Design a Questionnaire You Can Actually Analyse

# How to Design a Questionnaire Students Can Actually Analyse

Most student questionnaires are not ruined by bad questions. They are ruined by a missing plan for what happens to the answers once they are collected. This guide works backwards from the analysis: you decide how each answer will be entered, coded, and reported before anyone fills in a form. The steps below are written for final-year undergraduate and postgraduate students in Uganda running a first independent study, but the logic applies to any small survey project.

A note on evidence. The sources cited here are questionnaire design and validation studies from outside East Africa [S4][S6][S7]. They support general points about construct definition, validated instruments, and multidimensional scales. They do not describe Ugandan fieldwork conditions, and none of them gives numeric rules for item counts or scale lengths. Where this guide states a procedure rather than a finding, it is reasoning from general methodological principles, not a cited result. I mark the difference as I go.

1. Why Most Student Questionnaires Fail at the Analysis Stage

The failure is usually not visible in the questionnaire. It becomes visible when you open the spreadsheet.

Four defects surface after fieldwork. First, double-barrelled items — questions that ask about two things at once — cannot be split into one variable, so you have no clean column to analyse. Second, response options that overlap let two respondents with the same view choose different boxes, which makes their answers incomparable. Third, constructs named in your objectives are never operationalised into measurable items, so an objective sits in Chapter One with nothing in Chapter Four to answer it. Fourth, open-ended questions are included with no coding frame, producing paragraphs of text that cannot be counted or compared.

None of these defects is obvious from reading the questionnaire aloud. A questionnaire can look professional, be approved by a supervisor, and still be unanalysable. The cost is paid in a weakened results chapter, and redoing fieldwork is usually impossible within a student timeline.

The fix fits in one sentence: decide the analysis first, then write items that produce exactly the data that analysis needs. This is a principle of reasoning, not a finding from the supplied sources. The closest supplied support is the general idea that questionnaire creation can follow a systematic framework rather than an ad hoc process [S6].

For the wider question of how the questionnaire fits your overall design, see How to Choose a Research Design for a Student Project.

2. Step 1 — Restate Your Objectives as Constructs You Can Measure

An objective is a sentence. A construct is something you can measure. Your first job is to convert one into the other.

From objective to construct. Take an objective such as "to assess student satisfaction with mathematics study materials." The construct inside it — satisfaction with study materials — has to be defined before it can be measured. A published study that designed and validated a questionnaire for exactly this purpose shows the step in practice: the researchers had to specify what satisfaction with materials meant before they could write items for it [S4]. One objective may need two or three constructs. Write them as a numbered list so you can refer back to them.

Write a working definition for each construct. One or two sentences, in your own words, with a source if you are borrowing a definition. State the boundary explicitly: what is inside the construct and what you are deliberately leaving outside it. A construct with no boundary will absorb every question you can think of.

Decide the direction of scoring before writing items. For each construct, write down what a high score means and what a low score means. This one note prevents reversed items from being scored the wrong way later, when you are tired and entering data at midnight.

Check the constructs against your research questions. Every research question should map to at least one construct, and every construct should serve a research question. Delete constructs that serve neither. This alignment step is covered in more detail in Research Objectives and Questions That Match.

3. Step 2 — Choose the Response Format from the Analysis You Will Run

The response format is an analysis decision, not a design preference. Choose it by asking what treatment each item will receive.

Match format to intended treatment. If you will compute means or run comparisons between groups, you need ordered response options with roughly even spacing between them. If you will report frequencies only, simpler categorical options are enough. If you will analyse text thematically, say so now and plan the coding frame described in Step 4.

Keep one format per block. Changing scale length in the middle of a questionnaire confuses respondents and complicates scoring. Multidimensional instruments often use several subscales; if you do this, label each block clearly so respondents know the frame has changed. A validated multidimensional questionnaire on digital fatigue among university students is an example of a multi-subscale instrument in practice [S7].

Decide whether to borrow a validated instrument. A published, validated questionnaire gives you items and scoring guidance that have already been tested elsewhere [S4][S6][S7]. Borrowing is not copying, but you still need permission or correct citation, and you must check that the items fit your population and context. If you adapt an instrument, record exactly what you changed and why. The supplied sources do not state this record-keeping rule; it is standard methodological practice rather than a cited finding. One supplied source does support the general principle that instruments should fit the needs of the target group [S8].

