This IRT approach has multiple advantages. First, within each cohort there is increased measurement reliability, because when multiple inventories have been administered to the same individual, scores can be estimated using the items from all relevant inventories. In addition, items can be differentially and optimally weighted if necessary, and items that do not fit the measurement model can be identified and omitted, thereby increasing power. Subgroups of individuals that were assessed with only a subset of items can now also be included in the study. Moreover, the IRT approach can statistically evaluate the extent to which different inventories actually measure the same construct. Lastly, IRT enables researchers to determine the extent of measurement invariance across cohorts: can scores across cohorts be quantitatively compared and therefore pooled and meaningfully used in a meta-analysis?
