As we have noted, the pluralists would deny the existence of interests which are different from expressed preferences. The behaviourist research methods used by the pluralists do not provide grounds for establishing the existence of a false consensus except through the investigation of what happens when the consensus breaks down. The pluralist position is well represented by Polsby, who points to the danger of observers claiming to know the real interests of citizens who do not acknowledge the existence of such interests (1980, p. 224). In Polsby's view, researchers need to be extremely cautious in going beyond people's expressed preferences in analysing power relationships, although he does accept that there may be situations in which people do not act to maximise their values or their interests. Against this position, Lukes maintains that people's expressed preferences are shaped by socialisation, education and the mass media, and their real interests can only be established by examining what they would choose when exercising choice relatively free of these constraints. Put another way, Lukes argues that real interests can only be identified in conditions of relative autonomy. The difficulty with this is how to create such conditions