Tag: questionnaire design

Most public health final year projects in Nigeria stand or fall on one document: the questionnaire in Appendix A. Your supervisor will skim Chapter One. The defence panel will read Chapter Four. But the external examiner will turn to the back, read your instrument item by item, and decide within four minutes whether your findings mean anything at all.

The problem is that almost nobody teaches you how to build one. You are told to “design a structured questionnaire” and left to it. So you open a laptop, write forty statements beginning with “Do you agree that…”, print two hundred copies, and only discover at the defence that half your items measured nothing you claimed to measure.

This guide takes you from an approved topic to a validated instrument in ten steps, with the actual text you would write at each stage. It assumes the standard Nigerian arrangement: a departmental project in Public Health, Community Health, Environmental Health or Nursing, a cross-sectional descriptive design, a supervisor with limited time, and a five-chapter format. Confirm the fine details with your own department, because instrument requirements vary between UNILAG, the University of Ibadan, ABU Zaria and OAU.

Step 1: Turn every research question into a numbered questionnaire section

Do not start by writing items. Start by writing the map. Take your research questions from Chapter One exactly as they were approved and lay them out as sections.

Suppose your topic is knowledge and practice of exclusive breastfeeding among nursing mothers attending a primary health centre. Your research questions might be:

1. What is the level of knowledge of exclusive breastfeeding among nursing mothers attending the centre?
2. What is the level of practice of exclusive breastfeeding among the respondents?
3. What factors are associated with non-practice of exclusive breastfeeding?
4. Is there a significant relationship between maternal knowledge and practice of exclusive breastfeeding?

That produces a questionnaire with exactly four content sections plus a demographic section:

• Section A — Socio-demographic characteristics
• Section B — Knowledge of exclusive breastfeeding (answers research question 1)
• Section C — Practice of exclusive breastfeeding (answers research question 2)
• Section D — Factors associated with non-practice (answers research question 3)

Research question 4 needs no section of its own. It is answered by cross-tabulating Section B against Section C, which is a statistical operation, not an instrument.

Write this map on one page and show it to your supervisor before you write a single item. It is the cheapest correction you will ever receive. Every item you later write must belong to a section, and every section must trace back to a research question. If an item does not fit anywhere, it does not go in the questionnaire, however interesting it is.

Step 2: Match the item type to the variable, not to habit

The single most common failure in Nigerian public health projects is putting everything on a four-point agreement scale. Knowledge is not an opinion. You cannot ask a mother whether she agrees that exclusive breastfeeding lasts six months — she either knows the recommended duration or she does not.

The KAP structure — knowledge, attitude, practice — gives you three different item types, and mixing them up destroys your analysis:

• Knowledge items take a correct answer. Use True/False/Don’t know, or multiple choice with one right option. These are scored: one mark for correct, zero for wrong or don’t know.
• Attitude items take an agreement scale. A four-point Likert scale — Strongly Agree, Agree, Disagree, Strongly Disagree — is the Nigerian default because it forces a direction and gives you the familiar 2.50 decision criterion.
• Practice items take a frequency or a yes/no behavioural report. “How many times in the last week did you…”, not “Do you agree that you should…”.

Write the item type beside every row of your section map before drafting. If you are unsure which analysis your item types will permit later, work backwards from the test you intend to run — the guide on choosing the right statistical test for your project shows which question shapes require which test, and therefore which item format you must collect.

Step 3: Write Section A so it earns its place

Section A is not filler. Every demographic variable you collect should be one you intend to cross-tabulate. Panels ask a brutal question about this: “You collected religion. Where does religion appear in Chapter Four?” If the answer is nowhere, you wasted the respondent’s time and your printing budget.

For a public health study, the defensible core is usually:

1. Age (in completed years, or in bands you will actually use)
2. Sex
3. Marital status
4. Highest level of education completed
5. Occupation
6. Monthly income band, if income is in your objectives
7. Parity or number of living children, for maternal health topics
8. Residence — urban, semi-urban, rural — if your sampling crosses settlements

One rule that saves a lot of pain: collect age as a number, not as a band, whenever you can. You can always collapse numbers into bands during analysis, but you can never recover the number from a band. The same logic applies to number of children and years of experience.

