Preferences of Crowdworkers

Abstract

This study examines the revealed and stated preferences driving crowdworkers’ engagement in crowdsourcing activities on Amazon Mechanical Turk (MTurk). Most existing studies on workers’ motivations on MTurk are based on self-reports and hence measure stated preferences. The results of these studies are therefore susceptible to social desirability bias. We let crowdworkers design their own task using an incentivized partial-profile conjoint analysis, which reduces the effect of social desirability and identifies revealed preferences. We also investigate potential differences between stated and revealed preferences among crowdworkers. Finally, we study potential differences in crowdworkers’ motivations in high- and low-stakes environments. We find that stated and revealed preferences of crowdworkers are inconsistent: in a field experiment, crowdworkers indicate that intrinsic motivation is the most important factor in working on MTurk, while in a survey they state that extrinsic motivation factors are decisive. This finding holds equally for crowdworkers in low-stakes and high-stakes environments. Our data also shows that crowdworkers in a high-stakes environment are more strongly motivated by intrinsically motivating job characteristics compared to workers in a low-stakes environment.

Keywords: Platform Economy, Crowdsourcing, Crowdwork, Revealed Preferences, Motivation, Social Desirability

JEL Classification: J4, M52, O3