How Does Physical Exercise Affect Academic Performance? The Mediating Role of Non-Cognitive Abilities Wenye Li, 1 Jijun Yao, 1 Shike Zhou 2 1. Nanjing Normal University, Nanjing 210023, China 2. Jiangsu Institute of Education Science, Nanjing 210013, China Abstract. Previous studies have found that good physical exercise can promote academic performance, but the underlying mechanism behind this lacks large-scale empirical data. Based on this, we used the 2020 Jiangsu Province academic quality monitoring data to construct an OLS regression. Then, the non-cognitive abilities variable was built with the NEO-FFI Theory, and the mediation effect diagram was drawn through the Amos22.0 software. The study found that: (i) Physical exercise brought academic improvement to elementary and middle school stu- dents. (ii) The influence mechanism of physical exercise was different between elementary school and middle school. Students at the elemen- tary school were directly and indirectly affected. (iii) In non-cognitive specific dimensions, openness played a significant and great mediating role. Best Evidence in Chinese Education 2021; 8(2):1103-1120. Doi: 10.15354/bece.21.or046. How to Cite: Li, H., Yao J., & Zhou, S. (2021). How does physical exercise affect academic performance? The mediating role of non-cognitive abilities. Best Evidence in Chinese Education, 8(2):1103-1120. Keywords: Physical Exercise; Non-Cognitive Abilities; Openness; Mediating Ef- fect Li, et al. How Does Physical Exercise Affect Academic Performance? Vol.8, No.2, 2021 1104 Literature Review and Questions NHANCING the physique of the national citizen has always been a livelihood issue of the country’s focus. As an essential means, physical exercise has been highly valued at different times. Since the launch of “Sunshine Sports” in 2007, the value of physical exercise has been universally recognized. However, the question parents and teachers are more concerned about is if physical exercise can pro- mote the improvement of academic performance and that is still unclear. Physical exer- cise can strengthen the body, but spend time on physical exercise instead of extracurric- ular learning, parents will inevitably worry that the physical exercise “occupies” the study time and causes children’s academic performance to decline. Especially in the critical stage of entering higher education, many schools will even suspend physical exercise courses and increase academic courses. In recent years, the “quiet large class breaks” have also become a microcosm of the embarrassing position of physical exer- cise. Physical exercise seems to be “unrelated” to improving academic performance, and there is a general situation of “emphasizing academics and neglecting sports” throughout China. In the embarrassing public opinion, what is the relationship between physical exercise and students’ academic performance? High-quality empirical research is urgently needed to find the answer to it. Especially in the context of reducing stu- dent’s burdens and increasing students’ efficiency, what physical exercise can bring to education is related to every student, parent, and educator. The discussion on the relationship between physical exercise and students’ ac- ademic performance can be traced back to the 1950s. Based on the sample data of stu- dents in third and fifth grades, Gleason et al. (1958) found a weak correlation between physical exercise and students’ academic performance. Subsequent studies have mostly adopted experimental research methods and found that physical exercise can promote academic performance to a certain extent. Such studies were often based on physiology or psychology for theoretical construction. From the perspective of Arousal Theory, it was believed that the physical exercise that students engage in could effectively im- prove their brain’s execution level and make them have better task performance. How- About the Authors: Wenye Li, Graduate Student, School of Educational Science, Nanjing Normal University, Nan- jing 210023, Jiangsu, China. E-mail: 841659607@qq.com Shike Zhou, Senior Teacher of Middle School, Jiangsu Institute of Education Science, Nanjing 210013, Jiangsu, China. E-mail: jyszsk@163.com Correspondence to: Jijun Yao, Professor, School of Educational Science, Nanjing Normal University, Nanjing 210023, Jiangsu, China. E-mail: yaojijun_njnu@163.com Funding: This research is the phased result of the “Research on the Relationship between Physical Exercise and Student Development” (No.: KYCX21_1301) of the Jiangsu Province Graduate Research and Practice Innovation Program. This study was supported by the Jiangsu Provincial Superior Discipline Construction Project (PAPD). Conflict of Interests: None. © 2021 Insights Publisher. All rights reserved. Creative Commons Non Commercial CC BY-NC: This article is distributed under the terms of the Crea- tive Commons Attribution-NonCommercial 4.0 License (http://www.creativecommons.org/licenses/by- nc/4.0/) which permits non-commercial use, reproduction and distribution of the work without further permission provided the original work is attributed by the Insights Publisher. E mailto:841659607@qq.com mailto:jyszsk@163.com mailto:yaojijun_njnu@163.com Li, et al. How Does Physical Exercise Affect Academic Performance? Vol.8, No. 2, 2021 1105 ever, the control of confounding variables in this kind of study was not comprehensive, and the attributes of a single discipline were evident. Thus, it had failed to achieve the intersection and breakthrough of theories between disciplines. With the development of measurement technology and the continuous progress of cognitive psychology and neurobiology, the studies on the relationship between physical exercise and the academic performance had become more systematic and ma- ture. On the basis of controlling some of the confounding variables, many researchers have found that students engaging in specific physical exercises can effectively improve their cognitive abilities or academic performance. The value of physical exercise in ed- ucation quality and fairness has been effectively demonstrated. For example, Fang Lim- ing, based on adolescent samples in the CEPS database, used the PSM method to con- trol the endogeneity of the samples and found that physical exercise can not only im- prove the academic performance of adolescents but also effectively improve the aca- demic performance of students from economically disadvantaged families and achieve educational equity. Such findings have also been verified in the study of Fang Chao et al. (2021). Quantile regression results