The purpose of this study was to compare Student Teams Achievement Division (STAD) with Traditional Pedagogy for the effects on third-grade students with respect to math achievement, problem solving ability, math learning motivation, learning satisfaction, and interest in math learning. The quasi-experimental design was utilized for this study. The study data were collected through questionnaire survey and interviews. The subjects of the study were 61 third-grade students from two classes of an elementary school in the Kaohsiung City. One class was chosen as the experimental group in which STAD was employed in the experimental instruction and the other was chosen as the control group in which traditional pedagogy was adopted. Data were collected during the period of experimental instruction and were analyzed afterwards. The main results are presented as follows: 1. The math achievement of students who received STAD method was significantly higher after the experimental instruction than before. As for the improvement at posttest from pretest in math achievement, students in STAD group performed better than students in traditional pedagogy group did, but the difference did not reach the significant level. 2. The problem solving ability of students who received STAD method were significantly higher after the experimental instruction than before. As for the improvement at posttest from pretest in problem solving ability, students in STAD group performed significantly better than students in traditional pedagogy group did, which meant the STAD group made much more improvement than the traditional pedagogy group did after the experimental instruction. 3. After the experimental instruction, students who received STAD method performed better in math problem solving interview than those who received traditional pedagogy did. Students who received STAD method were more capable of understanding the questions without interviewer¡¦s explanations. Compared with the control group, the STAD group gave correct solutions more frequently, and was able to provide more reasonable explanations to their solutions. Besides, the STAD group was willing to try various ways to solve the same problem. 4. After the experimental instruction, students who received STAD method had significantly higher math learning motivation than did before; as for the improvement at posttest from pretest in math learning motivation, students in STAD group also performed better than students in traditional pedagogy group. 5. After the experimental instruction, students who received STAD method had significantly higher math learning interests than did before; as for the improvement at posttest from pretest in math learning interests, students in STAD group also performed better than students in traditional pedagogy group. Finally, the researcher proposed several suggestions for the ducational application in classroom teaching and future studies.

本研究主要目的是以文獻分析、問卷調查、潛在成長模式等方法探討數學焦慮、數學學習動機與數學學業成就等三個變項的縱貫模式及因果結構模式。根據五百二十九位國三學生所填的五波調查問卷資料，進行兩部分研究。 研究一乃在估算數學焦慮、數學學習動機、數學學業成就的潛在改變量模式及 數學焦慮、數學學習動機、數學學業成就兩兩之間的因果結構模式。結果發現1.數學焦慮縱貫模式上，符合「焦慮遞增理論」；2.數學學習動機縱貫模式上，符合「動機先升後降理論」；3.數學學業成就縱貫模式上，符合「成就先升後降理論」； 4.數學焦慮與數學學習動機因果結構模式上，符合「動機焦慮交互激發效果理論」；5.數學焦慮與數學學業成就因果結構模式上，符合「焦慮成就交互抑制效果理論」；6.數學學習動機與數學學業成就因果結構模式上，符合「動機成就交互激發效果理論」。 研究二則依據研究一二變項因果結構統計驗證成立的三套理論，建構出三變項因果結構六模式，並驗證有無中介效果。結果顯示，1.前焦慮、中動機、後學業成就因果模式上，符合「動機未完全中介前焦慮、後學業成就理論」；2. 前動機、中焦慮、後學業成就因果模式上，符合「焦慮未完全中介前動機、後學業成就理論」；3.前焦慮、中學業成就、後動機因果模式上，符合「學業成就未完全中介前焦慮、後動機理論」；4.前學業成就、中焦慮、後動機因果模式上，符合「焦慮未完全中介前學業成就、後動機理論」；5. 前動機、中學業成就、後焦慮因果模式上，符合「學業成就未完全中介前動機、後焦慮理論」；6.前學業成就、中動機、後焦慮因果模式上，符合「動機未完全中介前學業成就、後焦慮理論」。 除了上述結果外，研究也對數學焦慮、數學學習動機與數學學業成就縱貫模式的趨勢與時間效果量，國三學生轉折點界定，數學焦慮、數學學習動機與數學學業成就適當模式產出及個別中介角色剖析有深入探討。 / This study, using literature review, questionnaire survey, and latent growth model, investigated the longitudinal model and causal model among math anxiety, learning motivation, and academic achievement. After collecting 529 students with 5 waves, I conducted two studies. The purpose of study 1 was to estimate the latent change model and the causal model of math anxiety, learning motivation, and academic achievement. Results showed that 1. The anxiety increasing theory was supported by the math anxiety longitudinal model. 2. The first increasing then decreasing theory was supported by the math learning motivation longitudinal model. 3. The first increasing then decreasing theory was supported by the math academic achievement longitudinal model. 4. The reciprocal activated effect theory was supported by the math anxiety and learning motivation causal models. 5. The reciprocal inhibitive effect theory was supported by the math anxiety and academic achievement causal models. 6. The reciprocal activated effect theory was supported by the math learning motivation and academic achievement causal models. According to the final three theories stated above and proved by study 1, I constructed the six causal models in order to verify the mediated effects of math anxiety, math learning motivation, and math academic achievement. I found that 1. Math learning motivation did not mediate fully the effects on early math anxiety and late math academic achievement. 2. Math anxiety did not mediate fully the effects on early math learning motivation and late math academic achievement. 3. Math academic achievement did not mediate fully the effects on early math anxiety and late math learning motivation. 4. Math anxiety did not mediate fully the effects on early math academic achievement and late math learning motivation. 5. Math academic achievement did not mediate fully the effects on early math learning motivation math and late math anxiety. 6. Math learning motivation did not mediate fully the effects on early math academic achievement and late math anxiety. In addition, the research also explored the longitudinal model trend and effect, the key period for 3rd year junior high school students, the proper models, and the mediated roles among math anxiety, learning motivation, and academic achievement.