Psychometric evaluation of ecological momentary assessment items for mood in a non-clinical sample

Background.Smartphone-based Ecological Momentary Assessment (EMA) enables the real-time measurement of emotional states in daily life, reducing recall bias and capturing clinically meaningful fluctuations. However, evidence regarding the reliability and validity of EMA measures remains limited, and validated instruments are scarce, highlighting the need for EMA-specific psychometric evaluation.Objective.To assess the reliability, validity, and structural characteristics of a brief 11-item smartphone-based EMA of mood in a non-clinical sample.Methods.We used data from a randomized controlled trial evaluating a one-week digital self-efficacy training in 93 stressed Swiss university students. Baseline psychometric assessments included the Beck Depression Inventory II (BDI II), the Positive and Negative Affect Schedule (PANAS), the General Self-Efficacy Scale (GSE), the State and Trait Anxiety Inventory (STAI), and the Perceived Stress Scale (PSS). The EMA assessment included moods such as cheerful, irritated, anxious, happy, insecure, lonely, relaxed, sad, thoughtful, focused, and stressed. Analyses included descriptive statistics, internal consistency (Cronbach’s alpha), and external validity (correlations between baseline questionnaires and participant-level aggregated EMA ratings from the first 24 hours). Exploratory and confirmatory factor analyses and network analyses assessed the structure.Results.We considered the data of all 93 participants for the analysis. Participants (78.5% female) were on average 23.27 years of age (SD = 3.49). EMA items showed normal distribution, good internal consistency (α = 0.88), and low correlations (0.19-0.39) with the BDI II, the PANAS (positive affect subscale), and the GSE. Moderate correlations (0.40–0.48) were found with the PANAS negative affect subscale, the STAI, and the PSS. An exploratory factor analysis indicated two or three factors, while network analysis revealed positive and negative affect communities. Confirmatory analysis supported the network model as best fit (CFI = 0.99; TLI = 0.99; RMSEA = 0.01).Conclusion.Smartphone-based EMA of mood item set showed strong psychometric properties and distinguished positive from negative affect as well as depressive from anxious factors, supporting its use as a scalable tool for monitoring transient mood states in non-clinical samples.