Development of a Holistic Mental Health Item Bank for Use in the General Population

Abstract There is growing interest in the mental health of the public, but a lack of attention to measurement of this complex construct. We report on the development of a novel item bank fit for this purpose. Grounded in the dual continua model, this study aimed to (i) construct an item pool measuring deficit-based and strengths-based mental health constructs, (ii) explore its dimensionality and (iii) develop an item bank using item response theory (IRT). We argue that a holistic measure has optimal construct validity in the Australian general population, while minimising response burden. The present study was conducted in two phases. Phase 1 involved adapting and standardising items from existing validated instruments, resulting in 146 items. Phase 2 had two parts: (2A) exploratory factor analysis using a cross-sectional availability sample ( N =527, 78.8% female) to identify a provisional factor structure with 107 items; (2B) IRT calibration using a cross-sectional survey-panel sample matched to an Australian census by sex and age ( N =699, 50.6% female), resulting in a final 54-item item bank. A bifactor graded response model with a general factor ( General Mental Health ) and two specific factors ( Struggling and Thriving ) was fitted (RMSEA=0.04, TLI=0.99, CFI=0.99) and was generalisable to two other datasets. Provisional validation showed that the model is psychometrically sound with minimal item biases. Within the study’s limitations, such as matching only sex and age in the census-matched sample, the model appeared consistent with theoretical frameworks and other empirical studies. The item bank and its associated models provide a macro-level characterisation of individuals’ general mental health with a single standardised score, which may reduce the burden in administering and monitoring mental health longitudinally. Future studies should further explore the item bank’s psychometric properties and optimisation, such as developing computerised adaptive tests or short-form measures.