“…Further, some research has indicated that the performance of FCIP is different from transformation methods such as mean‐mean linking (MM; Loyd & Hoover, ), mean‐sigma linking (Marco, ), or characteristic curve linking (Haebara, ; Stocking & Lord, ) and may yield biased results when there is a shift in the population that leads to changes in the latent ability distribution (Baldwin, Baldwin, & Nering, ; Hu, Rogers, & Vukmirovic, ; Keller & Keller, ; Kim, ; Paek & Young, ). This effect occurs most often when only a small number of common items are employed (Arai & Mayekawa, ) and it depends on the software implementation (Kang & Petersen, ; Kim, ; Paek & Young, ). According to Stocking and Lord (SL; ), an alternative to the FCIP procedure is characteristic curve linking.…”