Participants in a study concerning social attitudes toward robots were randomly assigned a questionnaire form displaying a non-, semi- or highly anthropomorphic robot as a hidden intervention. Results indicate that asking people about their attitudes toward "robots" in general -- as done in some studies -- is questionable, given that (a) outcomes can vary significantly depending on the type of robot they have in mind, and (b) it is therefore easy to intentionally or unintentionally manipulate results by priming respondents with positive or negative examples.