In opinion surveys it is tacitly assumed that all respondents have opinions about the questions which are asked them and that all answers are based on well-formed attitudes. This study critically assesses these assumptions, both from a psychological perspective on the cognitive structuring of attitudes and from a sociological perspective on the social impact of different ways of expressing opinions. It is found that holding attitudes about the political and the medical domains is related to the social and cultural dynamics of these fields and of society in general. Hence, the opinion survey is not the neutral method it is usually taken for. It is necessary to search for other methods which can bring different aspects of opinions to the fore than those looked at in surveys.