We used multilevel mixed modeling to test the extent to which students’ music achievement scores were related to their reading and math achievement scores. Of the four levels examined (individual students, classrooms, schools, and districts), only individuals and districts accounted for a significant portion of the total variance in achievement scores. We studied several background variables potentially affecting academic achievement among individuals: grade level, gender, educational attainment of parents/guardians, free/reduced-price lunch, ethnicity, and urbanicity. We also incorporated district-level variables of sex, ethnicity, parent/guardian education level, and free/reduced-price lunch along with principal component scores for four district-level latent variables—district achievement, district behavior, available funds, and local revenue. Fourth through eighth graders (N = 1,081) from seven midwestern school districts participated in the study. Students sat for a representative portion of the first two of the Music Achievement Tests (MAT-1 and MAT-2). We developed separate models for reading and math achievement. No significant differences were found among the districts’ intercepts or slopes. With the aforementioned variables controlled for, both MAT-1 and MAT-2 (controlling for one another as well) demonstrated a strong relationship with reading and math achievement (ps < .0001).

中文翻译：

---

音乐成绩与阅读和数学成绩关系的多层次模型

我们使用多级混合模型来测试学生的音乐成绩与他们的阅读和数学成绩的相关程度。在检查的四个级别（个别学生、教室、学校和地区）中，只有个人和地区占成就分数总差异的很大一部分。我们研究了几个可能影响个人学业成绩的背景变量：年级、性别、父母/监护人的教育程度、免费/减价午餐、种族和城市化。我们还纳入了性别、种族、家长/监护人教育水平和免费/减价午餐的地区级变量，以及四个地区级潜在变量的主成分分数——地区成就、地区行为、可用资金和地方收入. 来自中西部七个学区的四年级至八年级学生 (N = 1,081) 参与了这项研究。学生参加了前两项音乐成就测试（MAT-1 和 MAT-2）的代表性部分。我们为阅读和数学成绩开发了单独的模型。各区的截距或坡度之间没有发现显着差异。控制上述变量后，MAT-1 和 MAT-2（也相互控制）都表现出与阅读和数学成绩的密切关系（ps < .0001）。各区的截距或坡度之间没有发现显着差异。控制上述变量后，MAT-1 和 MAT-2（也相互控制）都表现出与阅读和数学成绩的密切关系（ps < .0001）。各区的截距或坡度之间没有发现显着差异。控制上述变量后，MAT-1 和 MAT-2（也相互控制）都表现出与阅读和数学成绩的密切关系（ps < .0001）。