Uncovering Intentional Response Distortion: A Digital Behavioral Biometric Approach

Abstract

Self-report surveys are widely used in various contexts, but they are prone to methodological and measurement issues that can skew results and compromise data integrity. Inconsistencies in responses can lead to measurement bias, affecting a measure’s ability to accurately capture theoretical constructs. Unlike knowledge assessments—such as math quizzesor grammar tests—which have clear right or wrong answers, self-report surveys—such as health and wellness assessments, political polls, and pre-employment application assessments—do not have definitive correct answers. This inherent ambiguity, combined with methodological and measurement concerns, makes surveys vulnerable to response distortion, particularly impression management, where individuals consciously alter their responses to fit perceived social norms or expectations. Despite extensive research intended to identify intentional response distortion, current approaches fall short in robustly detecting and predicting this risk. This dissertation consists of three essays that use digital behavioral biometric techniques to detect and predict response distortion. Digital behavioral biometric data—such as response times, mouse speeds and distances, and answer-switching patterns—provide fine-grained, realtime insights into respondents’ decision-making processes. Analyzing these data across three contexts reveals that how respondents formulate their self-reported answers can reveal response distortion. The findings offer a novel approach to improve data quality and counteract the drawbacks of survey measurement errors. This research extends traditional theoretical boundary conditions about response distortion behaviors, and it proposes a low-cost, scalable artifact that organizations could implement in self-report questionnaires. Thus, the research makes important contributions to the information systems (IS) literature and offers promising practical benefits.