The inherited-template problem. If your department circulates a questionnaire template, audit it against your constructs rather than adopting it whole. Keep the items that map to your constructs. Rewrite or drop the rest. Note the changes for your methodology chapter. No supplied source addresses departmental templates; this is reasoning from the principle that items must serve constructs.

For how to justify these choices in writing, see How to Write a Research Proposal Methodology Section: Justify Every Choice.

4. Step 3 — Write Items That Produce One Variable Each

Each item should generate exactly one column of data with one unambiguous meaning.

One idea per item. Split any item containing "and" or "or" into separate items, unless the two parts are genuinely inseparable. The test: can a respondent reasonably agree with half the item and disagree with the other half? If yes, split it.

Non-overlapping, exhaustive response options. Options must not overlap, and they should cover the realistic range of answers. Include "prefer not to say" or "not applicable" only where it is analytically meaningful, and decide now how each will be coded. An option you cannot code is an option you cannot report.

Plain wording for a multilingual respondent group. Use short sentences, one clause where possible, and avoid idioms and double negatives. Uganda's language situation is genuinely complex: English is the official language and the main medium of instruction at university level, but for many students it is a second or third language, alongside a first language and often Luganda, Runyankole, Acholi, Ateso or another widely spoken language depending on the region. A questionnaire written in English may therefore be read by respondents whose command of English varies. If you plan to translate items, decide before the pilot which language versions you will produce and how you will check that they mean the same thing. None of the supplied sources covers translation procedure; treat this as standard practice.

Keep demographic items purposeful. Every demographic question should be one you will actually cross-tabulate or control for. If you cannot say how you will use it in analysis, cut it. This is the same discipline as the coding frame in the next step.

For sentence-level clarity in item wording, see Clearer Academic Writing Without Losing Rigour.

5. Step 4 — Draft the Coding Frame Before the Pilot

Build the spreadsheet structure and the code list now, so the pilot tests the coding plan as well as the questions.

Build the data template first. One row per respondent, one column per item, with a short variable name at the top of each column. Write the variable name as the item's meaning, not its number. "Satisfaction_materials" survives six months; "Q7" does not.

Write the code list for every closed item. For each item, record what number each response option becomes. Decide and record how missing answers, "not applicable," and "prefer not to say" are coded, and keep them distinct from each other. Collapsing these three into one code destroys information you may need later.

Write the coding frame for every open-ended item. Either drop open-ended items or write the categories you expect to use, with room to add categories during coding. State who will code the text and how you will handle disagreement if there is more than one coder.

Note the planned analysis next to each variable. Keep a short column in your planning document: frequency, mean, cross-tabulation, or theme. This is the check that closes the loop. Any variable with no planned analysis is either redundant or a gap in your plan.

The supplied sources do not cover data templates, code lists, or missing-value coding. These are standard data management procedures. For the analysis end of this loop, see Interpreting a p Value Without Overclaiming.

6. Step 5 — Pilot the Questionnaire and the Coding Frame Together

The pilot tests two things: whether respondents understand the items, and whether your coding frame survives real answers.

Pilot with people similar to your sample. Draw a small number of respondents from the same population, not classmates who already know your topic. Ask them to note any item they had to read twice or interpret. If your sample will include students from more than one university or region, try to pilot with a mix, so that wording problems tied to one setting show up before fieldwork.

Code the pilot data immediately. Enter the pilot responses into your template and apply the coding frame. Any answer that does not fit a code is a defect in the frame or the item. Fix it now, while you can still change the questionnaire.

Revise and record. Keep a dated log of every change: item dropped, wording altered, code added. This log becomes the instrument-development paragraph in your methodology chapter.

Pilot respondents are not your sample. For how the two differ, see How to Choose a Sample Size for a Student Survey.

7. Step 6 — Document the Instrument for Your Supervisor and Your Reader

A short instrument specification lets your supervisor approve the design and lets your reader understand your results.

What the specification contains. Constructs and their definitions; items mapped to constructs; response formats; coding rules; planned analysis per variable. One or two pages is enough.

Where it goes in the dissertation. Summarise it in the methodology chapter and place the full questionnaire in an appendix. Cite any borrowed or adapted instrument correctly [S4][S6][S7]. The citation rule itself is not supplied by those sources; it is standard referencing practice. See APA 7 Referencing Explained.

