Getting Started with Surveys on MTurk

Amazon Mechanical Turk
Jun 6 · Unlisted

Get fast, and more reliable responses to your survey or study with MTurk

MTurk is a great way to get fast, and more reliable responses to your survey or study and many sectors of the economy are able to leverage MTurk survey functionality. The private sector uses MTurk to gain consumer insights to guide business decisions or to conduct market research and customer preference surveys. Academic Survey Requesters from fields such as and psychology, marketing, political science, economics and sociology use MTurk for a wide variety of research studies. MTurk continues to be the biggest crowdsource marketplace worldwide and we want to make sure our Requesters have the information they need to get higher quality results from their MTurk surveys.

Linked Surveys: While you can create surveys directly in MTurk, many Requesters prefer to leverage third party survey tools to further streamline the process of creating and deploying their survey through MTurk. Learn more about deploying linked surveys in MTurk here. Many Requesters use Qualtrics, but options such as SurveyMonkey, Google Forms, and SurveyGizmo are also used frequently. Check the MTurk Blog for tutorials on using each of these survey tools. Each of these tools requires a few extra steps to ensure that your survey is correctly published on MTurk and that you can identify which Workers participated. Your use of these third party survey tools is at your own discretion. MTurk does not endorse the use of any specific third party survey tool and does not guarantee results and your satisfaction with these third party survey tool providers.

If you’re using the Survey Link template and hosting your survey on another website such as Qualtrics, SurveyMonkey, or your own website, you’ll need a way to match responses to the Worker information in MTurk. You can do this by providing Workers with a unique completion code and by collecting a Worker’s ID in the survey itself.

Completion Codes: If your survey tool allows you to provide a unique completion code when a survey is completed, you can then direct Workers to provide it in the MTurk task when they have finished the survey. You can then link the survey response to the Worker in your results, see the instructions for Qualtrics as an example. If you supply a nine digit code and a Worker provides a seven digit completion code, you may want to review that Worker’s responses. Additionally, varying the number of digits in your completion codes between surveys can make it easy to segment your survey results.

Worker IDs: Including a question in your survey that asks Workers for their Worker ID as part of the survey also enables you to identify and remove duplicate responses. Workers can generally find this on their Dashboard or in the upper left corner of the Worker website. When you get your results you’ll be able to match the Worker IDs.

Edit Survey Links: Take time to review the process for editing survey links, to avoid publishing the wrong survey to Workers. You will want to ensure that you follow best practices, take measures to prevent exposing Worker data, do not collect Worker Emails, and use Qualifications to select Workers for your Survey.

Qualifications: You may want to leverage Premium, System and/or Masters Qualifications to select Workers for your study. Requesters consistently tell us that using Qualifications is the most effective strategy for optimizing the quality of results on Amazon MTurk. It is highly recommended that you use Qualifications for surveys on MTurk. Using Qualifications allows you to select or create specific Worker eligibility requirements for your projects. This enables you to direct projects to specific groups of Workers on MTurk. When you select a Qualification, it means that you can include or exclude Workers who meet your Qualification standard and control which workers can contribute to your project. Read more about the importance of selecting Qualifications on the MTurk Blog. You can also create Custom Qualifications to include or exclude Workers across multiple survey batches or for longitudinal research.

Test: Finally, you can test your design before you publish your survey using our Requester Sandbox. Making sure the task functions correctly in advance will help you save time. Requesters have the highest degree of success with their surveys if they test surveys to a small number of Workers first. This will give you a chance to take a look at the responses and make sure you’re getting the results you’re looking for and validate that your questions are displaying as intended and the embedded completion code is populating as expected.

Please reach out to support at mturk-requester-support@amazon.com

Unlisted

    Amazon Mechanical Turk

    Written by