Conducting high-quality, face-to-face, computer-assisted personal interviews is a demanding task that often calls for experienced interviewers holding specialised expertise and the motivation to gather the best possible survey data. Unfortunately, some interviewers may attempt to navigate the challenges of the surveying by resorting to different types of undesirable interviewer behaviour ranging from minor infractions like speeding to more serious transgressions such as partial or even complete data fabrication. Given that even a small number of fabricated interviews can contaminate an entire dataset, it is imperative to swiftly identify the most serious forms of undesirable interviewer behaviour. With ethical and privacy considerations of interviewers and respondents in mind, we developed a novel approach called the Virtual Surrounding Impression (VSI), which allows the gravest forms of undesirable interviewer behaviour to be detected without resorting to actually recording location, audio or video data while taking account of ethical and privacy concerns. We show that the VSI approach enables the detection of instances where a single interviewer conducts multiple interviews at the same location or where the location changes during an interview, signalling possible fabrication, whether partlialy or full.
|
