vmw
Statistics bloody statistics - if I recall correctly all the surveys have been by voluntary participation - therefore can I assume not every user responded. Of those who respond are they weekend warriors or professionals where music is there major source of income?
That's not really relevant, because a large enough sample provides a cross-section that's
representative of SONAR users.
Now I could go on listing a range of questions where the answers could show a particular bias in the results; but that is not important. What is important is how the the questions are asked and how many multiple choice answers there are to select from in response to a neutral question.
There's not much opportunity for "leading answers" when the answers are simply numbers from 1 to 5. For example, with the question "How important is EUCON compatibility to you?" you could choose on a scale ranging from "not important at all" to "crucially important."
A question like "would you recommend Sonar?" - you might like Sonar but there is no caveat like "keep in mind it has low grade notation".
There
is the opportunity to add caveats because there's a comments section. People can say "I would recommend SONAR to anyone except those whose notation needs go beyond basics."
Surveys can be notoriously off the mark - just look at election polls around the world and see how many get it wrong.
Statistics is a science, which is why surveys give a margin of error. The further results fall outside the margin of error, the greater the odds of them being accurate. For example, if a survey indicates 80% of the respondents say "yes" and 20% say "no" with a margin of error of 5%, it's pretty safe to assume the "yes" faction is in the majority. There would be no certainty if it was 52% to 48% with a margin of error of 5%.
If you look at (for example) the results of Gallup's polls for presidential elections, the ones that ended up "wrong" were almost invariably within the margin of error and therefore, Gallup knew upfront the results had no guarantee of accuracy. Surveys can also be
very accurate - especially when they address the same subject over a period of years, which tends to "smooth out" statistical anomalies.