Without prentender that this list is exhaustive, will point out some ways to identify errors in questions or parts of a questionnaire.
- The questions used language or sociological categories that produce that respondents answered that they believe is expected to respond or misinterpret questions.
- The respondent believes that must be identify - for example in a closed with options question - what is the correct answer. Not involved in the sense of expressing their views by an error in the way of asking and sometimes in the mode of lead pollster. He believes that it is like a game that has to guess the correct answer option.
- Questions closed with options options are not mutually exclusive, either not exhausted all possibilities for response. So the alternatives are mutually independent and do not leave any aspect sociologically relevant is convenient to pilot test these questions (or the questionnaire). Occurs when a survey or a question has many instances for people that are not decided by any of the suggested alternatives, opting not to respond, point out missing some option or make verbalizaciones as "it depends" or "the question is poorly worded", or the "don't know" which isn't really a does not know. Sometimes reviewers and pollsters are right on wrong questions.
- The answers are elusive and notoriously false. It happens on sensitive issues, to find the right way to ask and have questions in the questionnaire. Also by the misuse of certain words in the question that lead to the above-mentioned responses by respondents.
- There is an overwhelming majority of favourable response or adverse responses.
- Redundancy: wasted resources asking something they already know.
These are some ways to identify mistakes in the questions of the questionnaires of opinion polls from the responses obtained after the field work.