UC Berkeley

BACKGROUND: Suicide treatment research has placed major emphasis on preventing behavior, an observable phenomenon. Suicidal ideation (SI), however, is a distressing experience that annually impacts >15 million Americans; yet the construct validity of SI has not been well established, limiting empirical attention. METHOD: We compared the coherence of three SI items (diminished wish to live, urge to die/controllability, suicide intent/expectations) and four related items across three adult samples (N = 314) that differed based on current SI severity and assessment time-scales (retrospective versus momentary). RESULTS: A measurement model with scalar invariance had acceptable fit (CFI = 0.969, RMSEA = 0.076), indicating that the SI items consistently loaded onto one construct across samples. The structural model with metric invariance also had acceptable fit (CFI = 0.956, RMSEA = 0.063) and three of four related items (burdensome, hopelessness, and fatigue, but not agitation) were associated with SI across samples. LIMITATIONS: Use of existing data limited the items considered. CONCLUSIONS: Suicidal ideation is a coherent construct independent of the items used to assess it, the assessments time-scale (retrospective versus momentary), or the severity of thoughts, and is worthy of greater empirical, clinical, and policy attention.