Survey respondents don’t always take surveys as seriously as researchers would like. Sometimes, they provide intentionally untrue, extreme responses. Other times, they skip items or fill in random patterns. We might be tempted to think this just introduces some random error into the estimates, but these responses can have undue effects on estimates of the wellbeing and risk of minoritized populations, such as racially and sexually minoritized youth. Over the past decade, and with a focus on youth who identify as lesbian, gay, bisexual, or questioning (LGBQ), a variety of data-validity screening techniques have been employed in attempts to scrub datasets of “mischievous responders,” youths who systematically provide extreme and untrue responses to outcome items and who tend to falsely report being LGBQ. In this talk, I discuss how mischievous responders—and invalid responses, more generally—can perpetuate narratives of heightened risk, rather than those of greater resilience in the face of obstacles, for LGBQ youth. The talk will review several recent and ongoing studies using pre-registration and replication to test how invalid data affect LGBQ-heterosexual disparities on a wide range of outcomes. Key findings include: (1) potentially invalid responders inflate some (but not all) LGBQ–heterosexual disparities; (2) this is true more among boys than girls; (3) low-incidence outcomes (e.g., heroin use) are particularly susceptible to bias; and (4) the method for detection and mitigation affects the estimates. Yet, these methods do not solve all data validity concerns, and their limitations are discussed. While the empirical focus of this talk is on LGBQ youth, the issues and methods discussed are relevant to research on other minoritized groups and youth generally, and speak to survey development, methodology, and the robustness and transparency of research.