| http://www.w3.org/ns/prov#value | - For 10 items, we collapsed response categories to achieve a minimal count in any response category of three or to achieve fit to a rank-scaled IRT model, excluding items because of local dependence, insufficient item fit, and DIF. An additional 44 items were excluded, including 22 items that were not intended for calibration (they were device, experimental, or performance items),21 other items, an
|