found that physical exercise can reduce the dif- ference between high and low quantile students and has a fair value that cannot be ig- nored. In addition, many researchers based on the sampling of regional samples also found that the frequency or duration of physical exercises of students was highly corre- lated with academic performance. The meta-analysis results also showed that allowing students to engage in physical exercise is conducive to improved academic performance. The studies, as mentioned above, have formed two sets of mechanisms of phys- iology and psychology in terms of the influence mechanism of physical exercise. The former emphasized the “strengthening” of physical exercise and believed that students improve their health through physical exercise and thus improve their academic perfor- mance. For example, Zhang et al. (2016) found that physical exercise can improve the quality of sleep of students and improve the academic performance of high school stu- dents. The latter focuses on the “mental health” effect of physical exercise. It was be- lieved that physical exercise effectively relieves students’ inner pressure and improves self-efficacy, which in turn enhances classroom performance and promotes academic performance. For example, the study of Zhang et al. (2021) selected peer relationship and self-confidence as the mediating variables and found that physical exercise promot- ed the improvement of both and thereby improved academic performance. Zhou (2018) used self-esteem as an intermediary variable and found that it played an essential role between physical exercise and academic achievement of college students. Taken together, the existing research has provided sufficient evidence for the positive effects of physical exercise. However, there are still apparent shortcomings behind these pieces of evidence. On the one hand, the sample size was not enough, and high-quality, large-scale data analysis was still needed to verify this positive effect. On the other hand, many studies were not adequate in controlling confounding variables. They failed to control variables such as family background, which may cause bias in the regression results. In terms of influence mechanism, there may be multiple influence paths between physical exercise and academic performance. Many studies often select a Li, et al. How Does Physical Exercise Affect Academic Performance? Vol.8, No. 2, 2021 1106 single intermediary variable. Most of them were psychological or physical variables related to cognitive ability but ignored the non-cognitive abilities highly correlated with students’ academic performance. As a concept corresponding to cognitive abilities that emerged during the development of human capital theory, non-cognitive abilities were often regarded as critical abilities in the 21st century (Huang et al., 2017; Xu, 2017). In empirical research, many empirical studies had also selected non-cognitive abilities as the core explanatory variable. However, the theoretical frameworks they were based on are relatively vague. They often defined self-efficacy, peer relationships, and other ex- istences unrelated to cognition in a broad sense as non-cognitive abilities. These con- cepts did not have the attribute of “ability.” Therefore, in terms of mechanism research, the number of existing studies on the mediating variables is insufficient. The theory is not yet systematic, and the exploration of non-cognitive abilities is not comprehensive. Based on this, this study will adopt the more popular NEO-FFI Theory (Neuroticism Extraversion Openness Five Factor Inventory). We subdivided non-cognitive abilities into openness, conscientiousness, extraversion, agreeableness and neuroticism (Costa et al.,1985; Meng, 2003; Wang, 1994). The following questions are verified and discussed by constructing the non-cognitive abilities variable and assuming it as a mediating vari- able between physical exercise and academic performance. 1. Can physical exercise improve students’ academic performance? 2. Can physical exercise promote students’ academic performance through non- cognitive abilities? 3. Which non-cognitive abilities does physical exercise mainly use to promote stu- dents’ academic performance? Methodology Research Ideas This study first verified the relationship between physical exercise and academic per- formance by constructing a regression model and calculating regression coefficients. Subsequently, intermediary variables are included to explore possible influence mecha- nisms and demonstrate the vital role of non-cognitive abilities in students’ academic performance. Finally, the mediation effect is calculated, and the path coefficient dia- gram is drawn. The effect size of each path is compared to explore the role of each di- mension in the non-cognitive dimension in more detail. Variable Selection Core Explanatory Variables  Physical Exercise Students need to answer the daily physical exercise frequency in this test. They choose from five options: “No”, “Less than 1 hour”, “1-2 hours”, “2-3 hours” and “3-5 hours”. Li, et al. How Does Physical Exercise Affect Academic Performance? Vol.8, No. 2, 2021 1107 In the elementary school sample, 26.9 percent of students failed to meet the one-hour exercise standard while in the middle school sample this accounted for 41.1 percent.  Non-Cognitive Abilities This study used the NEO-FFI Theory to subdivide sub-cognitive ability into five di- mensions: openness, conscientiousness, extraversion, agreeableness, and neuroticism. Among them, neuroticism is a negative personality, which has been transformed into an upbeat personality. Openness mainly refers to students’ ability to accept new things and use methods to solve problems. Conscientiousness primarily refers to the degree of ef- fort that students have to accomplish their goals and whether they persist in the end. Extraversion refers to the enthusiasm and vitality of students, higher extraversion indi- cates that students have better peer relationships. Agreeableness refers to the ability of students to show altruistic behavior. Finally, neuroticism refers to the power of students to control emotions. Since the above five dimensions are all latent variables, it isn’t easy to measure intuitively. Therefore, the five dimensions were all set with 3-4 ques- tions for measurement, and the principal component method was used to fit the score. The elementary school and middle school samples were tested separately. The α indica- tors of the five dimensions were all in the interval of [0.6, 0.8], indicating that the data has good reliability. In terms of validity, the RMSEA indicators of the five dimensions were all less than 0.07, meaning a good fit, and the CFI and TLI are both greater than 0.9, and the validity is good. More statistics can be found in Table 1.  