家庭作業研究的變色龍效應：家庭作業與學習成就的關係取決於測量與分析層次的選擇

The Chameleon Effect of Homework: The Relationship Between Homework and Learning Achievement Depends on Selection of Measurement and Level of Analysis

本研究透過「階層線性模式」（hierarchical linear modeling, HLM）與「長期追蹤交叉延宕模式」（cross-lagged panel modeling, CLPM），分析不同測量與不同分析層次的家庭作業與學習成就的關係。資料來自臺灣東部21個班級557位國中學生三個時間點的追蹤問卷調查。研究結果顯示，數學家庭作業時間對數學學習成就有負向預測效果，但兩者無雙向相互預測關係。其次，數學家庭作業努力對數學學習成就有正向預測效果，且兩者有雙向正向相互預測關係。再者，數學家庭作業情緒對數學學習成就有正向預測效果，且兩者有雙向正向相互預測關係。最後，數學家庭作業頻率對數學學習成就無預測效果，但聚集加總數學家庭作業時間對數學學習成就有正向預測效果。根據研究結果，本研究驗證了家庭作業與學習成就存在著「變色龍效應」（chameleon effect），兩者的關係取決於家庭作業測量與分析層次選擇的論點，並據以提出實務面與未來研究面的建議。

This study examined the relationship between homework and learning achievement across different measurement approaches and levels of analysis using hierarchical linear modeling and cross-lagged panel modeling. Data were collected from 557 junior high school students in 21 classes in Eastern Taiwan through three rounds of survey questionnaires. The results indicated that math homework time negatively predicted math learning achievement; however, no bidirectional predictive relationship was observed between math homework time and math learning achievement. Math homework effort and math homework emotion both positively predicted math learning achievement; each also exhibited a bidirectional positive predictive relationship with math learning achievement. Finally, math homework frequency did not predict math learning achievement, but aggregated math homework time showed a positive predictive effect on math learning achievement. These findings support the existence of the chameleon effect in the relationship between homework and learning achievement—the relationship depends on how they are measured and analyzed. This study offers practical implications and recommendations for future research.