Key Issues when Designing a Research Questionnaire

A research questionnaire has to be both reliable and valid, and that starts with using instruments that have already been tested in credible, peer-reviewed sources. The practical route suggested is to pull questionnaire items from quality academic databases—such as Emerald, Sage, Springer, Taylor and Francis, Cambridge, Oxford Science, and similar repositories—then verify that the specific questionnaire has evidence of reliability and validity. Beyond published testing, the instrument should be reviewed by experts: a supervisor, peers, colleagues, or subject specialists who understand the topic area. That expert check matters because even a well-known questionnaire can fail if it doesn’t fit the study’s context.

Context fit is treated as a make-or-break requirement. The transcript gives concrete examples of mismatch: studying social responsibility in public sector universities but using a corporate social responsibility questionnaire that includes profit-making items—an economic emphasis that aligns better with private, profit-oriented organizations than with universities. Another example targets creativity in banks: asking branch-level employees about creativity when creativity is typically associated with senior management. Even if the wording is correct, the wrong target group or wrong organizational setting can undermine the meaning of responses and distort results.

Once an appropriate instrument is selected, item design becomes the next priority. For structural equation modeling workflows (including tools like SmartPLS, AMOS, LISREL, and Mplus), the guidance is to include at least four items per construct, with a preference for five or more, but not so many that the questionnaire becomes unwieldy. Items are described as indicators that measure an underlying construct (sometimes also called variables or questions). Because items can be removed during SEM estimation, having a small buffer—often five to six items per construct—is recommended. At the same time, the questionnaire should stay concise: avoid multi-page surveys (four to five pages) and aim for roughly a two-page format when feasible.

Conciseness is not just about length; it’s also about presentation. Formatting in a word document—using narrow margins, landscape layout, and readable fonts (for example, 10–11 instead of 12)—can allow many items to fit without sacrificing legibility. The transcript also stresses that item wording and response wording must be clean: correct grammar, punctuation, and understandable phrasing, with additional review by an English expert when needed.

Response options must match the item type. Yes/no questions and overlapping statements like “Do you like your job?” or “Do you love your job?” can be difficult to analyze in SEM, where metric or scaled responses are preferred. The recommended pattern is to use Likert-style agreement scales (e.g., strongly disagree to strongly agree) for statements such as “I love my job” or “I would like to switch my job.” Finally, the questionnaire must align with how constructs are modeled: decide whether a construct is single-dimensional or multi-dimensional based on the study’s conceptualization, and ensure the selected instrument matches that definition. The transcript closes by emphasizing that constructs should be defined and conceptualized before searching for or building questionnaire items, so there’s no mismatch between the construct’s intended scope and the measures used.