The inaccuracy of data from online surveys: A cautionary analysis

Online recruitment methods for survey-based studies have become increasingly common in social science research. However, they are susceptible to a high rate of participation by fraudulent research subjects. The current study identified fraudulent (i.e., “fake”) participants in an online research study of parents of 13 to 18-year-old adolescents, and compared demographic, anthropometric, and subjective health data between “fake” (N = 1084) and “real” (N = 197) participants. Of 1,281 subjects who started the eligibility survey, 84.6% were coded as “fake.” “Fake” participants were less diverse in race/ethnicity and more diverse in gender. Their depression symptoms were inflated, but ratings of perceived health were comparable to “real” participants. Well-established correlations, such as that between BMI and perceived health, were not replicated with “fake” participants. Online surveys are highly vulnerable to fraudulent research subjects whose participation compromises the validity and interpretability of results. The discussion provides a guide and recommendations for improving data quality in online survey research.