Quality Control Questions in Surveys

Maintaining high-quality data in your research is critical. Within Glass, you have multiple tools to help ensure respondents are attentive, thoughtful, and providing meaningful answers.

This page covers the different types of quality control (QC) checks you can include, how and when to use them, and best practices for reviewing data without disrupting the respondent experience.

Even in well-designed surveys, issues like speeding, satisficing, or careless responses can occur. Quality control (QC) questions help you:

• Identify respondents who aren’t paying attention

• Flag or remove low-quality responses before analysis

• Ensure your insights are accurate, reliable, and trustworthy

We recommend including at least one to two open-ended questions in every survey. Open-ended responses show how engaged a respondent is with the content. Irrelevant, nonsensical, or extremely short answers may indicate low-quality participation.

Example: “In a few sentences, please explain why you chose your answer.”

Open-ended questions allow you to review written responses during fieldwork or analysis and identify respondents who may need to be removed or flagged.

Pro Tip: Review open-end responses early in fielding to catch quality issues before too many completes are collected.

Trap questions are designed to confirm that respondents are reading and following instructions. These are typically multiple-choice and come in two main types:

• Instructional Trap “To show that you are paying attention, please select the option ‘Strongly agree.’”

• Logical Trap “Which of the following is not a color? (Blue, Green, Elephant, Yellow)”

Trap questions can either:

• Use QUOTAS to terminate respondents immediately who fail the check (for high-incidence studies), or

• Flag responses on the backend for data cleaning and quality scoring later.

Pro Tip: Avoid overusing trap questions so you don’t annoy and lose real respondents – one per 5–10 minutes of survey length is ideal.

You have two main ways to apply QC outcomes in your dataset:

1. Immediate Termination Automatically remove respondents who fail a trap question.

• Best for high-incidence or large-sample surveys.

• Keeps data clean during fielding.

• NOTE: This must happen within the first 5-10 questions. Late terminations will result in your survey being flagged and potentially deactivated due to negative respondent experience. 

2. Backend Flagging (Quality Score)
Instead of terminating respondents, store their QC performance as a variable.

• Best for harder-to-source audiences or longitudinal work.

• Allows flexibility to exclude or weight respondents later in analysis.
  

• Include 1–2 open-ended questions in every survey.

• Add at least one trap question for surveys longer than 5 minutes.

• Avoid overuse – too many QC checks can frustrate legitimate respondents.

• Keep wording clear, simple, and professional.

• Always review QC outcomes during your data cleaning step (see Raw Data Export Guide)