What to say if you adapted an existing instrument. State the original source, what you changed, and why the change was necessary for your population. Do not claim the adapted version is validated unless you have validated it yourself. None of the supplied sources states this rule; it follows from the fact that validation is a property of a specific instrument in a specific population, which is why the validated instruments cited here report their own validation studies [S4][S7].

8. Common Questions About Questionnaire Design for Student Projects

"My supervisor gave me a questionnaire. Do I still need to do this?" Yes. Audit the inherited instrument against your constructs and coding frame, and document any changes. The audit is what makes the instrument yours.

"How many items per construct?" There is no universal number, and the supplied sources do not give one. They describe validated instruments [S4][S6][S7] but none states an item-count rule. Aim for enough items to cover the construct's boundary, and check coverage against your definition rather than a target count.

"Can I add open-ended questions?" Only if you have a coding frame and the time to apply it. Otherwise, convert them to closed items. An open-ended question with no coding plan produces text you cannot use.

"What if I have already collected data?" You can still build the coding frame retrospectively, but note the limitation plainly in your methodology chapter. The frame will fit the data less well than if you had designed it first.

For framing these limitations honestly, see How to Write a Research Problem Statement.

9. Checklist: What Must Be Finished Before Fieldwork Starts

Tick these off the week before data collection begins.

• Constructs listed and defined, each mapped to a research question.
• Response format chosen and justified by the planned analysis.
• Items written one idea at a time, with non-overlapping options.
• Data template built, one column per item, with readable variable names.
• Code list written for every closed item, including missing-value rules.
• Coding frame written for every open-ended item.
• Pilot completed and pilot data coded successfully.
• Change log dated and complete.
• Instrument specification ready for your supervisor.

For the proposal stage that precedes all of this, see How to Write a Research Proposal: A Practical Guide for University Students in Uganda. For where the specification ends up in the final document, see How to Structure a University Dissertation, Chapter by Chapter.

Note on sources. The supplied sources support general points about construct definition, validated instruments, and multidimensional scales [S4][S6][S7], and one supports the principle that instruments should fit the target group's needs [S8]. None addresses questionnaire design in Uganda or East Africa, and none provides guidance on sample size or item counts. Where this guide states a procedure rather than a finding, it is reasoning from general methodological principles. If your supervisor or examiner expects a cited methods text for those procedures, the sources listed here will not cover that need.

Sources used

• [S4] Prof. Dr., Universidad de Sevilla; Ramírez, Teresa González; García-Hernández, Alién; Dr., Universidad de las Ciencias Informáticas, 2021. "Design and Validation of a Questionnaire to Assess Student Satisfaction with Mathematics Study Materials." International Journal of Instruction. doi:10.29333/iji.2022.1511a
• [S6] Schrepp, Martin; Thomaschewski, Jörg, 2019. "Design and Validation of a Framework for the Creation of User Experience Questionnaires." International Journal of Interactive Multimedia and Artificial Intelligence. doi:10.9781/ijimai.2019.06.006
• [S7] Alva, Lucy Angélica Yglesias; Estrada-Alva, Luis; Montero, Lidia M. Lizarzaburu; Miranda-Troncoso, Adriana Elena; et al., 2025. "Digital Fatigue in Peruvian University Students: Design and Validation of a Multidimensional Questionnaire." Journal of Educational and Social Research. doi:10.36941/jesr-2025-0109
• [S8] Liakina, Natallia; Michaud, Gabriel, 2018. "Needs analyses for task-based curriculum design: How useful can it be for general purpose L2 courses?" Nouvelle Revue Synergies Canada. doi:10.21083/nrsc.v0i11.3996

Citation check summary. Every bracketed citation in the draft was checked against the supplied list. All in-text citations — [S4], [S6], [S7] and [S8] — correspond to supplied keys and are used for claims those sources can reasonably support. No citation was rewritten or removed, because none was found to be invalid. Sources S1, S2, S3 and S5 are supplied but not cited anywhere in the draft, so they are excluded from the "Sources used" list above.

Related reading

• How to Choose a Sample Size for a Student Survey
• How to Write a Research Proposal Methodology Section: Justify Every Choice
• Research Objectives and Questions That Match