Deviant interviewer behavior is a potential hazard of interviewer-administered surveys, with interviewers fabricating entire interviews as the most severe form. Various statistical methods (e.g., cluster analysis) have been proposed to detect falsifiers. These methods often rely on falsification indicators aiming to measure differences between real and falsified data. However, due to a lack of real-world data, empirical evaluations and comparisons of different statistical methods and falsification indicators are scarce. Using a large-scale nationally representative refugee survey in Germany with known fraudulent interviews, this study tests, evaluates, and compares statistical methods for identifying falsified data. We investigate the use of new and existing falsification indicators as well as multivariate detection methods for combining them. Additionally, we introduce a new and easy-to-use multivariate detection method that overcomes practical limitations of previous methods. We find that the vast majority of used falsification indicators successfully measure differences between falsifiers and nonfalsifiers, with the newly proposed falsification indicators outperforming some existing indicators. Furthermore, different multivariate detection methods perform similarly well in detecting the falsifiers.