1. Background and purpose of the study
Academic performance is affected by multiple factors, including individual characteristics, family background and educational resources. The purpose of this study was to systematically analyze the effects of gender, test preparation curriculum, ethnicity, parental education level, and lunch type on students' math, reading, and writing scores, and to provide data support for educational equity policies.
2. Methodological selection and basis
1. Data distribution characteristics
Although the histograms of math, reading, and writing scores appear to be approximately normally distributed to the naked eye, the Shapiro-Wilk test shows that all the score data do not conform to the normal distribution (p < 0.05), and there are a small number of extreme values at both ends of the data, resulting in a "truncated" tail and a slight rightward bias , the two endpoints of the Q-Q diagram deviate from the red line.
Therefore, this study is not considered to be suitable for parametric test methods such as t-test or ANOVA.
2. The necessity of non-parametric testing
Based on the results, the following non-parametric methods are used for analysis:
Mann-Whitney U test: Used to compare two independent samples, this study was used to compare the effects of gender differences, test preparation on performance.
Kruskal-Wallis H test: Used for comparisons between multiple groups, this study was used to compare racial differences
Spearman rank correlation: used to analyze the relationship between ordinal and continuous variables, this study was used to compare the relationship between parental education level and achievement
3. Visualization methods
The visualization of this study was implemented by Python pandas, seaborn, scipy, matplotlib, and all graphs included statistical annotations to enhance the readability of the results.
3. Task analysis and results
Task 1: Boxplot of gender distribution
|
|
|
|
| 101768 | <0.001 |
|
| 160508 | <0.001 |
|
| 169956 | <0.001 |
|
Task 2: The impact of exam preparation courses on grades
Removing some extreme values, U=150320, p<0.001 was obtained, and the average score of students who completed the test preparation course was significantly higher than that of those who did not complete the three subjects (about 8 points higher on average ). And the dispersion is relatively less (SD=13<14.2), and the play is more stable.
Task 3: Differences in performance by race/ethnicity
|
|
| Kruskal-Wallis 检验 |
Group A | 65.5 | 14.7 | H = 9.10, p < 0.001 |
Group B | 67.1 | 13.9 | |
Group C | 63.0 | 14.4 | |
Group D | 69.2 | 13.3 | |
Group E | 72.8 | 14.6 |
|
H=9.1, Group E students had the highest average grade, Group A had the lowest and was significant.
Task 4: The relationship between parental education and grades
|
|
|
|
some high school | 65.1 | r = 0.19 | <0.001 |
high school | 63.1 | ||
some college | 68.5 | ||
associate's degree | 69.6 | ||
bachelor's degree | 71.9 | ||
master's degree | 73.6 |
Conclusion: There is a weak positive correlation between parental education level and student achievement (r = 0.19), and there is a significant difference.
Task 5: Analysis of other factors
Effect of Lunch Type on Grades:
|
|
| Mann-Whitney U 检验 |
standard | 70.8 | 13.2 | U = 152822, p < 0.001 |
free/reduced | 62.2 | 14.5 |
|
Standard lunch students scored an average of 12.3 points higher than free lunch students.
Grade Correlation Heatmap:
Conclusion: The results of the three subjects are highly correlated, indicating that the learning ability is consistent as a whole. And the correlation between writing and reading is greater than that between writing and math and reading and math.
4. Summary of key findings
|
|
|
|
| Mann-Whitney U 检验 |
|
| Mann-Whitney U 检验 |
|
| Kruskal-Wallis H 检验 |
|
|
|
|
| Mann-Whitney U 检验 |
5. Suggestions and strategies
Gender-specific interventions
Strengthen language subject tutoring (reading, writing) for male students
Designing mathematical thinking training courses for female students to close the gender gap
Promote exam preparation courses
Provide free/low-cost resources, especially for underachievers
Racial Equity Policy
Additional academic support for Group A students (e.g. cram classes, tutor programs)
Analyze Group E's successes and replicate them to other groups
Parent Educational Involvement
Develop workshops for parents to increase educational engagement (especially for parents with high school education)
Provide family education guidance manuals to promote the optimization of the home learning environment
Optimization of lunch subsidy
Evaluate the practical effect of the free lunch policy, combined with comprehensive measures such as after-school tutoring
Provide nutritional supplementation programs for students from low-income families to improve their learning status