サッカー日本代表、やりましたね！強豪国コロンビア相手に2－1での勝利、興奮しました。2018年6月7日付のFIFAランキングを見ると、コロンビアは16位、日本は61位ですから、大金星と言っても過言ではないでしょう。ところで、様々ある世界ランキングの中で、日本が長らくトップクラスに位置しているもの、それが教育です。今回は、そんな世界に冠たる日本の教育について、全国学力テストの分析結果を英語で紹介していきます。
Japan has been widely acknowledged as a successful country in terms of students’ academic achievements. In addition to its strong performance on average, from an international comparative perspective based on the OECD’s PISA, Japan has actualised a high level of equity; although socio-economically advantaged students are more likely than others to show better academic outcomes, the association between socio-economic status (SES) and performance is relatively weak. While not a few studies have revealed the factors of Japan’s success via analysing the PISA data, in-depth investigations within Japan have not been developed/disseminated adequately.
Meanwhile, as explained in the previous article, the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan, has been carrying out the National Assessment of Academic Ability (the Assessment), a nationwide student/school survey, since 2007. As part of this initiative, I quantitatively analysed the survey data whilst conducting qualitative research focused on “effective schools.” As a result, evidence suggested that 1) SES had significant impacts on students’ achievements, but 2) various teaching/learning activities contributed to mitigating such unequal situations, and 3) the association between SES, teaching/learning activities, and performance was a bit different between primary school and lower secondary school students.
With this in mind, in this paper, I will introduce some findings of the quantitative analysis. (Don’t worry, I will write about research on “effective schools” in the next article!)
- 1. Analytical Framework: Identifying the Determinants of Academic Ability
- 2. Analysis Result: Primary School Students
- 3. Analysis Result: Lower Secondary School Students
- 4. As a Foundation for Further Research…
1. Analytical Framework: Identifying the Determinants of Academic Ability
The main goal of the quantitative analysis of the Assessment data was to identify the determinants of students’ academic ability (see the previous article for more details). In this light, I conducted a multilevel regression analysis using test scores in science (dependent variables) and answers of questionnaires for students and schools (independent variables) with the following framework (Figure 1). This analysis provided evidence to examine which independent variable (potential factor) significantly affected students’ achievements in primary and lower secondary schools respectively. For details of the number of respondents and a complete list of variables including their definitions and descriptive statistics, see the full report of the analysis.
Figure 1: Analytical Framework (Multilevel Regression Analysis)
Note: There are two types of test: Type A (subject knowledge) and Type B (practical skills). As they assess different dimensions of academic ability, I analysed the impact of independent variables on Type A score and Type B score respectively.
Source: The MEXT Expert Meeting (modified by author)
2. Analysis Result: Primary School Students
The table below (Table 1) summarises the analysis result concerning primary school students. (For more detailes of a multilevel regression including coefficients and other statistical information, see the full report.) In this table, “A” and “B” in the top row signify Type A and Type B tests. “〇” indicates that an independent variable in the left column demonstrates a significant positive impact on test scores (A and B respectively) after controlling for other variables, whereas “▼” means a significant negative impact. Blank cells suggest that independent variables do not show any significant association with test scores. For example, students who frequently read books outside of schools (fourth from the top of student-level variables) are more likely to show better performance in both Type A and Type B even after accounting for the influence of other factors. Meanwhile, those who attend disadvantaged schools (i.e. the share of students who receive financial/language assistance is high) are likely to suffer from low achievements as compared to students of more advantaged schools.
Table 1: Summary of The Impacts of Student-level and School-level Variables on Test Scores (Primary School Students)
Source: The MEXT Expert Meeting (modified by author)
Overall, student-level variables such as lifestyle (regularly eat breakfast, sleep/get up early), learning hours outside of schools, interests in science and societies, the use of shadow education, and experience of collaborative learning, presentation, and experiments have significant positive effects on performance regardless of test type. Further, while socio-economic disadvantages of a school (in the form of shares of financial/language assistance recipients) show negative influences, various teaching/learning activities underpin students’ achievements. Among others, teaching advanced knowledge on science (beyond the basic knowledge specified in the Courses of Study), encouraging students to design and record experiments by themselves, and providing a sufficient amount of homework demonstrate significant positive impacts on both Type A and Type B.
Indeed, the figure below (Figure 2) illustrates mean test scores of students by the frequency of recording/considering experiment results (in addition to designing and discussing them) in four different school groups classified based on the share of students who receive financial assistance. For instance, in schools where no students need financial assistance (indicated as “0%” at the top of the figure), the mean test score of students who “very frequently” record/consider experiment results is 65.5, whereas that of those who has “never” experienced such activities is 56.0 (and this difference is statistically significant). Similarly, even in very disadvantaged schools where more than 30% of students receive financial assistance, children who are instructed to do experiments actively are more likely to demonstrate stronger performance (mean test scores of “very frequently” and “never” are 57.1 and 50.0 respectively).