Step 4: Write knowledge items with a defensible correct answer

Every knowledge item needs a source you can name. If a panel member asks “Who says that is the correct answer?”, the answer must be a guideline, not your opinion. For public health topics, the usual authorities are the World Health Organization, the Nigeria Centre for Disease Control and Prevention, the Federal Ministry of Health, or the National Primary Health Care Development Agency.

Worked examples for the breastfeeding study:

• Exclusive breastfeeding means giving the baby only breast milk, with no water, for the first six months. [True / False / Don’t know]
• A baby on exclusive breastfeeding should be given water during hot weather. [True / False / Don’t know]
• Breastfeeding should begin within one hour of delivery. [True / False / Don’t know]

Notice the second item is deliberately false. A knowledge section made entirely of true statements can be passed by any respondent who simply ticks “True” all the way down, and your panel will spot that. Aim for roughly a third of your knowledge items to be false statements.

Then decide your scoring rule in advance and write it into Chapter Three. A common and defensible rule: fifteen knowledge items, one mark each, with 0–7 classified as poor knowledge, 8–11 as fair, and 12–15 as good. State that the cut-offs follow Bloom’s original criteria of below 50 per cent, 50–74 per cent and 75 per cent and above, and cite a published Nigerian study that used the same cut-offs. Never invent cut-offs after seeing your data — that is the fastest route to a correction.

Step 5: Write attitude and practice items that cannot be misread

Six drafting rules, each of which fixes a mistake that appears in almost every first draft:

1. One idea per item. “Exclusive breastfeeding is healthy and convenient” is two items. A mother who finds it healthy but inconvenient cannot answer.
2. No negatives in the stem. “I do not believe formula is unnecessary” is unanswerable. Disagreeing with a double negative is guesswork.
3. No leading language. “Do you agree that exclusive breastfeeding is the best choice for your baby?” tells the respondent what to say.
4. No technical vocabulary. Replace “complementary feeding” with “giving other food alongside breast milk” unless your respondents are health workers.
5. Anchor the recall period. “In the last seven days” or “since your last child was born” — never a bare “usually”.
6. Include two or three reverse-scored items so you can detect straight-lining, and remember to recode them before analysis.

Keep the whole instrument to a length a real respondent will finish. For a community sample, forty to fifty items across all sections is the practical ceiling. Fifteen minutes is the limit at a busy primary health centre, and a mother holding a baby in a queue will abandon anything longer.

Step 6: Adapt a validated instrument rather than inventing one

This is the step that separates a strong project from an average one, and it takes an afternoon. For most public health topics, someone has already built and validated an instrument.

Search Google Scholar and AJOL for your construct plus “questionnaire” or “scale”, restricted to studies in Nigeria or West Africa. When you find an instrument in a published article, three things follow. You can state its origin in Chapter Three. You can quote its reported reliability coefficient. And you can defend your items by pointing to prior use rather than to your own judgement.

Adapting is legitimate and expected — you change wording for your setting, drop irrelevant subscales, and add items for your local context. What is not legitimate is copying an instrument and presenting it as your own design. Cite the original authors in your instrument paragraph and in your reference list, following the referencing style your Nigerian university actually requires. Adaptation without attribution is a plagiarism finding waiting to happen.

Before you commit to a construct, it is also worth checking how heavily it has already been studied in Nigeria; the method in counting existing Nigerian studies on your topic tells you within an hour whether validated instruments exist for it.

Step 7: Establish content validity with three lecturers

Nigerian departments overwhelmingly expect face and content validity established by expert review, and the standard is three experts: your supervisor plus two other lecturers, ideally one from measurement or biostatistics.

Do it properly rather than by informal chat. Give each reviewer a copy of the instrument with your research questions and objectives attached, and a rating column asking them to mark every item as Relevant or Not relevant. Then compute the Content Validity Index: for each item, the proportion of reviewers who rated it relevant. Items scoring 1.00 with three reviewers stay. Items where two of three agreed are revised. Items where only one agreed are dropped. The scale-level index is the average across retained items, and anything from 0.80 upwards is comfortable to report.

Keep the signed rating sheets. Several departments require them as an appendix, and even where they do not, a panel member who asks how validity was determined is answered instantly by producing them.