Explained Variable The academic performance middle school students participating in were divided into Chinese, mathematics, English, biology, geography, and physics; and the elementary school was divided into Chinese, mathematics and English. Considering the academic burden and other reasons, each student in the middle school only tested two subjects and the elementary school only tried one subject. Considering that there are apparent differences between subjects, it is impossible to simply take the subject scores of the students as the total score. Moreover, regression by the subject may cause the lack of a large area of samples. Therefore, after collecting the data, we decided to fit a regression model based on other items in the questionnaire that may be related to the results; and then performed multiple imputations on the results of the test subjects that the sample did not take to generate the general scores. Such methods are often used to deal with large areas of missing values in the sample. After getting the results of each subject, considering that the policies for the high school entrance examination in various regions of Jiangsu Province are different, and the examination methods of biology and geogra- phy are also different, the middle school’s general scores only used the total scores of Chinese, mathematics, English and physics. In contrast, the elementary school used the total scores of Chinese, mathematics and English. As a result, the average score of all subjects in middle school was 555.22, and the average score of all topics in elementary school was 515.19. Li, et al. How Does Physical Exercise Affect Academic Performance? Vol.8, No. 2, 2021 1108  Control Variable To avoid missing variables cause biased estimation of regression coefficients, this study incorporated students’ family background variables, demographic variables, and school background variables into the regression model to control bias. In sociological research, students’ stock of family capital can influence students’ academic performance through intergenerational inheritance. These capitals cover three forms: economy, culture, and society. In this study, we averaged the number of students' reports about the items owned by student families and used it as proxy variables of economic capital. Students needed to answer the number of particular items (such as televisions, computers, cars, etc.) owned by the family. A total of nine questions were set. In terms of society capital, using the “Chinese Occupation Classification Dictionary (2015)” as a reference, the parents’ occupations were assigned, sorted (scores of 1-8) and standardized, and the highest score of the parents was taken as the social capital in the family background. Finally, the parents’ academic qualifications represented the cultural capital. First, the parents’ academic qualifications were converted into years of education and assigned values of 9, 12, 16, and 19. After standardization, the highest scores of both parents were taken as the cultural capital of the student’s family. In addition, the gender of stu- dents was 1 for girls and 0 for boys; the only child status was 0 for only children and 1 for non-only children. The school type was 1 for public and 0 for private. The school area was 1 for urban and 0 for rural areas. The specific sample descriptive statistics were shown in Table 1.  Model Setting This research will first conduct descriptive statistics on samples of the fifth and eighth grades in Jiangsu Province, locate the physical exercise situation, and construct an OLS regression model, Including core explanatory variables, control variables, and media- tion variables to analyze the positive effects of physical exercise and the role of non- cognitive abilities in the impact mechanism. The specific model expression is as follows: Among them, Scorei is the average grade of the subject of the i-th student. β1 is the regression coefficient of the student’s physical exercise situation. β2 is the regres- sion coefficient of students’ non-cognitive abilities. β3 is the regression coefficient of the control variable. Finally, εi is the residual of the regression equation.  Data Source and Processing This study used the 2020 monitoring data of students’ academic quality in compulsory education in Jiangsu Province. The project is carried out once every two years. Two- Li, et al. How Does Physical Exercise Affect Academic Performance? Vol.8, No. 2, 2021 1109 Table 1. Descriptive Statistics. Elementary School Mean Standard Deviation Minimum Maximum Only Child 0.63 0.48 1.00 2.00 PE Time 2.88 0.80 1.00 5.00 Public School 0.64 0.48 0.00 1.00 Urban 0.87 0.34 0.00 1.00 Female 0.48 0.50 0.00 1.00 Economic Capital 0.03 0.99 -3.04 4.21 Social Capital 0.69 0.29 0.00 1.00 Cultural Capital 0.49 0.33 0.00 1.00 Openness 0.03 1.00 -4.19 0.80 Responsibility 0.04 0.98 -3.88 0.89 Extraversion 0.01 1.00 -4.26 0.79 Pleasant 0.05 0.97 -4.55 0.71 Neuroticism 0.03 0.98 -3.41 0.84 Middle School Mean Standard Deviation Minimum Maximum Only Child 0.55 0.50 1.00 2.00 PE Time 2.62 0.66 1.00 5.00 Public School 0.86 0.35 0.00 1.00 Urban 0.66 0.47 0.00 1.00 Female 0.47 0.50 0.00 1.00 Economic Capital 0.57 0.10 0.25 1.00 Social Capital 0.60 0.29 0.00 1.00 Cultural Capital 0.31 0.30 0.00 1.00 Openness 0.00 1.00 -3.39 1.16 Responsibility 0.01 1.00 -2.87 1.49 Extraversion -0.01 1.00 -3.33 1.38 Pleasant 0.04 0.94 -4.00 0.98 Neuroticism -0.01 0.99 -2.41 1.38 stage stratified sampling was used to conduct a sample survey of students, teachers, and school leaders in 13 cities in Jiangsu Province. Among them, a sample of students was randomly selected from the sampled schools, and a total of 213,334 questionnaires for fifth-grade students and 147,432 questionnaires for ninth-grade students were collected. After deleting the samples that failed the polygraph test and missing important infor- mation, a total of 156,657 valid elementary school questionnaires and 127,738 useful student questionnaires from middle school were obtained. Results Regression Analysis Result of Physical Exercise In this study, students’ average scores were selected as dependent variables, and model 1, model 2, model 3, and model 4 were constructed respectively. Model 1 only incorpo- Li, et al. How Does Physical Exercise Affect Academic Performance? Vol.8, No. 2, 2021 1110 Table 2. OLS Regression