Citizens’ survey responses can be unreliable and influenced by perceived expectations. Researchers must understand this behaviour, clearly set the context, and define key terms to improve corruption measurement.