Designing your survey

Surveys

After developing a clear plan, you can begin to design your survey.

Designing your survey involves choosing the survey method, writing your questions, and ensuring your survey structure is clear. A well-designed survey will be easy for participants to complete and will ensure the information you collect can be analysed effectively.

Choose your survey delivery method

Surveys can be delivered in different ways, depending on your target population. The most common methods include:

  • Online

  • Telephone

  • Postal (or paper)

  • Face-to-face

Online surveys are a common method, but they risk excluding people with limited digital access or who aren’t digitally confident.

Alternative options include paper or postal surveys, which may work well if you are collecting data at a community event, or a GP waiting room. However manually processing the data will take longer, and there is a greater risk of errors in participants’ answers (e.g., handwriting that is difficult to read or respondents selecting more than one option when only one is allowed).

Telephone surveys allow you to reach populations who might not be able to engage online or in-person but require more resource, time, as well as skills to confidently administer a telephone survey and navigating permissions to contact potential respondents via phone.

A few key things to consider:

  • Who is your audience? Think about your target population’s age, literacy levels, digital access, and any accessibility or language needs they might have that could impact taking part

  • What resources do you have available? Consider how much time, budget, and staff resources are available to manage alternative methods of data collection which might take more time (e.g., paper surveys)

  • How many responses do you need? Some methods can provide higher response rates than others (e.g., face-to-face) so it is important to think about how many people you want to reach.

  • How sensitive is your topic? Asking questions about particular topics or experiences (e.g., mental health, sexuality, financial difficulties, reproductive health, experiences of victimization) may need a more personal approach where you are able to offer more support.

Decide on the type of survey

When designing your survey, it is important to consider not only how you will collect data (e.g., your method) but also the type of survey you are carrying out. The type of survey refers to the overall structure and timeframe of your survey, rather than how you will deliver it.

There are various survey types to choose from. Your objectives, resources, and how you plan to use the results will influence which type of survey you choose to carry out.

Survey Type What is it? When do I use it? Example
Cross-sectional Aone-off snapshot at a single point in time When you need quick insights into current attitudes, behaviours, or experiences from a sample of a population Asurvey of parents about how often their child enjoys and participates in P.E. lessons
Longitudinal Follows the same group of people over a (often extended) period of time When you want to understand how one group’s behaviours or health evolves and changes
over time Surveying the same group of children and parents each term for two years to track changes in enjoyment and confidence in P.E. lessons
Repeated cross-sectional The same survey repeated over time with a different group each time Whenyou want to monitor population-level trends without tracking the same groups of people Running the same survey every year, but with new sets of parents as children move through different year groups
Pre-post (before and after) Data collected from the same group before and after an event, intervention or programme When you want to measure change (e.g., in knowledge, confidence, or behaviour) linked to a specific programme or service Surveying parents before and after a new after-school P.E. programme to see if their child’s enjoyment and wellbeing has improved

Writing good survey questions

Good survey questions are the foundation of useful data. Well-written questions limit misinterpretation and help respondents to interpret the questions in the same way. Poorly worded questions can confuse respondents and impact data quality.

The key to writing good questions is keeping them clear, simple, and focused. Each question you ask should have a clear purpose and directly link to your overall objectives.

It is also important to consider inclusivity when writing your questions. This involves reflecting on any biases we might have, and how these could shape the questions ask or how we ask them. Our positionality1 (how our own background, identity, values and experiences can shape how we design and interpret research) can unintentionally influence our research, how it is conducted, and the results.

A few practical tips:

  • Check your assumptions: Before finalising questions, ask yourself if you are assuming all respondents share a particular identity or experiences.For example, instead of asking “do you live with your mum and dad?” consider asking “who do you live with?”.

  • Use neutral and respectful language: Avoid negative or deficit-based wording (e.g., “why do you not exercise?”) and use more open framing (e.g., what makes it difficult for you to be physically active?”). See How do I make sure my survey is inclusive? for further guidance. #TODO

  • Pilot your survey questions: Testing your survey questions with a range of people from different backgrounds can highlight where language or response options might exclude people.