Results. Elementary School Model 1 Model 2 Model 3 Model 4 (Constant) 484.50*** 509.99*** 501.13*** 520.70*** (0.548) (0.786) (0.867) (0.893) Physical Exercise 10.78*** 10.14*** 8.69*** 3.47*** (0.184) (0.176) (0.176) (0.185) Non-Only Child -23.48*** -20.74*** -19.81*** (0.302) (0.303) (0.305) Public School -2.07*** 1.29*** 0.75 (0.307) (0.314) (0.318) Urban 19.32*** 13.87*** 12.33*** (0.534) (0.529) (0.535) Female 8.24*** 6.71*** 4.19*** (0.287) (0.285) (0.289) Economic Capital 11.49*** 10.29*** (0.154) (0.156) Social Capital 8.06*** 6.79*** (0.545) (0.553) Cultural Capital 11.06*** 9.51*** (0.498) (0.505) Openness 10.34*** (0.242) Responsibility 4.17*** (0.229) Extraversion -0.21 (0.245) Pleasant 5.76*** (0.213) Neuroticism 0.22 (0.188) R² 0.03 0.11 0.17 0.27 F 3431.12 3236.64 3149.63 2832.05 Middle School Model 1 Model 2 Model 3 Model 4 (Constant) 503.42*** 545.11*** 466.53*** 497.41*** (0.833) (1.032) (1.534) (1.533) Physical Exercise 4.51*** 3.70*** 3.50*** -3.58*** (0.309) (0.296) (0.292) (0.296) Non-Only Child -20.90*** -13.41*** -12.42*** (0.396) (0.398) (0.39) Public School -33.70*** -24.22*** -19.99*** (0.56) (0.556) (0.544) City 20.07*** 10.62*** 10.14*** (0.418) (0.423) (0.416) Female 16.87*** 14.81*** 10.97*** (0.391) (0.384) (0.385) Economic Capital 81.31*** 59.93*** (2.168) (2.129) Social Capital 12.26*** 11.28*** Li, et al. How Does Physical Exercise Affect Academic Performance? Vol.8, No. 2, 2021 1111 (0.816) (0.801) Cultural Capital 44.22*** 36.71*** (0.807) (0.793) Openness 27.95*** (0.303) Responsibility 9.27*** (0.284) Extraversion -10.49*** (0.325) Pleasant 4.44*** (0.319) Neuroticism -7.23*** (0.23) R² 0.002 0.09 0.16 0.29 F 213.088 2288.96 2653.349 3159.02 Note: () is the standard error of the coefficient; * represents the significance level, where * is P < 0.1, ** is P < 0.05, *** is P < 0.01. rated physical exercise variables; models 2 and 3 incorporated background control vari- ables to estimate the positive effects of physical exercise; model 4 incorporated non- cognitive abilities variables to build a full model, analyzed the positive impact, and ini- tially discuss the impact mechanism of physical exercise. Before regression, the white collinearity test was performed on the sample data. There was no collinearity problem between the variables, and there was no obvious heteroscedasticity. The regression co- efficients and standard errors of the samples are shown in Table 2. From the results in Table 2, physical exercise had different effects in elemen- tary schools and middle schools. In models 1-3, physical exercises all positively pre- dicted the academic performance of elementary and middle school students. As the model variables increased, the regression coefficients decreased. The regression coeffi- cient in the elementary school sample was always positive; in the middle school sample, after adding non-cognitive variables, the student’s academic performance was signifi- cantly negatively predicted by the physical exercise, the regression coefficient was - 3.60, and the significance level reached 0.01. On the whole, regardless of whether there was a mediating effect, the positive impact of physical exercise existed in the elemen- tary school stage. However, after controlling the influence of background variables and non-cognitive abilities in the middle school stage, the positive effect of physical exer- cise turned into an “inhibition” effect. Each time a student’s physical exercise time was increased by one unit, their academic performance dropped by 3.60 points instead. Regression Analysis Results of Non-Cognitive Abilities In addition to the high correlation between physical exercise and students’ academic performance, this research also demonstrated the high correlation between non- cognitive abilities and academic performance. First of all, from model 4 of the elemen- tary school and middle school samples, we found that the three dimensions of openness, Li, et al. How Does Physical Exercise Affect Academic Performance? Vol.8, No. 2, 2021 1112 conscientiousness and agreeableness all positively and significantly predicted students’ academic performance, which was consistent with the results of foreign empirical re- search. The effect sizes of the three dimensions were further calculated. In the elemen- tary school sample, the effect sizes of the three dimensions were 0.195, 0.076 and 0.104, respectively, while the middle school sample were 0.379, 0.129 and 0.056, respectively. The above results showed that non-cognitive abilities could help students improve their academic performance. In its interior, the effects of each personality trait were different in different stages, but the openness was always in a higher position. That means stu- dents who is enthusiastic about new things and willing to challenge often promotes aca- demic performance the most. Regression Analysis Results of Other Variables In terms of background variables, while controlling for the influence of other variables, the results of all models showed that students with non-only-child status had lower aca- demic performance than students with only-child status. This may be because only- child families often have more investment space for students’ academic or non- academic tutoring. In addition, the academic performance of urban students was higher than that of rural students, and the regression coefficient interval was located in [9.53, 19.99], indicating that the issue of regional education equity needs to be improved. Pre- vious studies had agreed that a good family background would improve students’ aca- demic performance. In this study, the higher the stock of economic, social, and cultural capital of a student’s family, the higher their academic performance would be, con- sistent with previous research results. Analysis of the Impact Mechanism of Physical Exercise From the regression coefficients of model 1 to model 4, the coefficient of physical exer- cise gradually decreased, especially after adding the five dimensions of non-cognitive abilities; the coefficients of elementary school and middle school decreased significant- ly, which preliminarily explained mediating effect statistically. In order to visually pre- sent the role of each non-cognitive dimension in the influence mechanism, amos22.0 software was used to construct an intermediary model to draw a path diagram. The re- gression path diagram is shown in Figure 1. From the path diagram of the elementary school sample, the five non-cognitive dimensions would be significantly positively predicted by the physical exercise, and the regression coefficients were