Psychological Distress in Older Adults in Saudi Arabia: A Psychometric Evaluation of the Arabic 12-Item General Health Questionnaire (Ar-GHQ-12)

This study evaluated the psychometric properties of the Arabic 12-Item General Health Ques-tionnaire (Ar-GHQ-12) among older adults in Saudi Arabia, focusing on its factorial structure, internal consistency, and the feasibility of testing measurement invariance across sex. The sample comprised 221 participants aged 55 years or older, randomly divided into non-overlapping sub-samples for exploratory factor analysis (EFA; n = 110) and confirmatory factor analysis (CFA; n = 111). The Initial EFA used principal axis factoring (PAF) with oblimin rotation, followed by per-mutation-based parallel analysis using polychoric correlations and additional exploratory solu-tions. CFA compared five competing measurement models using robust maximum likelihood es-timation (MLR). Although the initial Kaiser criterion indicated three factors, parallel analysis fa-vored a two-dimensional solution; the exploratory solutions; however, showed suboptimal overall fit. In the CFA subsample, the bifactor model had the most favorable global fit indices (robust CFI = .979, TLI = .968, RMSEA = .041, SRMR = .045), while the wording-based two-factor model also showed favorable fit (robust CFI = .963, RMSEA = .049) and was preferred by BIC. Given the modest CFA subsample size and the differing model-selection criteria, these results do not estab-lish the superiority or stability of either factor structure. Bifactor-specific indices (ECV = .541, ω_H = .642, ω = .874, PUC = .545) did not provide sufficient evidence of essential unidimensionality. In-ternal consistency was acceptable to good for the negative-wording (α = .824; ω = .831) and posi-tive-wording (α = .758; ω = .760) item groups, although these estimates do not establish that the groups represent distinct psychological subscales. Measurement invariance across sex could not be reliably evaluated because the multigroup bifactor model produced an inadmissible solution. These findings provide preliminary, sample-specific psychometric evidence for the Ar-GHQ-12 among older adults but do not establish a definitive factorial structure or essential unidimension-ality. Further validation in larger, independent samples, including assessment of criterion-related validity and measurement invariance, is needed before clinical screening recommendations can be made.