  • Acknowledge your own lens: Be as reflexive as possible. For example, note in your planning documents or discuss with your team how your perspective may have shaped your survey design, and be clear about how you have addressed this (e.g., through peer review, piloting, or co-production).

Remember: Always consider the respondent’s perspective. Would they understand exactly what you are asking and know how to answer?

Below are some do’s and don’ts to keep in mind when writing your questions.

Do:

  • Keep questions short and simple

  • Use plain English

  • Ask one question at a time

  • Make sure each question is relevant to your objective

  • Use neutral wording so you don’t lead respondents to a particular answer

  • Start your survey with easy, non-sensitive questions before moving to more personal or complex questions

  • Use inclusive language

  • Check the readability of your questions (either with a colleague, a resident, or using a tool like Hemingway)

Don’t:

  • Use acronyms or jargon. If you have to use acronyms, always explain what they mean.

  • Ask unnecessary questions. Don’t ask anything you won’t use or “just in case”

  • Use double-barreled questions (e.g., is the service friendly and affordable). Split questions into different parts if you have to.

  • Ask leading questions (e.g., do you agree that…)

  • Use long sets of closed questions. If you have to, break up questions with open-text options or visual breaks.

  • Use ambiguous wording (e.g., do you regularly exercise?)

Choosing response options

There are various ways people can respond to your survey questions – and the type of response options you provide will impact the data you get in return.

There are generally two different types of questions which impact the type of responses people can choose from.

  • Closed questions give respondents a fixed set of answers to choose from. These questions are analysed quantitatively and can help us to compare between sub-groups because everyone is answering in the same way, but they don’t allow us to gather more detailed information.

  • Open questions allow respondents to answer in their own words and give whatever answer they like. These questions are also referred to as ‘free text’ or ‘open text’ questions. Open questions can provide richer data and are useful when you want more detailed information, but they typically analysed qualitatively (e.g., using thematic analysis).

Understanding how you will use your survey responses will help you identify the best format and response options. Think about the level of detail or type of information you want to collect from your sample. Some of the most commonly used response options you might come across are listed below.

  • Multiple choice (e.g., select all that apply)

  • Limited choice (e.g., select one option)

  • Likert scales (e.g., strongly agree → strongly disagree). These are often best for measuring frequency, satisfaction, agreement, or quality. 5-point Likert scales are the most common, but if you need more detailed responses, you might want to consider a 7-point scale.

  • Rating (e.g., rate your satisfaction with the service from 1 to 10)

  • Ranking (e.g., please rank the following in order of importance to you)

  • Frequency (e.g., never, less than once a month, 1 to 3 times a month, at least once a week)

Including a neutral response option, like ‘I don’t know’ or ‘I’m not sure’, can help reduce the risk of forcing respondents to answer inaccurately.

If you are including a ‘frequency’ response option, make sure your response options do not overlap. For example, ‘1 to 3 times per year’ and ‘3 to 6 times per year’ overlap at ‘3’. Instead, use clear, distinct ranges such as ‘1 to 3 times per year’ and ‘4 to 6 times per year’.

Even when using closed questions, try and include space for respondents to add their own comments or reflections where possible.

Adding an optional open text (e.g., please use this space to tell us more about…) offers respondents the opportunity to clarify, add context, or expand upon their answers to a previous question. However, it is important to use open text questions with purpose. Not every closed question response will need an explanation, and open text questions can increase the time required for both respondents and analysts.

Using validated measures and existing instruments

A validated measure is a set of questions (sometimes referred to as an instrument) that has been tested to make sure it is reliable and actually measures what it is designed to measure.

It is not always necessary or best practice to write your own survey questions. Many high-quality validated measures already exist, and it might be appropriate for you to use these instead of creating a survey question from scratch.

