The importance of examining measurement properties in ecological momentary assessment research: An illustrative example in suicide research.

Ecological momentary assessment is a burgeoning methodology within suicide research, allowing for investigations of dynamics in suicidality and its risk factors in naturalistic settings. Less attention, however, has been paid to carefully examining whether measurement invariance is present across participants and prompts. This study examined the measurement properties of items assessing affective states in 237 adults with severe suicidal ideation (Mage = 27.12, 61.2% cisgender women, 86.9% White, 38.4% bisexual/pansexual). Utilizing latent Markov factor analysis, four questions were addressed: (a) how many measurement models (states) underlie the data? (b) how do these states differ? (c) were momentary suicidal intent and lifetime suicide attempts predictors of transitions between states? and (d) for whom does invariance hold? Results indicated that the best-fitting model had three states (R2entropy = .93). State 1 (Demoralization, 59% of observations) consisted of one factor characterized by high distress and lower arousal. State 2 (Agitated Arousal, 28%) consisted of one factor characterized by high distress and arousal. State 3 (Content, 13%) consisted of one factor characterized by low distress and absent arousal. Although these states were very stable across observations, momentary suicidal intent and suicide attempt history predicted transitions. Only a minority of participants (11.1%) remained in the same state throughout the entire study; thus, within-person and between-person invariance were insufficient. These findings underscore the importance of meticulous inspection of measurement properties when conducting intensive longitudinal research, as observations cannot be validly compared if measurement invariance is not met. (PsycInfo Database Record (c) 2025 APA, all rights reserved).