Step 8: Pilot on ten per cent and compute Cronbach’s alpha

Pilot on a group similar to your respondents but not part of your final sample — a different ward, a different health centre in the same LGA. Ten per cent of your calculated sample size is the conventional Nigerian pilot, so a sample of 300 means about 30 pilot respondents.

The pilot does three jobs. It tells you how long the instrument takes. It surfaces items respondents misunderstand, which you catch by watching where they hesitate or ask questions. And it gives you the data for your reliability coefficient.

Compute Cronbach’s alpha separately for each Likert-scaled section, not for the whole questionnaire at once. Alpha across sections measuring different constructs is meaningless. Knowledge sections scored right/wrong technically call for the Kuder-Richardson formula, though many Nigerian departments accept alpha as an approximation — ask your supervisor which your department expects. If a coefficient comes back low, the diagnosis and the repair are set out in the guide to what counts as a good Cronbach’s alpha, and the fix is almost always deleting one or two bad items rather than rewriting the section.

Step 9: Write consent, and get ethical clearance before you distribute

Health research on human subjects in Nigeria requires ethical clearance, usually from your institution’s Health Research Ethics Committee or from the State Ministry of Health where you are collecting data. Students routinely underestimate how long this takes — two to six weeks is normal, and it must happen before distribution, never after.

Your instrument needs a consent page carrying the study title, your name and department, the purpose in plain language, the expected duration, a statement that participation is voluntary and can be withdrawn at any time, an assurance of anonymity, and a contact. Because you are handling personal data, the Nigeria Data Protection Act 2023 applies to you: collect only what you need, do not put names on questionnaires, and keep completed copies secure.

The delivery decision matters here too. A paper instrument administered at a health facility reaches a different population from an online link, and the trade-offs are compared in the piece on collecting project data with Google Forms, KoboToolbox or paper. For clinic-based public health samples, paper administered on site almost always yields the better response rate.

Step 10: Write the instrument paragraph for Chapter Three

Here is the paragraph, ready to adapt. Substitute your own numbers and sources:

Data were collected using a structured, self-administered questionnaire adapted from Adeyemi and Okafor (2022) and modified to suit the study setting. The instrument comprised four sections. Section A collected socio-demographic data across eight items. Section B measured knowledge of exclusive breastfeeding through fifteen items scored as True, False or Don’t know, with one mark awarded for each correct response, giving a maximum score of 15; scores were classified as poor (0–7), fair (8–11) and good (12–15) following Bloom’s criteria. Section C measured practice through twelve items on a four-point Likert scale ranging from Strongly Agree (4) to Strongly Disagree (1), with a decision criterion of 2.50. Section D contained ten items on factors associated with non-practice. Content validity was established by three experts in the Department of Public Health, yielding a scale-level Content Validity Index of 0.89. The instrument was pilot-tested on thirty nursing mothers at a comparable primary health centre who did not form part of the main study, and reliability coefficients of 0.81 for Section C and 0.78 for Section D were obtained using Cronbach’s alpha, indicating acceptable internal consistency. Ethical approval was obtained from the Health Research Ethics Committee, and written informed consent was obtained from every respondent.

That paragraph answers, in advance, every instrument question a panel is likely to ask. It belongs in the instrumentation subsection of Chapter Three, and it sits inside the wider structure explained in the guide to writing Chapter Three section by section.

Four mistakes that cost marks at defence

1. Items that do not map to any research question. The panel checks this. Every orphan item invites the question of why you collected it.
2. Reporting one alpha for the entire questionnaire. It signals that you ran the analysis without understanding it.
3. Piloting on people who are also in the main sample. It contaminates your data and an alert examiner will catch it.
4. Changing the instrument after data collection has started. Responses to two different versions of an item cannot be pooled. If the pilot showed a problem, fix it before, not during.

Turn your validated instrument into a finished Chapter Three

Once your items are written, your experts have rated them and your pilot has produced a coefficient, the remaining work is drafting — turning decisions you have already made into the formal prose your department expects, then keeping it consistent as your supervisor sends corrections back.

Tesify was built for exactly that stage. It drafts your chapters against your own approved topic, objectives and instrument, keeps your citations and bibliography consistent as the draft moves, and lets you rework a section after supervisor comments without restarting the chapter. It is a way to write your own project faster, not a way to submit someone else’s.

Start your project on Tesify for free and draft Chapter Three around the instrument you have just built.

Frequently asked questions