Predicting, Preventing, and Projecting Attrition in a Longitudinal and Crowdsource-Recruited Sample

• In Event: Advances in Experimental Criminology

Abstract

Reducing attrition is imperative to improving the validity of any longitudinal research study. Researchers have identified and developed numerous tools and techniques to minimize attrition although it is unclear if the same methods are effective with crowdsource-recruited (e.g., Amazon Mechanical Turk (Mturk), Prolific, and CloudResearch) samples. Using longitudinal data from Mturk we examine the magnitude and correlates of attrition. We also propose ways to prevent it, predict participant commitment, and put forward best practices for researchers utilizing crowdsource-recruited samples.

Authors

• Kassandra L. Cordero Treston, University of Central Florida

• Meghan Mitchell, University of North Dakota

• Erica Fissel, ICF

• Jacqueline Woerner, University of Central Florida