本研究以多階段動態評量為基礎結合教學流程來發展國小二?驟解題補救教學模式，並探討其教學現場運作效益。篩選後正式對象共42人，採不同處理組間實驗設計，動態評量組接受漸進提示與解題歷程導向補救教學之連續中介，直接教學組接受解題歷程導向補救教學；參照組則給予自我練習。結果發現，藉由中介階段解題表現區分及組型分析，動態評量不僅能支援補救教學所需之試探篩選、診斷處方功能，且可針對個體學習弱處進行後續補救教學中介。去除前測影響後，除兩個處理組之補救教學促進效益均明顯優於參照組外；動態評量組亦顯著優於直接教學組，效果值與關聯強度均達大幅水準，且具13%連續中介效益之累積加成解釋變異，顯示多階段動態評量更能發揮補救效益。

The researcher wants to investigate the feasibility and effectiveness of Multi-Stage Dynamic Assessment (MSDA) model be applied to develop alternative remedial instruction model for mathematical problem solving. The study utilizes a multi-group experimental design. The participants include 42 students are screened out by pre-testing, and assigned 14 to each of the three groups. The first group is mediated by multi-session dynamic assessment (DA), the second group is given the direct instruction (DI), the third group takes self-practice. Then all groups accept post-testing. Next, the data are analyzed by adequate statistical methods. The effectiveness of ability screening, performance discrimination, deficiency diagnosis and prescription for subsequent remedial instruction are found in the DA group. The simple gain scores of the first and second group are greater than the third group has significantly, after all the performances of the DA group are greater than the DI group has significantly, too. The effectiveness of cumulative facilitation are found in the DA group also.