What is a Questionnaire? Definition, Examples & Best Practices

A questionnaire is a research instrument made up of a set of questions intended to collect data from respondents. It can be administered on paper, online, by phone, or in person. The questions are fixed in advance, which means every respondent answers the same items in the same way.

The term comes from the French word for "inquiry" and entered widespread use in the social sciences in the late 19th century. Today, questionnaires underpin a huge range of data collection activities, from academic research and clinical assessments to customer feedback forms and employee engagement studies.

A questionnaire is what respondents interact with directly. It is not the same as a survey, though the two words are often used as though they were interchangeable. The distinction matters, particularly when describing research methods or reporting results.

This is one of the most searched distinctions in research methodology, and the answer is straightforward.

A questionnaire is the tool: the document or digital form that contains the questions.

A survey is the process: selecting a sample, distributing the questionnaire, collecting responses, and analysing the data.

Every survey uses a questionnaire. But not every questionnaire is part of a survey. A medical intake form is a questionnaire. A GP practice sending that form to 2,000 patients, collecting the responses, and analysing patterns across the group is a survey.

The confusion persists because most online survey platforms, including MindProbe, use the word "survey" to describe what is technically the questionnaire you are building. That is an industry convention, not a methodological claim.

When writing up research, use the correct term. If respondents answered a set of fixed questions, they completed a questionnaire. If that questionnaire was part of a structured data-collection exercise with a defined sample, it was a survey. If you want to learn more, check out our thought-piece here.

Questionnaires are categorised by question format, administration method, or purpose.

Structured questionnaires use closed questions with fixed response options. Multiple choice, rating scales, and yes/no questions all fall here. The responses are easy to quantify and compare across large samples.

Unstructured questionnaires use open-ended questions that allow respondents to answer in their own words. The data is richer but harder to analyse at scale.

Semi-structured questionnaires combine both. A rating scale question might be followed by an open text box asking the respondent to explain their score.

Self-administered questionnaires are completed by the respondent without an interviewer present. Online forms, paper forms, and postal questionnaires are all self-administered.

Interviewer-administered questionnaires are read aloud by a researcher who records the answers. This format allows clarification but introduces the risk of interviewer bias.

A questionnaire is the right tool when you need consistent, comparable data from multiple respondents and when the questions can be fully specified in advance.

Use a questionnaire for:

• Measuring attitudes, opinions, or satisfaction across a group
• Collecting standardised demographic or behavioural data
• Running pre/post assessments in training or health programmes
• Gathering product or service feedback at scale
• Academic research requiring reproducible data collection

A questionnaire is less suitable when the research goal is exploratory. If you do not yet know what questions to ask, an interview or focus group will generate more useful insights. According to Dillman, Smyth, and Christian in their widely cited "Internet, Phone, Mail, and Mixed-Mode Surveys" (4th ed., 2014), questionnaire design errors are the single largest source of measurement error in survey research.

The practical threshold is roughly this: if you are asking the same questions to more than ten people and intend to compare their answers, use a questionnaire. Below that number, a conversation will usually serve better.

1. Start with your analysis, not your questions. Before writing a single question, specify exactly what you will do with each answer. If a question does not connect to a decision or output, cut it.
2. Use one idea per question. Double-barrelled questions produce uninterpretable data. "How satisfied are you with the price and quality?" cannot be answered clearly because price and quality are separate things.
3. Match the scale to the construct. A five-point Likert scale works well for agreement and satisfaction. A numeric scale from 0 to 10 is better for likelihood or intensity. Picking the wrong scale introduces measurement error that no amount of analysis can fix.
4. Pilot before you launch. Send the questionnaire to five to ten people representative of your target group. Ask them to flag anything confusing. Small wording issues that seem obvious to the designer often confuse respondents who lack context.
5. Keep it short. MindProbe's estimated completion time feature displays how long a questionnaire takes before it goes live. Aim for under ten minutes. Completion rates fall noticeably beyond that threshold, and the respondents who drop out are rarely a random subset of your sample.

• Leading questions. "How much did you enjoy our event?" assumes enjoyment. "How would you rate your overall experience at the event?" does not.
• Ambiguous wording. "Recently" and "often" mean different things to different people. Use specific time frames and frequencies: "in the past 30 days" and "more than three times a week."
• Missing a neutral option. For attitude questions where some respondents genuinely have no opinion, forcing a directional answer produces false data. Include a "Not applicable" or "No opinion" option where it is genuinely relevant.
• Too many open-ended questions. Open text fields are valuable but slow to complete and expensive to analyse. Reserve them for questions where fixed options would lose important nuance.
• Skipping question order logic. General questions should precede specific ones. Sensitive questions should come after rapport has been established through easier items. Use skip logic to route respondents past questions that do not apply to them.