To design a dissertation survey or questionnaire, start from your research questions, decide what each construct needs to measure, use validated scales where they exist and write any new items clearly, then pilot the whole instrument before launch.
This guide walks through each step: planning, choosing existing instruments, question types, Likert scales, wording and order, reliability and validity, sample size, ethics and the analysis plan.
The Six Steps of Questionnaire Design
- List your research questions and the constructs each one involves.
- Find validated scales for each construct, and get permission if needed.
- Write any new items, plus demographic questions you will actually analyze.
- Order the questionnaire logically and keep it as short as possible.
- Pilot it with people like your participants and revise.
- Plan the analysis for every item before you launch.
Start from Your Research Questions
Every question in your survey should earn its place by helping to answer a research question. Items that are "interesting to know" lengthen the survey, lower completion and rarely appear in the results chapter.
Build an item map before you write anything. It keeps the instrument focused and later becomes a useful appendix.
| Research question | Construct | Measure | Planned analysis |
|---|---|---|---|
| RQ1: Is workload related to burnout? | Workload; burnout | Validated workload subscale; validated burnout scale | Correlation, regression |
| RQ2: Does burnout differ by shift pattern? | Shift pattern | One categorical item | ANOVA or Kruskal-Wallis |
| RQ3: What support do nurses want? | Preferred support | One open-ended item | Content or thematic analysis |
This illustrative map uses a nursing example. Notice that every measure links to a question and an analysis.
Use Validated Scales Where You Can
For established constructs such as burnout, self-efficacy, wellbeing or job satisfaction, a validated scale is almost always better than writing your own. It has evidence of reliability and validity, and your results can be compared with other studies.
- Search the literature and test databases for scales used with populations like yours.
- Check the license: some scales are free for research, while others require permission or a fee.
- Use the items and response options exactly as published; changing wording can affect validity.
- If you must adapt a scale (for example, translating it or changing the context), say so and justify it.
- Cite the scale's source and report its reliability in your sample.
Question Types and When to Use Them
| Type | Example | Best for | Watch out for |
|---|---|---|---|
| Single choice | Which shift pattern do you mainly work? | Categories such as job role or setting | Options must be exhaustive and not overlap |
| Multiple choice | Which support services have you used? (select all) | Several applicable answers | Harder to analyze; plan coding in advance |
| Likert-type item | "I feel emotionally drained by my work" | Attitudes, perceptions, agreement | Single items are ordinal |
| Rating scale | Rate your workload from 0 to 10 | Intensity or quality | Label the ends clearly |
| Numeric | How many years have you been qualified? | Counts and continuous variables | Set sensible limits in the survey tool |
| Open-ended | What one change would help most? | Explanations and new ideas | Time-consuming to analyze; use sparingly |
Designing Likert Scales
Likert-type items ask respondents how far they agree with a statement. They are widely used, but small design choices matter.
- Number of points: five and seven points are the most common. Evidence on which is better is mixed; consistency with your validated scales matters more.
- Midpoint: a neutral option lets genuinely neutral people answer honestly; removing it forces a choice. Researchers disagree, so justify your decision.
- Labels: label every point, or at least both ends, with balanced wording (strongly disagree to strongly agree).
- Reverse-worded items: can reduce automatic agreement, but can also confuse respondents. Reverse-score them before analysis.
Single Likert items are ordinal. Summed or averaged multi-item scales are commonly analyzed with parametric tests, which is standard practice in many fields, but state and justify your approach.
Writing Clear Questions: Worked Rewrites
Poor wording is the most common reason survey data cannot answer a research question. These rewrites show the usual problems.
Double-barrelled. "My manager is supportive and approachable." Fix: split into two items, one for supportive and one for approachable.
Leading. "How much has the excellent new rota improved your wellbeing?" Fix: "How, if at all, has the new rota affected your wellbeing?"
Double negative. "I do not disagree that breaks are not taken." Fix: "I am usually able to take my breaks."
