Many published studies in academic journals, newspapers, and online indicate that survey research (any instrument or procedure that asks questions of respondents) generates conflicting answers. This often occurs when an asking instrument contains numerous questions on a particular topic. Answers to some questions affirm the phenomenon (behavior, opinion, etc.) investigated, whereas other responses imply, or explicitly state, the opposite.
Also, conflicting answers are common when different instruments pose questions about the same phenomenon. For example, a number of survey researchers using different surveys asked about the similarities and differences between public and private organizations. “[T]his research . . . refute[s] widely held . . . assumptions about similarities and differences between public and private organizations but . . . in some ways support[s] such assumptions” [Rainey and Bozeman, “Comparing Public and Private Organizations”, J-Part, 2000, p. 447]. Paraphrasing The Rolling Stones, you can–in survey research, if nowhere else–have it both ways round: sun and moon, both A and ~A.
There are many reasons why survey research generates conflicting answers: different samples, respondents’ biases, question wording, etc. See the above-mentioned Rainey and Bozeman article and an online article about surveys of U.S pet ownership for discussions of reasons for this particular limitation/weakness of survey research.
Regardless why survey research produces conflicting answers, the fact that it does means that you shouldn’t ask to find out what’s really going on. See my book, The Problem with Survey Research, for a complete statement of the limitations/weaknesses of survey research and, in Part Six, brief descriptions of observation and other procedures that produce reliable information. On the latter point, see also my course syllabus, PA 544, Qualitative Methods: How to Find Out What’s Really Going On.