Chapter 6. Operationalisation, Constructs and Questionnaire Design
6.1 From abstract concept to variable
Operationalisation defines how a theoretical construct becomes observable data. Specify the conceptual definition, dimensions, indicators, response format, scoring, permissible range, handling of reverse-coded items, and interpretation of higher or lower scores. A construct may be reflective (indicators are understood as manifestations of an underlying trait) or formative (indicators jointly define the construct). The distinction affects measurement evaluation.
6.2 Use existing instruments carefully
When possible, select a well-supported instrument that fits the construct and population. Verify the original source, item wording, scoring instructions, validation evidence, population, language and permissions. An instrument validated in one country, occupation or language is not automatically valid in another. Translation may require forward translation, independent back-translation or committee review, cognitive interviews and pilot testing. Do not reproduce copyrighted scales without permission when permission is required.
6.3 Questionnaire structure
- Participant information and consent, presented before questions.
- Eligibility screening where needed.
- Clear, neutral instructions and an estimated completion time.
- Main construct measures grouped in a logical sequence.
- Demographic or organisational questions that are justified by the analysis plan.
- Optional final comment, if useful and ethically approved.
- Thank-you statement and contact route for questions or withdrawal, consistent with the approved protocol.
6.4 Item-writing principles
- Ask one question at a time; avoid double-barrelled items.
- Avoid leading, loaded or judgemental wording.
- Use a clear time frame where recall is required.
- Ensure response options are mutually exclusive and collectively adequate.
- Include “not applicable” or “prefer not to say” only when conceptually and analytically appropriate.
- Avoid unnecessary reverse-worded items; they can confuse respondents and create method effects.
- Use language appropriate to the target population and test comprehension.
- Do not ask for identifiable information unless it is essential and approved.
6.5 Likert-type responses and scoring
A single Likert-type item is ordinal: categories have order but equal spacing cannot be assumed automatically. A multi-item scale is often analysed as approximately continuous when its construction, distribution, reliability and use justify that choice. Report how item responses are combined, the minimum number of answered items needed for a score, and whether higher scores mean more or less of the construct. Reverse-code only items designated in the instrument's scoring instructions.
For a 1–5 item reverse-coded on a 1–5 scale, the transformed value is 6 minus the original response. For a 1–7 scale, it is 8 minus the original response. Check the permitted range and missing-value codes before recoding.
6.6 Pilot and cognitive testing
A pilot tests logistics, clarity, completion time, response patterns and data workflow; it is not necessarily a definitive validation study. Cognitive interviews ask a small number of target participants how they understand questions and choose responses. Revise ambiguous wording before the main study. If pilot participants are to be included in the main dataset, define this in advance and ensure the instrument and protocol did not materially change.
6.7 Common method bias
When the same respondent reports predictors and outcomes in the same questionnaire at the same time, shared method effects may inflate or distort associations. Procedural safeguards include separating predictor and outcome sections, assuring confidentiality, using clear non-evaluative wording, varying measurement sources when feasible, and reducing evaluation apprehension. A single statistical test cannot prove that common method bias is absent.
6.8 Instrument documentation
Keep a measurement table listing each construct, operational definition, scale source, item count, response anchors, scoring rule, prior reliability/validity evidence, adaptation, permission status and planned evaluation in the current sample. Put the complete questionnaire in an appendix when permitted, and cite the original instrument source.