Online Sampling (Part I)

Online surveys are often criticised for being wide-of-the-mark in terms of sampling of the interviewees. As a matter of fact, even that is false since no sampling at all is possible on the web. People will choose to respond to your survey if it is at all visible and that is certainly the most difficult bit. If people have clicked massively on your survey banners and if you think yourself lucky then you'd better think twice.

Most likely, those who have answered your questions are those who have had time to read your pages carefully (students or senior citizens, typically) and not a representative sample of the total population which visits your website. Even worse, they could just be there for the inevitable web freebie you will promise to offer them, and they won't give a damn about the questions you're asking. All they want is enter the prize-draw !

Secondly, it's difficult to cross-check whether your respondents are who they pretend to be. In fact, it's even worse than surveys carried out via the telephone (CAPI*). Thirdly, quite a few CAWI** software won't let you check that a respondent hasn't filled in twice your form.

Lastly, there are other minor drawbacks such as the limitations in terms of length of the surveys, which is a permanent bother for any interviewer but even more so for the online interviewer. Indeed, it can get pretty difficult to read a web page which is too long, and therefore all the aspects linked to the layout of your html surveys are fundamental.

Do check your CAWI software for any bugs or glitches which would make your page too long, too weird or even totally unreadable. Last but not least, the navigation between your questions should be as clear as possible in order to enable your filters to work correctly. Even so you will still get people who will fail to read the instructions and who fill out the questions for which they are not required to answer. You will have to check your software again inorder to be able to modify manually all these unwanred answers which make your numbers cease to add-up. Your software should also allow you to filter your responses and even add up these filters inorder to answer complex 'what-if' questions.

Although there are all these known limitations to online Internet surveys ( check the Hawaï '99 Conference Minutes [ 5-8/01/1999 ] for more details ) that does not mean that they cannot teach you anything. Do all these biases lead you to think that online surveys are no use at all? I wouldn't agree less with that statement. Conversely, I think that online Internet surveys can be the best thing on earth as long as you follow the rules. That's what we'll be seeing with the second part of this article.

Notes:

• *CAPI: Computer Aided Phone Interviewing, i.e. the piece of software that will enable you to design, generate, administer and analyse a survey based on the telephone,

• **CAWI: Computer Aided Web Interviewing, i.e. a similar piece of software as CAPI but as a Web-authoring version. Additionally, good CAWI software should also encompass the hosting of both your survey and the responses to your questions

Proceed to part II