本研究旨在發展K-3學童適用之記憶力測驗，作為學習困難初步的診斷工具。測驗 內容包含（１）語詞記憶（立即與延宕）；（２）數字序列記憶；（３）物體辨識（立即與 延宕）及（４）工作記憶等四部份。常模樣本包含 60 所學校 K-3 年級共 1,235 名學童， 另外以資源班 440 名及啟智班 339 名學童作為初步效度研究的對照討論樣本。一、三、四 子測驗的平均（點）二系列相關依序為.55，.61 和.41；內部一致性α為.75，.85 和.64。 語詞記憶有明顯的順序效應；物體辨識測驗中動物的獨特性和受試學童喜愛的程度有些微的 影響；工作記憶變異較大，因此它與其他子測驗的相關也較高（.34 至.46 ）。各子測驗男 女學童的差異並不明顯，其中語詞記憶和數字記憶廣度，女學童比男學童略高，但解釋量低 於 1%。 各子測驗年級間的差異都是顯著的，由幼稚園到三年級平均得分逐年上揚，年級變 項對各子測驗變異的解釋量依序為 7.5%，7.8%，11.6% 和 15.4%。 大致看來作業內涵愈複 雜，年級的差異就愈明顯。常模、資源班和啟智班的差異幅度更大，班別對各子測驗變異的 解釋量依序為 19.3%，17.0%，17.8% 及 25.2%。 其中年級和班別的交互作用在物體辨識和 工作記憶測驗上是顯著的，即各類班別隨年級上揚的幅度並不平行，換言之各類學童成長的 速率有差異。研究中同時提出各子測驗疑似記憶薄弱的決斷分數，以為進一步協助必要性的 警示。　

　　　　　The purpose of this study is to develop a K-3 children memory test (CMT) to diagnose the characteristics of learning disable students. The facets of short term, long term and working memory are all included. There are 4 sub-scales in CMT: (1) vocabulary memory; (2) digit span; (3) animal identification; and (4)working memory. The norm is based on 1235 K-3 children selected from 60 schools. Another 440 learning disabled children and 339 mentally retarded children are also included as the contrast groups for the discussion of discriminatory validity. The psychometric information of each sub-scale is presented. Since some of the stimuli are meaningful and relevant to children's experience, the cognitive issues of sequential effect, familiarity, or preference effect are analyzed as part of the preliminary construct validity discussion. The gender differences are very minor (less than 1% variance accounted). The grade differences are substantially significant. The variances accounted by grades are 7.5%, 7.8%, 11.6% and 15.4% for the four sub-scales respectively. The differences among contrast groups (norm, LD and MR) are even larger. The variances accounted by groups for the sub-tests are 19%, 17%, 18% and 25% respectively. The interaction effects between grades and groups are significant for animal identification and working memory tests. The grade differences are not parallel for the three groups. In other words, the increasing sizes between neighboring grades are not the same for children of subtypes. The cut-off point of each sub-scale is also provided for the screen purpose.