What makes a good survey? Here is a list of things to consider.

The primary survey software is Webropol (Sign in OAMK / University of Oulu). In addition, you can make lighter surveys, for example using MS Forms.

Tips for pollsters:

1. Make your survey short rather than long. Respondents often leave a long, time-consuming survey unfinished.
2. Just ask what you need to know.
   For example, if you send personal links, you don’t have to ask for the respondent’s email address, because you already have it. (Note: survey invitations can only be sent to personal e-mails if you have permission to do so. The permit can be a separate research permit, for example Oamk’s research permit practice. Or the addresses may have been obtained from a personal data register whose privacy description states that one of the purposes of using the contact information is to use the data in question).
3. Make sure that the instructions, questions and statements you give are simple, precise and easy to understand. Then you will get more reliable answers and more respondents will be able to answer all the questions in your survey.
   • If you know the author, don’t use a passive
   • Summarise. Avoid stiff language and long, unclear sentences.
   • Avoid double negation in statements: “I don’t think it’s impossible..” = “I think it’s possible..”
   • An introductory question distorts the results. Aim for a neutral style
4. Ask only one thing at a time.
   – For excample “What school grade do you want to give to the ICT services help website and customer service?”?”
   – Split the question in two, as you can only measure one thing at a time:
   “Give a school grade to the ICT help website” and
   “Give a school grade to ICT customer service.”
5. Put the easiest questions at the beginning.
6. Think about the structure of the survey: Put the questions on the same topic one after the other. The questions can be either on their own pages or grouped on one or more pages.
7. If your survey has more than one page, make it clear how many questions are left. Already in the survey invitation, tell the respondent how long they should reserve for completing the survey.
8. When using ready–made answer options, check that the answer options are comprehensive and mutually exclusive. 
   For example if the answer options “0 – 1” and “1 – 2” both have 1; it should only be in one of them.
9. You can categorize the exact answers into reports, but you can no longer break down the answers that have already been categorized at the survey stage afterwards. So think carefully in advance about what accuracy is enough.
10. Make sure that there is a suitable answer for all respondents.
    You can let the respondent choose ‘something else; which option, in which case they can write their own answer. If necessary, offer “I don’t know” or “I don’t want to say” as the last option. (For example if you ask the respondent’s gender, you can offer the options of man, woman, other, and I don’t want to say.) If you use the Likert scale for statements, always place the “I don’t know” option as the last option: If you do not offer the Likert option in question. Respondents use a neutral answer if they don’t know what to answer and it can distort the results.
11. If the questionnaire contains conditional questions, make it clear to the respondents. A conditional question means that if you answer one question in a certain way, you will either be shown or you will not be shown another question. (This is possible, for example. Webropol’s survey tool. In some other tools, it may be the case that all respondents are given the same form, in which case you have to offer a “I don’t know/doesn’t apply to me” option.)
12. Survey reporting is easier the more you have used ready-made or numerical answer options. When asking for textual answers, make them optional, if possible. Free-form answers are laborious from the point of view of processing, but they can contain particularly useful things.

« Back