The Foundation of Survey-Based Research
A poorly designed questionnaire produces biased, unreliable, or uninterpretable data — no matter how sophisticated your statistical analysis. This guide covers everything you need to build a valid and reliable measurement instrument for your thesis.
Use Existing Scales Where Possible
Whenever available, use previously validated scales. This saves time, strengthens your methodological foundation, and allows comparison with existing literature. Only develop a new instrument if no suitable existing scale covers your construct.
Designing Likert Items
- Use 5-point (Strongly Disagree to Strongly Agree) or 7-point formats.
- Each item should measure a single concept — avoid double-barreled statements (e.g., "fast and reliable service").
- Include reverse-scored items to reduce acquiescence bias.
- Each subscale should have at least 3–4 items.
- Use clear, simple language appropriate to your target population.
Content Validity
Content validity ensures the scale adequately covers the construct domain. Collect ratings from a minimum of 5–7 subject-matter experts. Calculate the Content Validity Index (CVI); a value ≥0.78 per item is required.
Pilot Testing
Administer the questionnaire to 30–50 participants before your main data collection. During the pilot: identify ambiguous or poorly understood items, check Cronbach's Alpha, and review item-total correlations. Revise accordingly.
Ethical Approval
Research involving human participants requires ethics committee approval. Submit your questionnaire and informed consent form along with your research protocol. Participants must provide written or electronic consent before completing the survey.
Reverse Coding in SPSS
Reverse-scored items must be recoded before analysis: Transform → Recode Into Different Variables. For a 5-point scale: 1↔5, 2↔4, 3→3. Failure to recode will invalidate reliability and factor analysis results.