本研究的目的在利用電腦多媒體的優勢及認知成份分析的明確架構，發展電腦化圖形推理測驗（The Computerized Figure Reasoning Test, CFRT），實徵驗證認知成份架構對測驗效度論述的重要支持意涵。研究中採用八項成份進行CFRT 作業的發展，針對國小五、六年級共498 名學生進行施測，實徵檢驗認知 成份對作業難度的解釋力。研究中操作性界定認知成份的內涵與編碼方式，以多元迴歸進行資料分析，結果顯示八項認知成份模式對實際難度參數的解釋量約為90%，充分顯示認知成份分析對於試題難度來源解釋與操弄的應用潛力。 本研究所提出的圖形推理認知成份架構，可作為未來圖形推理作業自動化產出的基礎，降低電腦化圖形推理題庫資源發展的成本，進一步提高圖形推理能力診斷與介入應用的可能。

The purpose of this study is to propose an eight-component-model to develop a computerized figure reasoning test (CFRT). The components included are distribution of three values rule, distribution of two values rule, constant in column/raw rule, quantitative pairwise progression rule, figure addition or subtraction rule, figure movement or rotation rule, perceptual identification, rule explicitness. The CFRT was administered to 498 5th and 6th graders. The results suggest that the cognitive components included can predict around 90% variances of the difficulty parameters. The cognitive component based test development approach is successfully implemented in this study. The approach demonstrates it’s direct and strong support for construct validity. The eight components investigated in this study might also serve as the basis for further compensatory intervention for students with low figure reasoning ability.