ReDem’s advanced quality checks work best when your questionnaire is strategically structured. Below are best practices for designing surveys that unlock the full power of ReDem’s quality scores.

​
Open-Ended Questions

Open-ended questions are highly effective for quality assurance and fraud detection. We recommend including at least two mandatory open-ended questions in every survey.

🤔 Why they matter:

  • Mandatory answers filter out fraudsters: Fraudsters and disengaged respondents often skip or give meaningless answers to open-ended questions. Making them mandatory helps eliminate them early.
  • AI-evaluable content: Avoid generic questions like “Anything else you’d like to share?” Instead, ask questions that provide context, enabling ReDem’s AI to evaluate response quality.
  • Detecting AI-generated content: Emotional or opinion-based questions are especially helpful, as AI systems often struggle to convincingly express authentic emotion or personal views.
  • Strategic placement: Place one open-ended question at the beginning and one at the end. This helps evaluate attention and consistency across the interview.
  • Duplication detection: ReDem checks for identical or similar open-ended responses both within and across interviews to flag suspicious patterns.
Use our ChatGPT-powered OES Question Master to generate high-quality open-ended questions for your survey.

​
Grid-Questions

Grid or matrix-style questions allow for pattern recognition in click behavior, helping detect inattentive or fraudulent respondents.

👍🏼 Design recommendations:

  • Include at least 1 grid question with a minimum of 7 items and 4 or more response options (e.g., a Likert scale).
  • Ensure enough variability: A higher number of statements helps differentiate meaningful from arbitrary responses.
  • Balance options and items: More items = fewer needed options, but never fewer than 3.
  • Add inverted statements: Mix positive and negative phrasing to reveal inconsistent or bot answer patterns (e.g., straightlining or zigzagging).

​
Time Durations

ReDem’s Time Score is more accurate when detailed timing data is available.

🤨 What to capture:

  • Total Interview Duration (LOI): Used to flag unusually fast or slow respondents.
  • Section/Page-Specific Timing: Helps identify suspicious timing behavior, especially in longer or complex sections or pages.
  • Avoid brief sections: Skip timing for yes/no or demographic questions.
  • Account for interruptions: Be aware of idle time manipulation by bots trying to simulate realistic durations.

​
Trap Questions

Trap questions support the Coherence Score by identifying respondents who overclaim, contradict themselves, or fail attention checks.

✅ Best practices:

  • Use Sparingly: Limit to a maximum of two trap questions per survey to avoid participant frustration.
  • Design Subtly: Don’t make trap questions obvious. If respondents detect them, they may adjust their answers unnaturally.
  • Mix formats: Use variations like:
    • Repeat questions with rephrased wording
    • Instructions to select a specific option
    • Fake or nonsense options (e.g., fictional brands)
  • Design for contradiction detection: Structure your questionnaire in a way that allows contradictions to emerge. This helps ReDem’s Coherence Score detect dishonest or inattentive respondents more effectively.
  • Don’t rely on traps alone: Trap questions should complement, not replace, broader ReDem checks. An isolated error shouldn’t unfairly disqualify high-quality respondents.