Surveys are not an easy undertaking. Not for lack of answers, but for lack of questions.
Complexity seems to lie in numerical and statistical procedures. However, the best statistician will not get anything out of skewed or faulty data.
Why then are so many surveys so badly written ?
For instance :
- many questions have multiple parts (“what is your opinion about this restaurant’s service?”). How do you know whether the client reacted to the great maître d’h or the horrendous waitress?