心理科學領域內的客觀測量：Rasch模型之特點及發展趨勢

Rasch模型是在國外學術界受到廣泛關注和深入研究的一個潛在特質模型。該模型為解決心理科學領域內測量的客觀性問題提供了一個可行性很高的解決方案。而國內關於Rasch模型的理論探討和應用研究卻並不多見。不同於一般項目反應理論，Rasch模型要求所收集的數據必須符合模型的先驗要求，而不是使用不同的參數去適應數據的特點。Rasch模型的主要特點 (包括個體與題目共用標尺、線性數據、參數分離) 確保了客觀測量的實現。未來關於Rasch模型的研究方向包括多維度Rasch模型、測驗的等值與鏈接、計算機自適應性考試，大型應用測量系統 (比如Lexile系統) 等等。
Rasch model is a latent trait model which has drawn international interest among researchers. It provides a promising solution to ensure the objective measurement in psychological science. However, the research and applicatoin of Rasch model are not as popular as expected among domestic scholars. Unlike general IRTs that adopt a "the model fits data" position and use different parameters to accommodate the idiosyncrasies of the data set, the Rasch model requires that "data fit the model". Its unique features including the same metric shared by persons and items, data linearity, and parameter separation ensure the achievement of objective measurement. The foci of future development of Rasch model include multidimensional Rasch model, test equating and linking, computer adaptive testing, and Rasch-based measurement system such as Lexile framework. Copyright © 2010 中國科學院心理研究所.