Using a validated measure has several advantages:

  • Reliability: as they have been tested beforehand, using validated measures can reduce the risk of misleading results

  • Comparability: using a measure that another study has used, or that has been used in a national survey, can allow you to benchmark your findings against other datasets

  • Efficiency: using a validated measure can save time and resources that would be used developing your own bespoke questions

Many of the areas we focus on in our work are complex and can be measured in a variety of different ways. Using ‘tried and tested’ instruments can reduce the burden of creating bespoke measures and enable us to collect data that aligns with national standards.

Below are some examples which are widely used in UK health research. We recommend exploring some of these tools to see whether they are suitable for your survey. If you would like to discuss using a validated measure in your survey, please get in touch with

  • General Health Questionnaire (GHQ-12)

  • Patient Health Questionnaire (PHQ-9)

  • Short Form 12 Patient Questionnaire (SF-12)

  • ONS4 Personal Wellbeing

  • Warwick-Edinburgh Mental Wellbeing Scale (WEMWBS)

  • Alcohol Use Disorders Identification Test (AUDIT-C)

  • Child Eating Behaviour Questionnaire (CEBQ)

  • Strengths and Difficulties Questionnaire (SDQ)

Some resources for finding and exploring existing validated measures:

Structuring your survey

How you choose to structure your survey is equally as important as the questions you ask. Well-structured surveys that flow nicely can reduce the risk of respondents abandoning the survey and increase your response rate.

Start your survey with simple, easy to answer questions

Beginning your survey with simple questions (e.g., demographic questions, or questions with a yes/no response) can build momentum and give respondents confidence to continue the survey. Respondents might be more likely to answer sensitive or complex questions later in the survey.

Group questions by theme

Jumping between questions about different topics can be frustrating for respondents. If you are asking a lot of questions, grouping these by theme can help a survey naturally flow from one section to another. For example, ask all of your questions about physical activity and then ask questions about diet.

Add branching or skip logic to your survey

Using a branching or skip logic feature can improve the flow of your survey. Branching allows you to control which questions a respondent sees based on their responses to a previous question. Essentially, this means you can customise the respondents’ survey path and ensure they only see questions which are relevant to them.

For example, if a respondent says that they do not smoke, they won’t be asked to answer any questions about their smoking behaviour. This can make surveys shorter for some respondents as well as remove the risk of collecting any unnecessary data.

Use appropriate question formats

It is important to choose the right type of question for the information you want to collect. Matrix or grid style questions allow respondents to answer multiple questions on the same scale and can reduce cognitive load and answer time.

Multiple choice or drop-down response options can be useful for questions which have lots of response options.2

Test the structure of your survey

Always test your survey with multiple people before it’s formally launched. You could ask colleagues or a similar group to your target population. This will help ensure your survey flows well and there are no errors.

Top tips for effective survey design

  • Be clear about purpose: Always provide clear explanations as to why you are carrying out the survey and what their information will be used for (e.g., to inform policy, or make a change to a service). People are much more likely to take part if they feel their contribution will make a difference.

  • Keep it simple: Avoid difficult, confusing and complex questions or instructions. Don’t ask more questions than you need to and only collect information you will actually use

  • Reduce effort for respondents: Keep the survey short and straightforward. Tell respondents upfront how long your survey will take to complete

  • Limit personal data requests: Only ask for personal information if it is absolutely necessary. Always explain clearly why you are collecting personal information from respondents and how it will be used (e.g., to identify any differences between particular groups). Reassure participants that their answers will remain confidential.

  • Test before launching: Always pilot your survey to check for any unclear wording, missing response options, or technical issues. For larger surveys (e.g., over 50 participants) you could consider cognitive testing3 with a group similar to your target audience. This helps ensure questions are understood as intended and can improve the quality of your data. Even small misunderstandings in wording can lead to biased or inaccurate results.

Further information

Back to top

Footnotes

  1. Positionality refers to an individual’s worldview, background, and identity, and the position they adopt in a research context.↩︎

  2. Matrix, grid, and drop-down responses are not offered in the current version on Microsoft Forms.↩︎

  3. Cognitive testing is a method of capturing respondents’ thought processes and understanding in responding to questions. This method can help identify any issues with your survey and improve the reliability of your survey data.↩︎