In This Article

  1. Introduction
  2. Defining Objectives
  3. Question Types
  4. Writing Good Questions
  5. Sampling Methods
  6. Common Biases
  7. Conclusion

Introduction

Survey design is the process of creating questionnaires to collect data from respondents. A well-designed survey produces reliable, valid data that can inform business decisions. Poor design leads to biased or unusable results.

Effective surveys balance the need for comprehensive data with respondent burden—surveys that are too long or complex will have low completion rates and poor quality responses.

Defining Research Objectives

Before designing questions, clearly define:

  • Research questions: What do you want to learn?
  • Target population: Who are you studying?
  • How data will be used: What decisions will it inform?
  • Key variables: What must you measure?
Tip: For every question you include, ask: "What will I do with this data?" If you can't answer, don't include it.

Question Types

TypeDescriptionBest For
Multiple ChoiceSelect one from optionsCategorical data, easy analysis
CheckboxSelect all that applyMultiple selections allowed
Likert ScaleAgreement scale (1-5, 1-7)Attitudes, satisfaction
Rating ScaleNumeric scale (1-10)Intensity, likelihood
Open-EndedFree text responseExploratory, detailed feedback
RankingOrder items by preferenceRelative preferences
MatrixGrid of related questionsMultiple items, same scale

Likert Scale Example

"I am satisfied with the customer service I received."

  • 1 - Strongly Disagree
  • 2 - Disagree
  • 3 - Neutral
  • 4 - Agree
  • 5 - Strongly Agree

Writing Good Questions

Do's

  • Use simple, clear language
  • Ask one thing at a time
  • Provide balanced response options
  • Include "Don't know" or "Not applicable" when relevant
  • Use consistent scales throughout
  • Order questions logically

Don'ts

  • Double-barreled questions: "How satisfied are you with price and quality?" (asks two things)
  • Leading questions: "Don't you agree that our service is excellent?"
  • Loaded questions: Questions with emotionally charged words
  • Jargon: Technical terms respondents may not understand
  • Absolute words: "always," "never" (rarely accurate)

Example: Improving a Question

Bad: "How satisfied are you with our fast and reliable delivery?"

Good: Split into two questions:

  • "How satisfied are you with our delivery speed?"
  • "How satisfied are you with our delivery reliability?"

Sampling Methods

MethodDescriptionPros/Cons
Simple RandomEvery member has equal chanceUnbiased; may miss subgroups
StratifiedDivide into groups, sample eachEnsures representation
ClusterSample groups, survey all in groupCost-effective for geography
ConvenienceSample whoever is availableEasy; potentially biased
QuotaFill quotas for each segmentEnsures diversity; not random

Sample Size Considerations

  • Larger samples = more precision
  • Consider confidence level needed (typically 95%)
  • Consider margin of error acceptable (typically ±3-5%)
  • Account for expected response rate

Common Biases to Avoid

  • Selection bias: Sample doesn't represent population
  • Non-response bias: Those who respond differ from non-responders
  • Social desirability bias: Respondents give "acceptable" answers
  • Acquiescence bias: Tendency to agree with statements
  • Order effects: Earlier questions influence later answers
  • Primacy/Recency: First or last options chosen more often

Conclusion

Key Takeaways

  • Start with clear research objectives
  • Choose question types appropriate for your data needs
  • Write clear, unbiased questions—one concept per question
  • Avoid leading, double-barreled, and loaded questions
  • Select sampling method based on resources and needs
  • Be aware of common biases and design to minimize them
  • Pre-test your survey before full deployment