Co-worker transparency in a microtask marketplace

  • Authors:

  • Peter Kinnaird;Laura Dabbish;Sara Kiesler;Haakon Faste

  • Affiliations:

  • Carnegie Mellon University, Pittsburgh, Pennsylvania, USA;Carnegie Mellon University, Pittsburgh, Pennsylvania, USA;Carnegie Mellon University, Pittsburgh, Pennsylvania, USA;Carnegie Mellon University, Pittsburgh, Pennsylvania, USA

  • Venue:
  • Proceedings of the 2013 conference on Computer supported cooperative work
  • Year:
  • 2013

Quantified Score

Hi-index 0.00

Visualization

Abstract

Workers in microtask work environments such as Mechanical Turk typically do not know if or how they fit into a workflow. The research question we posed here was whether displaying information about the number of other workers doing the same task would motivate better or poorer work quality. In experiment 1, we varied the information about co-workers presented to the worker and the number of his or her co-workers: "you" or "you alone" are doing a task, or "you" plus 5, 15, or 50 co-workers. We compared these conditions with a no-social information control. In experiment 2, we crossed the number of co-workers (5 vs. 50) with the type of incentive (individual or group). Results show that visual presentations of co-workers changed workers' perceptions of co-workers, and that the more co-workers participants perceived, the lower their work quality. We suggest future work to determine the kinds of co-worker information that will reduce or increase work quality in microtask settings.