similar. All were significant at the level of 0.01, indicating that physical exercise can effectively shape all aspects of personality traits. The second half of the path was significant at the 0.01 level. The mediation effect calculation showed that the mediation effect of openness, conscientiousness, and agreeableness was more prominent, which were 28.43%, 10.95%, and 14.00%, respectively. But unlike elementary schools, there were certain differences in the effects of physical exercise on the five non-cognitive dimensions in the middle school samples. Figure 2 showed that in terms of regression coefficients, the entire five dimensions Li, et al. How Does Physical Exercise Affect Academic Performance? Vol.8, No. 2, 2021 1113 Figure 1. Elementary School Stage Road Map. Figure 2. Middle School Stage Road Map. Li, et al. How Does Physical Exercise Affect Academic Performance? Vol.8, No. 2, 2021 1114 could be affected by physical exercises, which were all significant at the level of 0.01. In the second half of the path, the entire dimension could predict academic performance while the extraversion and the neuroticism predict negatively. Among them the open- ness could predict academic performance much more strongly than other dimensions. After calculating the mediation effect, the mediation effect of the openness was the largest, reaching 41.17%, second of which is extroversion, only reach 13.52%. Consci- entiousness, agreeableness and neuroticism were 12.45, 6.44% and 8.61%, respectively. Moreover, the direct effect is negative. The above results showed that students at the elementary school could directly improve academic performance through physical exer- cise and could also indirectly improve academic performance to a greater extent by shaping the personality traits of openness, conscientiousness, and agreeableness. Through shaping extraversion and neuroticism personality, physical exercise could indi- rectly promote academic performance to a lesser degree. However, middle school stu- dents are under the burden of academic burden, the relationship between academic per- formance and physical exercise was very delicate. On the one hand, physical exercise did not directly promote the improvement of academic performance. On the other hand, physical exercise could greatly indirectly promote academic performance by shaping openness personality whose mediating effect was more significant than the direct effect of physical exercise. Discussion Physical Exercise Can Directly or Indirectly Promote the Improvement of Academic Performance “Sports have the power to change the world.” However, physical fitness is just the tip of the iceberg of the value of physical exercise. In the regression and mediating effect analysis of this study, physical exercise, whether in the elementary school sample or the middle school sample, was found to directly or indirectly improve academic perfor- mance. In the elementary school sample, physical exercise could reduce the burden and increase adolescents’ academic work efficiency and alleviate the educational equity brought about by the differences in students’ family backgrounds. This was consistent with the early research results of Yin et al. (2014), focusing on the positive effects of long-term physical exercise. However, we have again demonstrated this view under the influence of other variables under better control. In addition, this study also found that the direct effect of physical exercise in middle school samples is negative, which is con- trary to the conclusions of many studies (Fang, 2020; Fang et al., 2021; Lu et al., 2014). The reason is that, on the one hand, we believe that this research has selected more in- termediate variables and brought the variable of non-cognitive abilities into the regres- sion model. This effectively stripped off the effect of physical exercise statistically and obtained more accurate coefficients (Wen et al., 2004; Wen et al., 2014). On the other hand, in theory, middle school students and elementary school students have differences in psychology, physiology, and cognition, or there may be Li, et al. How Does Physical Exercise Affect Academic Performance? Vol.8, No. 2, 2021 1115 two different influence mechanisms. This ultimately leaded to the fact that physical ex- ercise cannot directly promote students’ academic performance among middle school students. However, the mediating effect of non-cognitive abilities allows physical exer- cise to improve students’ academic performance indirectly. Therefore, physical exercise seems to “squeeze” students’ academic time but the relationship between academic time and performance is not linear (Shen, 2014; Zou & Tan, 2013). Therefore, finding the best advantages of academic time and proper planning of time investment can maximize the marginal effect of chemical industry time and effectively exert the positive impact of physical exercise. Therefore, the concept of family education should shift from the original “focusing on academics and neglecting sports” to “grasp both academics and sports”; to achieve mutual complement, reduce burdens and increase efficiency. Differences Exist in the Impact Mechanism of Physical Exercise between Elementary School and Middle School. As mentioned above, middle school and elementary school may have two different in- fluence mechanisms. The academic performance of elementary school students is di- rectly and positively promoted by physical exercise and can be positively promoted by physical exercise through the five non-cognitive dimensions. On the other hand, the improvement of the academic performance of middle school students depends on physi- cal exercise through the indirect effects of openness and agreeableness. Research on the role of non-cognitive abilities in the influence mechanism of physical exercise is scarce in China. The only study that failed to consider the multidimensional and complex con- cept of non-cognitive abilities, instead of using a single variable, the results obtained are not robust (Dong et al., 2020). This study built the dimension of non-cognitive abilities based on the NEO-FFT Theory and deeply analyzed the role of each sub-dimension in the influence mechanism. Middle school students rely on indirect effects, and we be- lieved it might be that middle school students have higher non-cognitive abilities than elementary school students. Previous studies found that students of different ages have significant differences in the scores of the five dimensions (Nie et al., 2011), which made the marginal effect of physical exercise on non-cognitive abilities