Sampling methodology and observed data patterns in ecological momentary assessment: A study of nonsuicidal self-injury among high-risk young adults.

Prior research using ecological momentary assessment (EMA) has been mixed in comparing signal- and event-contingent protocols' influence on research findings. No work has compared these approaches for high-severity, low-frequency behaviors for which event-contingent EMA is recommended, such as nonsuicidal self-injury (NSSI). We examined EMA data from 36 community-recruited, high-risk young adults; participants completed 1 week each of signal- and event-contingent EMA in randomized order. Event-contingent methods evinced higher NSSI urge mean and variability, but fewer total NSSI behaviors, compared to signal-contingent methods. NSSI correlates, such as suicidal ideation and negative affect, were more highly endorsed, but no more variable, across event- versus signal-contingent surveys. Associations between constructs at the same survey varied in strength by protocol type but inconsistently across examined predictors. Given the small sample size, and risk of Type 2 error, replication is needed in larger, more diverse samples. Results highlight the substantive impact of EMA sampling approach on observed data characteristics and associations, with implications for future EMA research examining high-severity, low-frequency behaviors. (PsycInfo Database Record (c) 2026 APA, all rights reserved).