本研究主要在探討有無補習之高年級低成就學生在數學文字題學習成就上的差異，與其學校及補習班教師教學型態對數學文字題的影響。本研究之五、六年級低成就生分別接受該年級之數學文字題測驗，該測驗由低、中、高閱讀理解難度文字題組成，有補習低成就生的補習班與學校教師分別接受教學型態量表。研究結果主要發現如下：（一）補習與未補習之五年級數學低成就生在數學文字題測驗總分及其學習保留量皆沒有明顯差異；但有補習的六年級低成就生數學文字題總表現顯著高於沒有補習者，並較具學習保留效果；（二）補習僅對於六年級數學低成就生在低、中閱讀難度文字題得分有顯著的幫助。（三）對五年級低成就生而言，補習班教師愈傾向使用建構教學取向，有補習者的數學文字題成就總分及中、高難度文字題的作答表現則越高，而其直接教學取向對中難度題具有負向預測力；在六年級部分，智力對六年級低成就生的數學文字題總分有正向預測力，而教學型態則不具預測力。但對低閱讀難度的數學文字題而言，智力與學校教師越不採用直接教學取向的程度可共同預測低成就生數學文字題低難度題得分，其解釋力為 20%；而補習天數對五、六年級數學低成就生的數學文字題測驗總分皆沒有顯著預測力。

This study investigates differences in mathematics word-problem solving performance among 5th and 6th grade mathematics underachievers with and without cram schooling, and the effect of their elementary school teachers’ and cram school teachers’ teaching approaches on word-problem solving performance. The word-problem test is comprised of three parts, including problems requiring low, average, and high level of reading comprehension. Participants were given word-problem tests appropriate to their grade level. Students’ mathematics teachers in cram schools and elementary schools were given a questionnaire on their teaching approaches. First, there were no significant differences in math word-problem solving performance and learning retention between 5th grade mathematics underachievers with and without cram schooling. However, word-problem solving performance and learning retention of the 6th grade underachievers with cram schooling were significantly better than those without. Regression analysis of math word-problem performance in 5th grade underachievers revealed that the more cram school teachers used a constructivist teaching approach, the higher their scores were on the total, average, and high level math word-problem tests. In contrast, the more cram school teachers used the direct teaching approach, the lower scores math underachievers got on the average level word-problem tests. The 6th grade math underachievers’ performance can be explained by their intelligence. Moreover, the direct teaching approach was demonstrated to have a negative effect on underachievers’ performance solving low level word-problems. In addition, word-problem solving performance was not related to the number of hours per week they learned mathematics in cram schools.