decline. There- fore, the promotion of academic performance will not be as apparent as the elementary school stage. But on the whole, there is a very high correlation between physical exer- cise and non-cognitive abilities. Physical exercise effectively shapes students’ positive personality, helps students accept new things and better restraint ability, and finally, through these abilities, effectively promotes academic performance. Open Personality Traits Play a Key Role The immense theoretical contribution of this research is to compare the effects of the five dimensions of non-cognitive abilities rather than simply discussing whether non- cognitive skills play a role. At present, the academic circle is still inconclusive as to Li, et al. How Does Physical Exercise Affect Academic Performance? Vol.8, No. 2, 2021 1116 which dimension of the five personality can effectively promote the improvement of academic performance. Under the premise of the independent variable of physical exer- cise, this study found that the mediating effect of openness personality is the greatest. This can show that physical exercise can help shape students’ openness personalities and it also shows that students with more assertive openness personalities are more like- ly to exert the positive effects of physical exercise. Students with strong openness per- sonalities can often use “wisdom” to summarize their behavior (Liu, 2001), and they are willing to actively seek solutions to problems and win external support (An, 2017). Specifically, in this study, students with open personality traits are less afraid of diffi- culties, have higher expectations of themselves, and can effectively use the learning methods learned in the classroom. These positive behaviors can effectively promote the improvement of academic performance. In physical exercises, especially in competitive sports activities, students have ample opportunities and possibilities to try new methods while striving to achieve competitive goals and develop a behavioral pattern that uses new methods rationally and is not afraid of difficulties. This experience will also be transferred to classroom learning to help them no longer be fearful of the challenges in the classroom learning so that they can communicate with teachers and parents, seek methodological support, and ultimately achieve the improvement of academic perfor- mance. In other words, physical exercise can improve academic performance because it has the effect of enhancing students’ classroom performance, such as enhancing atten- tion, which is consistent with the conclusions reached by some foreign studies (Grieco, 2009; Mahar et al., 2006). Study Limitation and Perspective This study demonstrated the positive effects of physical exercise and analyzed the me- diating impact of non-cognitive abilities with the help of the NEO-FFI Theory. In re- sponse to the three problems established in the study, the following conclusions are drawn: physical exercise positively can improve students’ academic performance. Fur- thermore, this positive effect is achieved by physical exercise through non-cognitive abilities. Among them, the openness personalities shaped by physical exercise are the most important. But this study still has limitations. Firstly, the measurement of NEO-FFI Theory is incomplete compared with the measurement in psychological research and includes fewer questions. Secondly, the physical exercise variable is set as a level variable, not a continuous variable that can characterize the duration, which may make the research results miss some valuable findings, such as the optimal interval and nonlinear relationship of physical exercise. Thirdly, the direct effect in the midlle school sample was negative, which is somewhat contradictory to previous studies. In addition to the mediating effect of non-cognition, we believe endogeneity is what makes this happen and some matching methods should be used in later studies. Conclusion and Suggestion Li, et al. How Does Physical Exercise Affect Academic Performance? Vol.8, No. 2, 2021 1117 Change Ideas, and Promote the Development of both Academics and Sports Health and education are both essential components of human capital, and the two should be equally important. This study verified the positive effects of physical exercise. Educators need to realize that physical exercise has a role that other academic invest- ments cannot replace as a health investment. Therefore, we should abandon the educa- tional concept of “emphasizing academics over sports,” leaving enough time for stu- dents’ physical exercises and providing necessary support. Parents can regularly partic- ipate in physical conditioning with their children, enhancing the parent-child relation- ship and improving their academic performance. Teachers can use sports activities or competitions to cultivate children’s competitive spirit to develop positive behavior pat- terns and personality qualities and thus have better classroom performance. Pay Attention to Non-Cognitive Abilities and Open the “Black Box” of Physical Exercise This research has discovered the mediating role of non-cognitive abilities in the influ- ence of physical exercise, which shows that in the black box of physical exercise and academic performance, previous studies have indeed overlooked some key variables. As a critical skill, non-cognitive abilities will be one of the core qualities of future stu- dents. Therefore, all educators and researchers should integrate students’ non-cognitive abilities into scientific research and classroom teaching and strive to find efficient and quick ways to improve non-cognitive abilities. Pay Attention to the Cultivation of Openness Personali- ty and Realize Efficient Classroom Learning Open personality was found to play an essential role in the influence mechanism of physical exercise in this study, and it has a considerable effect on academic perfor- mance. Therefore, in daily family education, parents should attach importance to com- munication with their children and allow children to be open to competition and diffi- culties. Teachers should “teach students how to fish, not just give fish.” It is necessary to emphasize the importance of methods and let students develop the habit of using strategies to learn knowledge, which improves classroom efficiency and benefits stu- dents’ long-term development. Li, et al. How Does Physical Exercise Affect Academic Performance? Vol.8, No. 2, 2021 1118 References Álvarez-Bueno, C., Pesce, C., Cavero- Redondo, I., Sánchez-López, M., Garrido- Miguel, M., & Martínez-Vizcaíno, V. (2017) Academic