Utilising Computerised Adaptive Testing to Alleviate Respondent Burden in Maternal Stress Assessment

Maternal mental health plays a pivotal role in perinatal care, with far-reaching implications for both maternal well-being and child development. Amidst global challenges like the COVID-19 pandemic, the demand for effective mental health assessment has surged. In response, this study investigates the utility of Computerised Adaptive Testing (CAT) for profiling maternal stress. Using the Growing Up in Singapore Towards Healthy Outcomes (GUSTO) dataset, we focused on the Maternal Stress module, comprising the Perceived Stress Scale (PSS) and Parental Stress Index (PSI). We generated an item pool from both scales and employed the Graded Response Model (GRM) for calibration. A final item bank of 105 items was consolidated. Using the Concerto software, we devised a CAT questionnaire for test administration. We then used the 'Firestar' R package to simulate the CAT with various stopping criteria, revealing substantial question reduction (up to 84.8%) while still maintaining high correlations with true theta scores (up to 99.9%). Nonetheless, limitations include assumptions in simulations, item calibration, and a specific focus on maternal stress. This study underscores CAT's potential to streamline assessments, enhancing perinatal mental health evaluations, while signaling the need to further explore CAT’s potential in this domain.