Observed Behavior Patterns, Noncognitive Human Capital and Misreporting Behavior in Online Surveys

In this study, we examine misreporting behavior in online surveys using field experiments conducted within a large-scale online training program for rural teachers. We link digitally recorded data with survey responses and incorporate randomized controlled trials (RCTs) into the survey design. Noncognitive human capital is measured through both self-reported personality traits and proxies based on observed behavior patterns. Our findings reveal that the influence of observed individual characteristics on misreporting varies depending on the nature of the questions and survey specifics. Unobserved heterogeneity affects both survey participation and response accuracy, leading to sample selectivity bias. Noncognitive human capital inferred from observed behavior patterns consistently demonstrates a significant impact on misreporting, whereas self-reported personality traits are compromised by the same misreporting issues. However, behavior proxies may also capture factors external to survey respondents. Enhancing survey efficiency—for instance, by adjusting question saliency and optimizing question formats—can improve survey quality. Finally, using observed behavior patterns to adjust the raw sample leads to more reliable data. © 2025, The Authors. All rights reserved.