MTurk data for unemployment research. We did so through four distinct analyses in which we sought to replicate existing findings that have been established using unemployment field data. Findings indicate unemployed MTurk sample demographics closely reflect US national unemployed population rates, with the exception of the MTurk sample being substantially younger, on average. Moreover, analyses aimed at replicating the effects of latent resources (self- esteem, social support, and meaning in life) on well-being among the unemployed MTurk sample were highly successful. However, analyses aimed at replicating effects requiring a comparison of unemployed to employed MTurk respondents were less successful. Coupled with evidence of differences between the employed MTurk sample and US national employment population rates on age and marital status, study implications may say as much about the potential (lack of) generalizability of findings from MTurk data for employment research as they do about the potential generalizability of findings from MTurk data for unemployment research.