Abstract
Surveys are indispensable in population and public-health research for measuring knowledge, perceptions, and behaviours not captured in administrative data, especially in cardiovascular (CVS) research (e.g., diet, activity, medication adherence). However, poorly constructed questionnaires introduce systematic measurement errors and biases. This paper synthesizes best practices for designing reliable and valid self-report instruments, emphasizing a respondent-centred, systematic approach.
We present evidence-based strategies integrating (1) item generation through operationalizing theoretical constructs into measurable indicators, (2) tailored item formulation addressing the four cognitive stages of survey response (comprehension, retrieval, judgment, and response formatting) to minimize measurement error, (3) questionnaire composition strategies for logical structure, flow, and managing contextual and order effects, and (4) systematic questionnaire testing and refinement through pre-testing (qualitative item refinement, individual item quality assessment, and holistic questionnaire appraisal), pilot testing with psychometric validation (reliability, validity, and item discrimination), and evidence-based item reduction and finalization. Practical tools (e.g., BRUSO checklist, cognitive interviewing, factor analysis) are highlighted throughout.
Applying these strategies yields questionnaires that are more interpretable, less biased, and better suited to intended analyses. Key operational recommendations include defining the constructs and using appropriate question items and types to measure them, prioritizing brevity and clarity, preferring closed items with exhaustive response categories, allowing for the capture of unforeseen responses, conducting pre-testing and pilot testing, and thoroughly evaluating the questionnaire.
High-quality questionnaire development is a resource-intensive, iterative process that combines qualitative insight and quantitative rigour. Adoption of the outlined evidence-based steps substantially reduces measurement error and enhances the utility of survey data for CVS and broader public-health research.