Researchers have long worried about a phenomenon where study participants give higher ratings on self-report scales the first time they take a survey compared to subsequent times, particularly for negative subjective experiences. Recent experimental evidence, using samples of U.S. college students, suggests that this initial elevation phenomenon is due to an upward bias in people’s initial responses. Such bias potentially undermines the validity of many research findings. However, more recent studies have found little evidence in support of the phenomenon. To investigate the robustness of the initial elevation phenomenon, we conducted the largest experiments to date in diverse online samples (N = 5,285 across three studies, from Prolific.co). We observed an initial elevation on self-reports of negative subjective experiences such as mood and mental and physical health symptoms. Our findings show that the threats to validity posed by the phenomenon are real and need to be reckoned with.