Figure 2: Mean Test Scores (Type B) of Primary School Students by the Frequency of Recording/Considering Experiment Results (in four different school groups based on the share of students receiving financial assistance)
Source: The MEXT Expert Meeting (modified by author)
Sadly, as we can see from the above table/figure, the achievement of students is significantly affected by the extent to which their schools are socio-economically (dis)advantaged. In fact, for example, the mean score of those who “very frequently” record/consider experiment results in the most disadvantaged schools (i.e. more than 30% of students receive financial assistance) is lower than that of those who “seldom” do such activities in the most advantaged schools (57.1 and 61.1 respectively and this difference is statistically significant). This suggests that it is difficult to completely overcome the socio-economic disadvantage via one specific pedagogy. Nevertheless, it is also true that various teaching/learning approaches have potential to enhance students’ academic performance at least to the degree of mitigating the influence of socio-economic circumstances. That is, by incorporating multiple activities that have proved to promote students’ achievements, we can expect greater impacts that cannot be realised by merely employing each measure independently.
3. Analysis Result: Lower Secondary School Students
What about secondary school? Table 2 is a summary of the analysis result of lower secondary school students, showing a very similar pattern to primary school students with some differences. Firstly, student-level variables, including learning hours, reading books, interests in science and societies, and experience of designing/presenting experiments, have significant influences on test scores. Secondly, the share of students who receive financial assistance in a school has negative impacts, whereas the proportion of language assistance recipients no longer shows significant effects. Thirdly, a variety of school-based teaching/learning activities, such as putting importance on learning discipline as a basis for quality learning and encouraging students to actively participate in designing/recording experiments, demonstrate positive associations with academic performance.
Table 2: Summary of The Impacts of Student-level and School-level Variables on Test Scores (Lower Secondary School Students)
Source: The MEXT Expert Meeting (modified by author)
The figure below (Figure 3) illustrates mean test scores of students by the frequency of designing experiments based on their own hypotheses in different types of schools as defined by the proportion of financial assistance recipients. The more frequently students consider hypotheses and design experiments by themselves, the higher mean test scores become regardless of school categories. However, there is also a very clear trend; students in more advantaged schools are likely to show better performance regardless of the frequency of experiment designing. Consequently, even though students “very frequently” design experiments based on their hypotheses, they cannot outperform those who have “never” experienced such activities in more socio-economically advantaged schools. For instance, the mean score of “very frequently” in schools where more than 30% of students receive financial assistance (45.8) is lower than that of “never” in schools whose share of assistance recipients is between 10% and 30% (47.2).
Here one may argue that the gap between the advantaged and the disadvantaged can be alleviated by combining various interventions (teaching/learning activities) as with primary school students. Nevertheless, it is worthwhile to recognise the possibility that the impact of economic circumstances in each school is intensified in lower secondary schools as compared to primary schools. Indeed, the magnitude of the share of financial assistance recipients is larger for lower secondary students than for primary school students in multilevel regression models (see the full report for more details). With this in mind, much broader approaches focused on socio-economic circumstances, in addition to school-based interventions, might be more important at later development/school stages to overcome the socio-economically established unequal situations.
Figure 3: Mean Test Scores (Type B) of Lower Secondary School Students by the Frequency of Recording/Considering Experiment Results (in four different school groups based on the share of students receiving financial assistance)
Source: The MEXT Expert Meeting (modified by author)
4. As a Foundation for Further Research…
What is the key to high academic performance in science in Japan? As noted, there are a variety of variables that affect students’ achievements. On the one hand, socio-economic conditions (quantified by the share of students who receive financial/language assistance in a school) have significant negative influences after controlling for other factors. That is, while Japan has been renowned for its relatively equal education system, unequal structures surely exist (especially at lower secondary schools). Nevertheless, on the other hand, this does not necessarily mean that we cannot do anything to address this unequal circumstance. Indeed, various activities at both individual and school levels, such as designing/reporting experiments, securing proper learning discipline, and providing a sufficient amount of homework, demonstrate positive effects on students’ outcomes even after accounting for other variables. Although it seems difficult to completely overcome the persistent socio-economic disadvantages, it is also reasonable to expect that such unequal situations can be greatly mitigated through multiple teaching/learning activities as indicated above. To carry out such interventions effectively, we should bear in mind that a broader collaboration among not only students/families and schools but also other stakeholders including local communities and private sectors are necessitated.
In the meantime, it is also important to note that there are many other interesting findings we can derive from the results of the quantitative analysis (although I could not introduce all of them in this article due to a space constraint…). In particular, comparative perspectives focused on the difference across gender, test type, school stage, and geographical location among others should lead to insights into the determinants of academic performance. Further, while a quantitative analysis enables us to statistically evaluate whether or not and to what extent independent variables contribute to test scores, it is not enough to elucidate “how/why” each factor affects students’ achievements. It is therefore essential to conduct a qualitative analysis and to consider its results in conjunction with quantitative evidence. Thus, in the next article, I will introduce essence from “effective school” research to clarify the key to equitable academic success in Japan. Stay tuned!
荒木啓史
Araki, Satoshi
HP: Sarthak Shiksha | Quality Learning for All
FB: https://www.facebook.com/SarthakEd
---------------------------------------------------------------------------------------------
サルタックは無料メルマガを開始しました。
・本ブログ記事の更新情報
・編集後記
・日本でのイベント情報
・インターンやボランティアの募集告知
などをお送りいたします。
下記のリンクにあるようなメルマガを配信しています。
ご興味がある方は下記のリンクよりご連絡ください。
またサルタックでは常時会員と寄付を受け付けています。
サルタックと共に多くの子どもたちに有意義な教育を届けるパートナーになりませんか。
ご関心がある方は下記のリンクより会員・寄付のオプションについての詳しい説明にお進みください。
---------------------------------------------------------------------------------------------