achievement and physi- cal activity: A meta-analysis. Pediatrics, 140(6):e20171498. DOI: https://doi.org/10.1542/peds.2017-1498 An, Y., Huang, Q., Gu, W., & Fu, H. (2017). The impact of open personality on post- traumatic growth of college students: a moderated mediation effect. Studies of Psychology and Behavior, 15(6):815-823. [Chinese] http://www.cnki.com.cn/article/cjfdtotal- clxw201706014.htm Bailey, R., Armour, K., Kirk, D., Jess, M., Pickup, I., Sandford, R., & BERA Physi- cal Education and Sport Pedagogy Special Interest Group. (2009). The educational benefits claimed for physical education and school sport: An academic review. Research Papers in Education, 24(1):1-27. DOI: https://doi.org/10.1080/026715207018098 17 Costa, P.T., & McCrae, R.R. (1988). Person- ality in adulthood: A six-year longitudinal study of self-reports and spouse ratings on the NEO Personality Inventory. Journal of Personality and Social Psychology, 54(5):853-863. DOI: https://doi.org/10.1037/0022- 3514.54.5.853 Dong, Y., & Zhu, C. (2020). A study on ado- lescents’ extracurricular sports on aca- demic performance: Also on the mediating effect of non-cognitive abilities. Journal of Sports Research, 34(6):52-62. [Chinese] DOI: https://doi.org/10.15877/j.cnki.nsic.20210 108.005 Fang, C., & Huang, B. (2021). Can physical exercise promote the cognitive develop- ment of adolescents? An Empirical Study Based on the Data of China Education Tracking Survey. Journal of East China Normal University (Educational Sciences), 39(3):84-98. [Chinese] DOI: https://doi.org/10.16382/j.cnki.1000- 5560.2021.03.008 Fang, L. (2020). The effect of physical exer- cise on adolescents’ cognitive ability and academic performance. China Sport Sci- ence, 40(4):35-41. [Chinese] DOI: https://doi.org/10.16469/j.css.202004004 Gabbard, C., & Barton, J. (1979) Effects of physical activity on mathematical compu- tation among young children. Journal of Psychology, 103(3):287-288. https://www.researchgate.net/publication/ 234574145_Effects_of_Physical_Activity _on_Mathematical_Computation_among_ Young_Children Gleason, G.T., & Gleason, G.T. (1958). The relationship between variability in physi- cal growth and academic achievement among third-and fifth-grade children. The Journal of Educational Research, 51(7):521-527. DOI: https://doi.org/10.1080/00220671.1958.10 882497 Grieco, L.A., Jowers, E.M., & Bartholomew, J.B. (2009). Physically active academic lessons and time on task: The moderating effect of body mass index. Medical Sci- ence in Sports and Exercise, 41(10):1921- 1926. DOI: https://doi.org/10.1249/mss.0b013e3181a 61495 Hinkle, J.S., Tuckman, B.W., & Sampson, J.P. (1993). The psychology physiology and the creativity of middle school aerobic ex- ercises. Elementary School Guidance Counseling, 28(2):133-145. https://www.jstor.org/stable/42869141 Huang, G., & Xie, Y. (2017). The impact of cognitive abilities and non-cognitive abili- ties on youth labor income returns. Chi- nese Youth Study, 29(2):56-64+97. [Chi- nese] DOI: https://doi.org/10.19633/j.cnki.11- 2579/d.2017.02.009 https://doi.org/10.1542/peds.2017-1498 http://www.cnki.com.cn/article/cjfdtotal-clxw201706014.htm http://www.cnki.com.cn/article/cjfdtotal-clxw201706014.htm https://doi.org/10.1080/02671520701809817 https://doi.org/10.1080/02671520701809817 https://doi.org/10.1037/0022-3514.54.5.853 https://doi.org/10.1037/0022-3514.54.5.853 https://doi.org/10.15877/j.cnki.nsic.20210108.005 https://doi.org/10.15877/j.cnki.nsic.20210108.005 https://doi.org/10.16382/j.cnki.1000-5560.2021.03.008 https://doi.org/10.16382/j.cnki.1000-5560.2021.03.008 https://doi.org/10.16469/j.css.202004004 https://www.researchgate.net/publication/234574145_Effects_of_Physical_Activity_on_Mathematical_Computation_among_Young_Children https://www.researchgate.net/publication/234574145_Effects_of_Physical_Activity_on_Mathematical_Computation_among_Young_Children https://www.researchgate.net/publication/234574145_Effects_of_Physical_Activity_on_Mathematical_Computation_among_Young_Children https://www.researchgate.net/publication/234574145_Effects_of_Physical_Activity_on_Mathematical_Computation_among_Young_Children https://doi.org/10.1080/00220671.1958.10882497 https://doi.org/10.1080/00220671.1958.10882497 https://doi.org/10.1249/mss.0b013e3181a61495 https://doi.org/10.1249/mss.0b013e3181a61495 https://www.jstor.org/stable/42869141 https://doi.org/10.19633/j.cnki.11-2579/d.2017.02.009 https://doi.org/10.19633/j.cnki.11-2579/d.2017.02.009 Li, et al. How Does Physical Exercise Affect Academic Performance? Vol.8, No. 2, 2021 1119 Jiang, Y., Yang, Y., & Xing, S. (2016). A review of research progress on sports promoting children’s academic achieve- ment and its mechanism. Journal of Phys- ical Education, 23(5):86-92. [Chinese] DOI: https://doi.org/10.16237/j.cnki.cn44- 1404/g8.20160721.002 Jin, Y. (2001). Imputation adjustment of miss- ing data. Journal of Applied Statistics and Management, 20(6):47-53. [Chinese] DOI: https://doi.org/10.13860/j.cnki.sltj.2001.0 6.012 Li, L, & Ji, L. (2016). Research progress on the impact of physical exercise on stu- dents’ academic performance. Journal of Beijing Sport University, 39(9):82-90. [Chinese] DOI: https://doi.org/10.19582/j.cnki.11- 3785/g8.2016.09.013 Li, L. (2018). Research on the influence of physical activity on the academic perfor- mance of elementary and middle school students in my country and its intermedi- ate variables. Dissertation; Shanghai: East China Normal University. [Chinese] http://cdmd.cnki.com.cn/article/cdmd- 10269-1018710743.htm Mahar, M.T., Murphy, S.K., Rowe, D.A., Golden, J., Shields, A.T., & Raedeke, T.D. (2006). Effects of a classroom-based pro- gram on physical activity and on-task be- havior. Medical Science in Sports and Ex- ercise, 38(12):2086-2094. DOI: https://doi.org/10.1249/01.mss.000023535 9.16685.a3 Nie, Y., Yang, A., & Zeng, M. (2011). The relationship between adolescents’ meta- cognition, Big Five personality and learn- ing adaptation behaviors. Psychological Development and Education, 27(2):151- 157. [Chinese] DOI: https://doi.org/10.16187/j.cnki.issn1001- 4918.2011.02.011 Pang, X. (2004). Discussion on related issues in the processing of missing data. Statis- tics and Information Forum, 11(5):29-32. [Chinese] DOI: https://doi.org/10.3969/j.issn.1007- 3116.2004.05.007 Peña-López, I. (2016) PISA 2015 Results (Volume II): Policies and Practices for Successful Schools, [M]. Paris: OECD Publishing:126-132. http://dx.doi.org/10.1787/9789264267510 -en Shen, X. (2014). Exploring the rationality of the study time of Shanghai middle school