Asking instruments (polls, surveys, interviews, focus groups, and all other types of asking) produce unreliable answers; that is, answers that may, or may not, be accurate. The only way to know if answers are accurate is to check or verify them with data from non-asking sources; for instance, from observations or experiments. Askers/survey researchers do not have this type of data; all they have is unreliable information.
Supporting evidence that asking instruments make answers unreliable is provided by numerous studies demonstrating (1) that asking instruments produce symbolic and unrealistic answers, (2) that each instrument produces different results for the same question, and that they (3) often generate inconsistent or conflicting answers, (4) much nonresponse and, with rare exception, (5) unrepresentative results.
Discussions and further documentation of how asking instruments make answers unreliable, as well as a full accounting of all the fatal flaws of survey research, can be found in my book, The Problem with Survey Research.