Attitudes towards inclusive education typically are measured by questionnaires. Self-reported attitudes are prone to various biases, however, that often are neglected in the interpretation of survey outcomes. The present study investigated two biasing influences experimentally: social desirability and question order. A contrast effect occurred. Global support for inclusive education was higher after participants had rated an item specifically asking about inclusion of a child with physical and intellectual disability, but only when the design of the questionnaire suggested that the institution allegedly conducting the survey was critical towards inclusive education. Expressed support for inclusion was generally higher when the institution allegedly conducting the survey appeared to support inclusive education and this support did not depend on question order. These findings show how fragile attitude measures can be. Several biases threaten their validity and must be taken into account in the design as well as interpretation of surveys.