Online surveys: lessons learned in detecting and protecting against insincerity and bots

Online survey panels (e.g., KnowledgePanel) and large-scale survey programs (e.g., World Values Survey) commonly used by social scientists may not support research on rare populations. For researchers interested in accessing specialized populations, openly accessible invitation links conducted in web mode may be necessary. However, surveys recruiting through online channels and administered online are vulnerable to infiltration by bots and insincere responses, and developing procedures and strategies for identifying and filtering out fraudulent responses is a pressing concern for this typeof survey research. This article describes our experience with an infiltrated sample, and provides a multifaceted data cleaning process that can be used to identify insincereand bot responses. Insincere and bot responses are becoming more sophisticated and less likely to be automatically detected by online survey vendors. Employing a multifaceted systematic methodology during the data cleaning stage, is recommended to ensure the validity and integrity of open-access survey data collected online.