The EDI-5 as an indicator of emotional distress: Instrument design and validation

From a sociology of health perspective, emotional distress is best approached as a dimensional population construct that goes beyond clinical diagnosis. This article presents the rationale, design, and validation of the EDI-5 (Emotional Distress Indicator), a brief unidimensional instrument (5 items) capturing core negative affect (tension, low morale, discouragement/depressed mood) while incorporating loneliness as a relational component and a reverse-keyed calm/tranquility item as a response-control marker. The scale was administered to three independent samples of students aged 14–18 years from the Canary Islands, Spain ( n = 287, 665, 1,168). Confirmatory factor analyses using WLSMV supported an essentially one-factor structure with good fit across samples (CFI = .993–.995; TLI = .986–.991; SRMR = .016–.023; RMSEA = .068–.082) and acceptable reliability (α = .768–.812; ω = .788–.824). Measurement invariance tests supported configural comparability across sex, age group, and samples, while full scalar invariance was not cleanly supported across sex and contexts. External validity evidence showed coherent associations with related constructs, including moderate-to-strong correlations with anxiety symptoms (GAD-7) and sleep difficulties, and negative associations with well-being indicators. As a benchmarking exercise, ROC analyses against probable moderate-severe anxiety (GAD-7 ≥ 10) showed good discrimination (AUC = 0.886 and 0.820), supporting a provisional general reference threshold of ≥ 15, with sex-stratified values suggesting ≥ 15 for boys and ≥ 16 for girls. Owing to its brevity and conceptual clarity, the EDI-5 is suitable for population monitoring and first-pass identification of elevated distress in educational and community settings, to be followed by more specific assessments when indicated. • EDI-6: a brief, unidimensional index of distress. • Grounded in stress process and allostatic load frameworks. • Validated in three youth samples (N=2,120) with one-factor CFA. • Reliable (α, ω ≥ .80) and partly invariant by sex, age, samples. • Strong construct validity and effective ROC cut-offs (≥18).