students based on PISA data. Research in Educational Development, 33(4):9-14. [Chinese] http://www.cnki.com.cn/Article/CJFDTot al-SHGJ201404005.htm Tomporowski, P.D., Lambourne, K., & Oku- mura, M.S. (2011).Physical activity inter- ventions and children’s mental function: An introduction and overview. Preventive Medicine, 52(S1):3-9. DOI: https://doi.org/10.1016/j.ypmed.2011.01.0 28 Wang, D. (1994). Classification of Big Five Factors in the Research of Personality Traits. Advances in Psychological Science, 12(1):34-41. [Chinese] http://www.cnki.com.cn/Article/CJFDTot al-XLXD401.006.htm Wang, J., Huang, Z., & Li, H. (2016). A meta- analysis of the influence of physical activ- ity on the learning performance of chil- dren and adolescents. Journal of Capital University of Physical Education and Sports, 28(6):560-565+571. [Chinese] DOI: https://doi.org/10.14036/j.cnki.cn11- 4513.2016.06.016 Wen, X. (2015). The effect of physical exer- cise on adolescents’ cognitive ability and academic performance: the history, cur- rent situation, and future of research. Chi- na Sport Science, 35(3):73-82. [Chinese] DOI: https://doi.org/10.16469/j.css.2015.03.010 Wen, Z., & Ye, B. (2014). Mediating effect analysis: method and model development. Advances in Psychological Science, 22(5):731-745. [Chinese] https://doi.org/10.16237/j.cnki.cn44-1404/g8.20160721.002 https://doi.org/10.16237/j.cnki.cn44-1404/g8.20160721.002 https://doi.org/10.13860/j.cnki.sltj.2001.06.012 https://doi.org/10.13860/j.cnki.sltj.2001.06.012 https://doi.org/10.19582/j.cnki.11-3785/g8.2016.09.013 https://doi.org/10.19582/j.cnki.11-3785/g8.2016.09.013 http://cdmd.cnki.com.cn/article/cdmd-10269-1018710743.htm http://cdmd.cnki.com.cn/article/cdmd-10269-1018710743.htm https://doi.org/10.1249/01.mss.0000235359.16685.a3 https://doi.org/10.1249/01.mss.0000235359.16685.a3 https://doi.org/10.16187/j.cnki.issn1001-4918.2011.02.011 https://doi.org/10.16187/j.cnki.issn1001-4918.2011.02.011 https://doi.org/10.3969/j.issn.1007-3116.2004.05.007 https://doi.org/10.3969/j.issn.1007-3116.2004.05.007 http://dx.doi.org/10.1787/9789264267510-en http://dx.doi.org/10.1787/9789264267510-en http://www.cnki.com.cn/Article/CJFDTotal-SHGJ201404005.htm http://www.cnki.com.cn/Article/CJFDTotal-SHGJ201404005.htm https://doi.org/10.1016/j.ypmed.2011.01.028 https://doi.org/10.1016/j.ypmed.2011.01.028 http://www.cnki.com.cn/Article/CJFDTotal-XLXD401.006.htm http://www.cnki.com.cn/Article/CJFDTotal-XLXD401.006.htm https://doi.org/10.14036/j.cnki.cn11-4513.2016.06.016 https://doi.org/10.14036/j.cnki.cn11-4513.2016.06.016 https://doi.org/10.16469/j.css.2015.03.010 Li, et al. How Does Physical Exercise Affect Academic Performance? Vol.8, No. 2, 2021 1120 https://d.wanfangdata.com.cn/periodical/xl xdt201405001 Wen, Z., Zhang, L., Hou, J., & Liu, H. (2004). Intermediary effect test procedure and its application. Acta Psychologica Sinica, 49(5):614-620. [Chinese] http://www.cnki.com.cn/Article/CJFDTot al-XLXB200405016.htm Xu, D. (2017). How colleges can change the fate of underachievers: family poverty, non-cognitive abilities, and initial income. Chinese Journal of Sociology, 37(4):90- 118. [Chinese] DOI: https://doi.org/10.15992/j.cnki.31- 1123/c.2017.04.004 Yao, C. (2018). The relationship between physical exercise behavior and academic performance of high school students in Nanjing. Chinese Journal of School Health, 39(6):851-853. [Chinese] DOI: https://doi.org/10.16835/j.cnki.1000- 9817.2018.06.014 Yao, H., & Ye, Z. (2018). Family background, education quality and the formation of students’ abilities: an analysis of a multi- layer linear model based on CEPS. Con- temporary Education and Culture, 10(4):70-79. [Chinese] DOI: https://doi.org/10.13749/j.cnki.cn62- 1202/g4.2018.04.012 Yin, H., Chen, A., Ma, Z., Li, X., & Liu, M. (2014). A follow-up study on the impact of two exercise intervention programs on the executive function of elementary school students. China Sport Science, 34(3):24-28+75. [Chinese] DOI: https://doi.org/10.16469/j.css.2014.03.001 Zhang, L., Li, H., & Wang, Z. (2018). A sys- tematic review of the impact of physical activity on the academic ability of chil- dren and adolescents. Chinese Journal of School Health, 39(12):1808-1813. [Chi- nese] DOI: https://doi.org/10.16835/j.cnki.1000- 9817.2018.12.013 Zhang, W., Jiao, Y., Xu, X., & Yan, J. (2016). The effect of physical exercise on aca- demic performance of high school stu- dents-the mediating effect of sleep quality. Journal of Physical Education, 23(6):135- 140. [Chinese] DOI: https://doi.org/10.16237/j.cnki.cn44- 1404/g8.20161010.005 Zhang, Y., Liu, J., & He, X. (2021). The ef- fect of physical exercise on adolescents’ academic performance and its mediation mechanism: An empirical analysis based on the follow-up survey of Chinese educa- tion. Journal of Shanghai University of Sport, 45(1):29-39. [Chinese] DOI: https://doi.org/10.16099/j.sus.2021.01.004 Zhou, Z. (2018). The relationship between physical exercise and college students’ ac- ademic achievement: the mediating effect of self-esteem. Zhejiang Sports Science, 40(3):84-89+101. [Chinese] http://www.cnki.com.cn/Article/CJFDTot al-ZJTK201803019.htm Zou, M., & Tan, D. (2013). A research review of student learning input and output eval- uation. Educational Measurement and Evaluation, 6(3):10-14. [Chinese] DOI: https://doi.org/10.16518/j.cnki.emae.2013. 03.006 Received: 30 June 2021 Revised: 22 July 2021 Accepted: 22 July 2021 https://d.wanfangdata.com.cn/periodical/xlxdt201405001 https://d.wanfangdata.com.cn/periodical/xlxdt201405001 http://www.cnki.com.cn/Article/CJFDTotal-XLXB200405016.htm http://www.cnki.com.cn/Article/CJFDTotal-XLXB200405016.htm https://doi.org/10.15992/j.cnki.31-1123/c.2017.04.004 https://doi.org/10.15992/j.cnki.31-1123/c.2017.04.004 https://doi.org/10.16835/j.cnki.1000-9817.2018.06.014 https://doi.org/10.16835/j.cnki.1000-9817.2018.06.014 https://doi.org/10.13749/j.cnki.cn62-1202/g4.2018.04.012 https://doi.org/10.13749/j.cnki.cn62-1202/g4.2018.04.012 https://doi.org/10.16469/j.css.2014.03.001 https://doi.org/10.16835/j.cnki.1000-9817.2018.12.013 https://doi.org/10.16835/j.cnki.1000-9817.2018.12.013 https://doi.org/10.16237/j.cnki.cn44-1404/g8.20161010.005 https://doi.org/10.16237/j.cnki.cn44-1404/g8.20161010.005 https://doi.org/10.16099/j.sus.2021.01.004 http://www.cnki.com.cn/Article/CJFDTotal-ZJTK201803019.htm http://www.cnki.com.cn/Article/CJFDTotal-ZJTK201803019.htm https://doi.org/10.16518/j.cnki.emae.2013.03.006 https://doi.org/10.16518/j.cnki.emae.2013.03.006 OriginalArticle-JijunYao-BECE_title_July2021 OriginalArticle-JijunYao--BECE_MaintextJuly2021