本研究的主要目的是探討目標層次、回饋訊息對國中生數學工作表現及學習動機的影響，是否會隨著受試者控制信念的不同而不同。為完成研究目的，本研究共進行了兩次預試與四次的實驗操弄。在實驗時，研究者將受試者隨機分派成四個實驗組：明確且困難目標/回饋訊息、明確且困難目標/無回饋訊息、模糊目標/回饋訊息、模糊目標/無回饋訊息，並依受試在國中生「數學學業成就責任量表」上的得分，取全體受試之前、後33％為內、外控者，總計樣本共有230位。蒐集所得的資料，本研究針對數學工作表現進行三因子共變數分析，並針對數學學習動機進行四因子混合設計變異數分析。

研究結果顯示，在數學工作表現方面，當受試者傾向內控，無論是接受明確且困難目標或是回饋訊息時，都會有較佳的數學工作表現。在學習動機方面：（1）在工作價值上，當操弄為指派明確且困難目標時，接受回饋訊息的受試者之工作價值高於無接受回饋訊息者。（2）在自我效能上，當操弄為指派明確且困難目標時，傾向內控的受試者之自我效能高於外控者。（3）在期望成功上，當操弄為指派明確且困難目標且附加回饋訊息時，傾向內控的受試者之期望成功會高於傾向外控者；其次，當操弄為指派明確且困難目標但無附加回饋訊息時，傾向外控的受試者其期望成功反而下降。（4）在正向情感上，當操弄為指派明確且困難目標時，傾向內控的受試者之正向情感高於外控者。（5）在負向情感上，當操弄為指派明確且困難目標時，傾向內控的受試者之負向情感也會低於傾向外控者。本研究根據研究結果在實際課室學習上的應用進行討論，並提出未來研究上的建議。

The purposes of this study were to explore whether the effecst of goal levels, feedback information on junior high school students’ performance of mathematic task and learning motivation were moderated by locus of control. For the purposes of this study, two trial-outs and three experiments were administrated. Subjects were sampled and assigned randomly to four experimental groups: specific and difficult goal/feedback information, specific and difficult goal/no feedback information, ambiguous goal/feedback information, ambiguous goal /no feedback information. To verify the hypotheses of this study, collected data were analyzed by a three-way ANCOVA and a four-way mixed design ANOVA.

Results showed that there was a significant three-way interaction among goal levels, feedback information and locus of control. Moreover, for learning motivation:(a)a three-way interaction among feedback information, locus of control and measuring stages on task value was significant, (b)a two-way interaction between goal levels and locus of control on self-efficacy was significant, (c)a four-way interaction among goal levels, feedback information, locus of control and measuring stages on expectancy for success was significant, (d)a two-way interaction between goal levels and locus of control on positive affect was significant, (e)a three-way interaction among goal levels, locus of control and measuring stages on negative affect was significant. There was a significant three-way interaction among feedback information, locus of control and measuring stages on negative affect. Finally, the research discussed how to apply the results in classroom teaching practice and provided suggestions for future research.