Exploring the structural characteristics of the Adult Social Care Outcomes Toolkit (ASCOT) and ASCOT-Carer [version 1]

Measurement models inform the approach to assess a measure's validity and also how a measure is understood, applied and interpreted. With preference-based measures (PBMs), it is generally accepted that they are ; however, if they are applied without preferences, they may be , or . In this study, we sought to empirically test whether the , or measurement model best describes PBMs of social care-related quality of life (SCRQoL) - specifically, the ASCOT and ASCOT-Carer. We also explored the network approach, as an alternative. ASCOT and ASCOT-Carer data were analyzed using confirmatory factor analysis and Multiple Indicators Multiple Causes models to test reflective, formative or mixed measurement models, respectively. Network analysis of partial correlations using the Gaussian graphical model was also conducted. The results indicated that the reflective measurement model is the worst fit for ASCOT and ASCOT-Carer. The formative or mixed measurement models may apply to ASCOT. The mixed measurement model was the best fit for ASCOT-Carer. The network analysis indicated that the most important or influential items were and (ASCOT) and and (ASCOT-Carer). The ASCOT and ASCOT-Carer are best described as formative/mixed or mixed measurement models, respectively. These findings may guide the approach to the validation of cross-culturally adapted and translated versions. Specifically, we recommend that EFA be applied to establish structural characteristics, especially if the measure will be applied as a PBM as a measure of SCRQoL. Network analysis may also provide further useful insights into structural characteristics.