This post by Robyn Kannemeyer first appeared on Thesislink in October 2016.
You may think that writing a survey is easy but believe me, if you are conducting a survey and you want to obtain data which is meaningful and straightforward to analyse, then consider the following ten tips. They will be help you to design and write your survey and may save you some heartache when you are analysing your data.
- Formulate your research aims or objectives before you design your survey. This is the most important first step because even if you design a well-constructed survey it will be meaningless if it doesn’t answer your research questions. Ensure you have fully researched your topic and have spent time reading about other relevant research or surveys that have been carried out.
- Construct the survey with the data analysis in mind i.e. you need to think about how the question might be answered by your respondents and then how you will analyse the results. The resulting data should test your hypotheses. This is a good way of pruning the length of the survey because a particular question may not provide useful data.
- Keep the survey to a manageable length. My first response when asked to participate in a survey is – how long will it take? Remember you are relying on a respondent to answer your survey in their own time so ‘If in doubt, leave it out’.
- Keep the instructions for the respondents clear, unambiguous and easy to follow. Explain the purpose of the survey at the outset and instruct your respondent as to how you want the question answered. There is a big difference between ranking and rating a response.
- Ensure the questions are brief, readable, and understandable so they can be easily answered. Bad grammar and spelling will detract a respondent from answering the question. Consider these points:
- Select the words carefully and keep them simple.
- Think about the intended audience that will be reading the questions. It is best not to use vague words as this encourages vague answers.
- Only ask one concept per question and make sure the question isn’t too general.
- Avoid ambiguity e.g. double negatives, two questions in one, or words with two different meanings.
- Don’t use leading questions i.e. don’t ask the question in such a way that you suggest the answer, as this may influence how the respondent answers the question, and lead them to respond in a way that is inconsistent with how they feel.
- Avoid prejudicial language.
- Be cautious with personal or ethical questions as the respondent may not feel like completing the survey if they are offended by the question.
- Think carefully about the order of the questions and how you structure the survey. The opening question should put the respondent at ease and encourage them to complete the survey. It may be better to put the sensitive questions at the end of the survey after you have built rapport with the respondent. Also, do not end the survey with a complicated question because the respondent may not be willing to answer it.
- Pilot- test the survey on a sub-sample of respondents. This will help you pick up any problems with the instructions, ambiguous questions, or whether any questions can be eliminated. This feedback is ‘pure gold’, especially if you can have a chat afterwards. You may have to do another pilot test if you make substantial changes.
- Revise the questions that cause difficulty and eliminate any questions that don’t contribute to the research questions you are trying to answer. You may even add new questions at this stage.
- Pre-code the responses. Closed questions are predominantly used in surveys and these can be pre-coded to enable the responses to be classified into meaningful categories. This helps you to think about how you will analyse the data. It is also important to think about how you will code the responses from open-ended questions. These can be more difficult to analyse. See May (2011, p. 114) for further information on categories and coding.
- Proof read the final version. Get someone reliable to read over the final version of your survey. You will be amazed at the little mistakes that can be missed e.g. your and you’re; its and it’s; allowed and aloud. Once your survey is live, it is too late!
It is important that you are prepared to write many drafts of your survey before you reach the final version. This will take time but it is all part of the process and will result in a much tighter, clearer and more useful survey. Don’t forget that you may also need ethics approval from your university so planning is essential. Statistics New Zealand has a useful guide for designing a good survey.
These tips comes with a precaution – don’t be surprised if you still receive some unusual survey responses.
May, T. (2011). Social research: Issue, methods and process (4th ed.). Berkshire, England: Open University Press, McGraw-Hill Education.
Statistics New Zealand (2015). A guide to good survey design (4th ed.). Retrieved 1 August 2019 from http://archive.stats.govt.nz/methods/survey-design-data-collection/guide-to-good-survey-design.aspx.