Tips for Conducting a Longitudinal Study

Longitudinal studies sample the same group of participants repeatedly over time. The MTurk Toolkit makes setting up and conducting longitudinal studies easy. To make your study as successful as possible, follow these tips.

CloudResearch's Mechanical Turk Toolkit makes it easy to set up longitudinal studies by allowing you to define eligibility criteria and email workers when successive waves of the study are ready. However, to conduct a successful longitudinal study you want to minimize attrition and make sure you know how to link worker responses across multiple waves of data collection.

Minimizing Attrition

There are several ways to minimize attrition in a longitudinal study on Mechanical Turk. 

        1. Let workers know up front that the study is longitudinal.
          You can use the custom instructions box when describing your HIT to tell workers the study requires multiple rounds of data collection. Within the survey, ask workers again if they would be interested in doing the rest of the project. Workers who are unsure of their commitment to participating in follow-up study are more likely to drop out of the study.
        2. Pay more across time points or offer a bonus for completing all time points.
          By using an incremental pay schedule or offering a bonus at the end, you can increase the incentive for workers to continue being part of your project. Be sure to let workers know about the pay schedule up front using the custom instructions box.
        3. Pay above average wages.
          Workers are more likely to return for successive ways of data collection not just if there is a bonus or increasing rewards to be earned, but also when the study's overall pay is above average. Average wages are close to minimum wage, so most longitudinal studies should pay above minimum wage.
        4. Email workers when each new study launches.
          Emailing workers lets them know when your study is available and provides them with a link directly to the study.

Linking Worker Responses Across Time Points

When you conduct a longitudinal study on Mechanical Turk, the easiest way to link participant responses across time is using Worker IDs. If you want to offer workers an extra layer of privacy, you can use Anonymized Worker IDs from CloudResearch. 


You can ask workers to enter their Worker ID in each survey or you can automatically have each Worker's ID added to your data file using embedded data within your survey creation platform. After the study ends, you can use the Worker ID to link participants responses across time.