Data integrity in an online world: demonstration of multimodal bot screening tools and considerations for preserving data integrity in two online social and behavioral research studies with marginalized populations

Internet-based studies are widely used in social and behavioral health research, yet bots and fraud from “survey farming” bring significant threats to data integrity. For research centering marginalized communities, data integrity is an ethical imperative, as fraudulent data at a minimum poses a threat to scientific integrity, and worse could even promulgate false, negative stereotypes about the population of interest. Using data from two online surveys of sexual and gender minority populations (young men who have sex with men and transgender women of color) we a) demonstrate the use of online survey techniques to identify and mitigate internet-based fraud, b) differentiate techniques for and identify two different types of “survey farming” (i.e., bots and false responders), and c) demonstrate the consequences of those distinct types of fraud on sample characteristics and statistical inferences, if fraud goes unaddressed. We provide practical recommendations for internet-based studies in psychological, social, and behavioral health research to ensure data integrity and discuss implications for future research testing data integrity techniques.