Context eects in the questionnaire measurement of national stereotypes have been demonstrated in many studies since the early pioneering work of Katz and Braly (1933). More recently, self-categorization approaches (e.g. Oakes, have been used to explain context eects, with the suggestion that variation in stereotype content can be conceptualized as variation in the comparative meaning of categories, as speci®ed by the metacontrast principle. However, the methodological consequences of stereotype context eects for survey responding remain poorly understood. Using an experimental questionnaire manipulation, the current study demonstrates empirically that varying the stereotype rating task presented to respondents can have knock-on eects on responding in other questionnaire modulesÐin this case, modules exploring attitudes to European integration and national and European social identities. It is suggested that stereotype rating tasks might activate some social identities whilst inhibiting the activation of others, and that these eects might be aected by the level of abstraction of the social identities concerned. # 1998 John Wiley & Sons, Ltd.