Vague time frame. "Do you often feel tired at work?" Fix: "In the past month, how often have you felt exhausted at the end of a shift?" with labeled frequency options.
Overlapping options. "Age: 18 to 25, 25 to 35, 35 to 45." Fix: "18 to 24, 25 to 34, 35 to 44, 45 or over".
Use plain language your respondents use themselves, avoid jargon and abbreviations, and ask about one thing per item. Include a "prefer not to say" option for sensitive demographic questions, and a "not applicable" option where some respondents genuinely cannot answer.
Order, Length and Layout
- Start with the participant information and consent.
- Open with easy, relevant questions that build engagement.
- Group items by topic and keep validated scales intact.
- Place sensitive questions later, once trust is established.
- Put most demographics at the end, unless they are needed for screening or routing.
- Finish with a thank-you and any debrief or support information.
Keep the survey short. Long questionnaires lose respondents and produce careless answers towards the end. Test the completion time in your pilot and cut anything not on your item map.
Online, use one topic per page, show a progress bar if your tool offers one and check the survey on a phone, since many respondents will complete it there. Use routing (skip logic) so people only see questions that apply to them, and test every route before launch.
Reliability, Validity and Sample Size
Examiners will ask how you know your questionnaire measures what it claims to, and whether your sample is large enough.
| Concept | What it means | How to address it |
|---|---|---|
| Internal consistency | Items in a scale measure the same construct | Report Cronbach's alpha or McDonald's omega; .70 is a common rule of thumb |
| Content validity | Items cover the construct fully | Use validated scales or an expert review |
| Construct validity | The scale behaves as theory predicts | Cite evidence from the scale's development; factor analysis in larger studies |
| Face validity | Items seem relevant to respondents | Check in the pilot |
For sample size, run a power analysis for your planned test. If you aim to estimate a percentage in a large population, about 385 responses give a margin of error of plus or minus 5 percent at 95 percent confidence, assuming random sampling. Most dissertation samples are convenience samples, so discuss generalisability honestly.
Ethics, Data Protection and Tools
- Get ethics approval before collecting any data, including pilot data.
- Explain purpose, voluntary participation, withdrawal and data use on the first page.
- Collect only the personal data you need; avoid names and email addresses unless essential.
- In the UK and EU, follow data protection law (UK GDPR or GDPR) and your university's storage rules.
- Use a survey platform your university approves; many provide Qualtrics, Microsoft Forms or similar.
Avoid offering incentives that could pressure participation without ethics committee approval, and never collect data from groups your approval does not cover.
How DissertationHorse Helps with Questionnaire Design
Send your research questions, constructs, any validated scales you plan to use and your ethics requirements. Your writer drafts or reviews a custom questionnaire with an item map, clear wording, a sensible order and a matching analysis plan, and can write the instrument section of your methodology.
We never complete surveys, invent respondents or fabricate data. Free revisions within the scope of your original request are included.
Want a Questionnaire That Measures What You Need?
Send your research questions, constructs and any scales you plan to use. Your writer drafts or reviews a custom questionnaire with an item map and analysis plan.
Order Your Questionnaire HelpFree revisions · Full refund if late · Written from scratch for your order
Frequently Asked Questions
Use validated scales for established constructs wherever possible. Write your own items only when no suitable measure exists, and pilot them carefully.
As few as your research questions need. Many dissertation surveys take 10 to 15 minutes to complete; test the timing in a pilot.
Both are widely used, and evidence is mixed. Follow the published format of any validated scale and stay consistent across your own items.
Avoid it where possible, since changes can affect validity. If you must adapt it, get permission if required and justify the change.
.70 is a common rule of thumb for acceptable internal consistency, though very high values may suggest redundant items. Report it for each scale in your sample.
It depends on your analysis. Run a power analysis for your main test, and be realistic about response rates when planning recruitment.
It is strongly recommended. A pilot reveals unclear items, routing errors and completion time before the main launch.
Yes, a few can add useful context. Plan how you will analyze them, and avoid calling the study mixed methods unless the qualitative data form a full strand.