Large-scale assessments (LSAs) rely on matrix-sampling designs that generate substantial planned missing item-response data, necessitating robust
imputation methods to support valid population inference. In this study, we
report on a two-step impu...