Internet notebook about my work: deep listening to facilitate positive change


Friday 25 June 2010

Web survey

A web survey is a good way to explore demand and how to best satisfy it. The tools are there and make it very easy. They are user friendly and are mostly free, but better use for a short period the upgraded versions. I learned the following five 'rules of the game'.
The respondents. Define whose opinion you need. If it is a small group of 10 to 20 people you already know, you may reconsider a web-survey and do telephonic interviews. If it is a large group of a few hundred or few thousand people it is important to know how much the issue is of concern to them. The closer they are to the issue the higher the response you may expect. 2% is a very high response for a large and ‘distant’ group. 20% response is a high for a smaller and 'closer' group. Make sure you have updated mailing lists.
The objectives. Define the two or three things you really want to get out of your questions. To do so it is best to first conduct telephonic interviews. This qualitative research will help you formulate what is really relevant to ask. 5-10 semi-structured interviews will do. Don’t ask more than 6-8 questions. Analyze and discuss the results with a few colleagues. Then design the web survey. If you don’t have time for interviews, at least discuss objectives and framework of the survey with colleagues.
The three parts of the survey. Start the survey with a concise formulation of the background and objectives. Explain what is in it for the participants, e.g. "it will help us to service you better"; "you will be credited in the publication as a contributor". As to the questions, keep their number as low as possible. Make sure you ask about the profile of respondents. Ask their email address to be able to keep in touch. In a choice matrix keep the number of choices around 5. More choices make reporting in a diagram difficult to read. If you need more choices use a new question. Ask to explain their answer when you use a closed question. The last question should be about what else respondents would like to say about the issue and what advice they have for you. Always end the survey with a thank you and indicate when and where respondents can see the results of the survey.
The test. Before you send out the survey always test its ‘usability’ among a few colleagues. Ask them to do the survey. This alone may bring to light some small mistakes. Also ask them about whether or not to use ‘compulsory’ questions. Are the questions clear? Do the open questions work or irritate? What other advice they have?
The message. The more personal the email with the request to participate, the more response. Use a sender they know, appreciate and has credibility. Keep the text in the mail short, but personal. If you use more mailing lists, apologize for possible cross-posting. Keep the time frame for the survey short (maximum three weeks). Send a reminder halfway the period for response.

No comments: