Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 8, No. 2, 141-155 2024 Publisher: Learning Gate DOI: 10.55214/25768484.v8i2.1763 © 2024 by the author; licensee Learning Gate © 2024 by the author; licensee Learning Gate * Correspondence: 18508318855@163.com A study on the revision of adolescent academic burden attitude scale Guoliang Deng Krirk University, International College,Bangkok,Thailand,10220; 18508318855@163.com (G.D.) Abstract: The problem of excessive academic burden among adolescents has been hindering the implementation of quality education, and the effective measurement of adolescents’ attitudes towards academic burden is particularly important for later research. Objectives: To revise the Academic Burden Attitude Scale so as to increase its convenience for future scholars. Methods: The questionnaire survey method was used to distribute and collect the Academic Burden Attitude Scale compiled by Zhang Feng et al., and 500 (boys = 301, girls = 199) middle school students in Sichuan, China were distributed and collected by exploratory factor analysis. Validation factor analysis was conducted to distribute and recycle 300 (boys = 179, girls = 121) middle school students in Sichuan, China. Results: Through the revision of the Academic Burden Attitude Scale, the Academic Burden Attitude Scale was finally obtained, which consisted of three subscales and a total of 35 items. There were four common factors with a root greater than 1 (1.004 - 4.140) in the cognitive component of academic burden attitude, and the cumulative contribution rate of the scale was 73.703%. The emotional component of academic burden attitude showed three common factors (1.198 - 4.034) with characteristic roots greater than 1, and the cumulative contribution rate of the scale was 73.094%. There were four common factors (1.054 - 4.538) with characteristic roots greater than 1, and the cumulative contribution rate of the scale was 72.236%. The overall reliability of the revised Academic Burden Attitude Scale was 0.929 and the halving coefficient was 0.815, indicating high reliability. The confirmatory factor analysis of AMOS 22.0 was used in each subscale of the questionnaire, and the structural validation analysis values reached the significance standard (P <0.001), and the fitting indicators of goodness reached the suitability standard. Discussion: The 35-item version of the Academic Burden Attitude Scale revised in this study meets the relevant statistical standards and can be used as an evaluation tool for related studies. Keywords: Academic Burden Attitudes, Adolescents, Scale Revisions. 1. Introduction “Academic Burden Attitude” refers to the stable behavior and psychological tendency of students under academic burden, and is (deleted) specifically defined into three basic dimensions of cognition, emotion, and behavior (Zhang et al., 2004) [1]. For a long time, the problem of excessive academic burden on primary and secondary school students has been hindering the implementation of quality education (deleted), and excessive academic burden will have a serious negative impact on children’s physical and mental health. Studies have shown that excessive academic burden can have a serious impact on the mental health development of adolescents, which is mainly manifested in prominent emotional maladjustment problems, mental health problems, and mental illness (Lin Guozhen, 2010) [2], and insecurity, sadness, nervousness, and uncontrollable impulsiveness (Dong Yan and Yu Guoliang, 2010) [3]. Epkins and Seegan (2015) [4] believe that students with negative academic burden attitudes, even if the objective burden is low, will feel the great pressure brought by the academic burden and the psychological burden will be heavier. Shi (2022) [5] said in the study that although there are many documents dedicated to reducing the academic burden of adolescents, the 142 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 8, No. 2: 141-155, 2024 DOI: 10.55214/25768484.v8i2.1763 © 2024 by the author; licensee Learning Gate actual situation of adolescents' academic burden is still not optimistic (deleted), and adolescents still have problems such as excessive learning tasks, excessive study time, and excessive learning pressure. The measurement of academic burden attitude has become an important tool for in-depth research. The accuracy and convenience of measurement tools have become important considerations for later scholars. The scale of “Attitudes of Middle School Students’ Academic Burden” developed by psychological researcher Zhang Feng et al. (2004) [1] includes three dimensions: cognition, emotion, and behavior, and the cognitive aspect mainly includes students' perception and evaluation of the severity of academic burden. The emotional aspect mainly includes students' emotional experience of academic burden, focusing on the examination of self-feelings. The behavioral aspect mainly includes the actions taken by students and the behavior patterns shown by students under academic load. From the compilation of the scale to the present, it has been verified by many scholars and is one of the most frequently used tools (Jin Yule and Zhao Ruixue, 2023) [6]. However, some scholars have put forward some opinions in the process of using it (Li Yajun, 2020) [7], which can further improve the scale. First, there are many questions on the scale, and during the research process, many middle school students reported that the scale was too long and it took too much time, and they were a little impatient to fill in the back (Wu Fang and Ren Yuanhang, 2023). Question 2: It has been almost 20 years since the questionnaire was compiled, during which the situation and social status of students have undergone great changes (Liu Axiu and Feng Lei, 2024). According to the survey results, it also shows that Cronbach's a score is higher after the deletion of individual questions, and individual questions can be finely adjusted. Therefore, this study began to revise the scale of “Academic Burden Attitudes of Middle School Students” compiled by Zhang et al. (2004) [1], so as to make it more convenient for later scholars to use. The number of items in the revised scale is reduced, the reliability is reduced due to too many questions, the efficiency of scholars is improved, and the factors of each subscale are refined, so that the goals of scholars are more accurate when intervening in the attitude of primary and secondary school students towards academic burden. 2. Research Subjects and Methods 2.1. Research Subjects 2.1.1. Subjects for Exploratory Factor Analysis The principle of exploratory factor analysis using convenient sampling, Including No. 4 High School of Yibin, Yibin Tianli School, Jiang’an County Han’an Junior High School, No. 3 High School of Jiang’an County, and Changning County Yujiang Junior High School, with 100 students selected from each school, and a total of 500 students were selected as the study subjects, including 301 boys and 199 girls, namely 170 in the first grade, 162 in the second grade, and 168 in the third grade. 2.1.2. Samples for Confirmatory Factor Analysis According to the principle of convenient sampling, 35 items were compiled into the “Academic Burden Attitude Scale” for redistribution. The scales were distributed and collected from the students of Yibin Tianli School, Jiang’an County Han’an Junior High School, and Jiang’an County Hongqiao Town Junior High School. 100 students were selected from each school, and a total of 300 students were selected as the sample for this study, including 179 boys and 121 girls.94 students in the first grade, 102 students in the second grade, and 104 students in the third grade. According to the principle of invalid questionnaire, the collation of this questionnaire is consistent with the principle of data collation collected from the first preliminary questionnaire. 2.2. Research Methods 2.2.1. Questionnaire Method This study adopted the Academic Burden Attitude Scale for Middle School Students compiled by Zhang Feng et al. (2004) [1], with a total of 62 items, including three subscales of cognition, emotional experience, and behavioral tendencies towards academic burden. The academic burden cognition 143 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 8, No. 2: 141-155, 2024 DOI: 10.55214/25768484.v8i2.1763 © 2024 by the author; licensee Learning Gate subscale includes five dimensions: academic workload, teaching and assessment methods, academic difficulty, exam ranking, and elder expectations; the academic burden attitude emotion subscale includes four dimensions: learning interest, learning anxiety, emotions under load reduction, and workload emotions; the academic burden attitude behavior subscale includes four dimensions: learning initiative, learning adaptability, learning planning, and learning tolerance. The three subscales of this scale have Cronbach’s a between 0.861 - 0.902, and the scale has high reliability and validity. This scale uses a 5- point scoring system, where 1 represents “completely inconformity”; 5 represents “complete conformity”, and the higher the score on the Academic Burden Attitude Scale, the more negative the student. In order to meet the needs of this study, the Academic Burden Attitude Scale was distributed and collected from 500 middle school students in Sichuan Province, China, and 500 valid questionnaires were collected, with an effective recovery rate of 100%. The Revised Version of the Academic Burden Attitude Scale was distributed and collected to 300 middle school students in Sichuan Province, China, and 264 valid questionnaires were collected, with an effective recovery rate of 88.0%. The distribution of questionnaires is mainly based on online questionnaires (questionnaire stars). (Deleted). Principle of elimination of invalid questionnaires: (1) More than half of the questions are not answered. (2) There is no obvious rule to answer, for example, “4” is selected for the entire dimension, or “12345” is repeated to answer the rule. 2.2.2. Mathematical Statistics SPSS was used to analyze the reliability of the data, exploratory factor analysis, Harman factor analysis was used for common method bias analysis, etc., and AMOS was used to analyze the confirmatory factors of the data, mainly analyzing the standards of factor models such as 2, df, RMR, GFI, NFI, CFI, RMSEA, SMR, etc. 3. Results & Analysis 3.1 Reliability and Validity Before Amendment The Cronbach’s α scores of the three subscales of the Academic Burden Attitude Scale ranged from 0.628 to 0.765, the Cronbach’s α score of each dimension of the Academic Burden Attitude Cognition Subscale ranged from 0.621 to 0.723, the Cronbach’s α score of each dimension of the Academic Burden Attitude Emotion Subscale ranged from 0.554 to 0.823, and the Cronbach’s α score of each dimension of the Academic Burden Attitude Behavior Subscale ranged from 0.612 to 0.721. In terms of the Cronbach’s α score after the deletion of the items, the Cronbach’s α scores of Items 1, 9, 11 and 22 of the Academic Burden Attitude Cognition Subscale were higher than that of relevant dimension before deletion. Items 30, 32, 37, and 41 of the Academic Burden Attitude Emotion Subscale scored higher than the Cronbach’s α coefficient of the dimension after removing the items. Items 48, 50, 54, and 56 of Academic Burden Attitude Behavior have higher scores than the Cronbach’s α coefficient of the dimension after deletion, and the above questions may need to be further modified. 3.2. Differentiation of Questions Before Amendment The differentiation of the questionnaire is an important criterion to measure the quality of the questionnaire preparation. Here, we use the “cut-off ratio” to analyze the discrimination of the questionnaire questions. The critical ratio value (CR value) method is used to test the significance of the difference between the average of the top 27% of the high group and the bottom 27% of the total score of each item, and if the result is not significant, it means that the item has low discrimination power and should be deleted (Wu Minglong, 2001). If the "cut-off ratio" of a question item does not reach a significant level, indicating that there is no significant difference between the high and low group scores of the item, it means that the item does not have good discrimination, indicating that the item cannot distinguish the degree of response of different participants and can be deleted (Andreu, 2020) [8]. 144 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 8, No. 2: 141-155, 2024 DOI: 10.55214/25768484.v8i2.1763 © 2024 by the author; licensee Learning Gate According to the statistical results of the mean equivalence t-test (two-tailed test), among the 62 items, the T values of V9 (t = -1.265, p = 0.207), V22 (t = -2.319, p = 0.121), V30 (t = 0.437, p = 0.663), V31 (t = -1.049, p = 0295), and V45 (t = -2.317, p = 0.171) did not reach a significant level, indicating that these items do not show good discrimination. So, we can consider deleting them when revising the questionnaire in the future. Table 1. KMO and bartlett test of the academic burden scale. Test method Valid values Cognitive subscale Emotion subscale Behavior subscale KMO test KMO values 0.957 0.938 0.892 Bartlett’s sphericity test Approximate chi-square 6991.430 5864.114 4415.182 Variance 300 190 136 Significance 0.000 0.000 0.000 3.3. Results and Analysis of Exploratory Factor Analysis 3.3.1. KMO And Bartlett’s Sphericity Test If the KMO value is below 0.60 as calculated by SPSS 25.0, it indicates that the sample data is completely unsuitable for the factor analysis (Angelica, et al., 2018) [9]. The KMO values of the three subscales of the Academic Burden Scale in this study ranged from 0.957 to 0.892, Approx. Chi-Square = 4415.182 - 6991.430, and all p values were < 0.001, indicating that the results of the Bartlett sphericity test reached a significant level (Zhao Shushen (2005)) [10]. The above results show that there are common factors among the correlation matrices of the population, indicating that the sample data in this study are suitable for factor analysis, as shown in Table 1. 3.3.2. Presentation of the Results of Exploratory Factor Analysis Cognitive components of academic burden attitude. Through data analysis and extraction of common factors, combined with gravel steep slope map, the cognitive component of academic burden attitude showed four common factors with characteristic roots greater than 1, and the cumulative contribution rate of the scale was 73.703%, and the exploratory factor rotation matrix for cognitive component of academic burden attitude, and the factor load value of the three items in the dimension of coursework was between 0.699 and 0.967. The factor load of the three items of coursework difficulty was between 0.768 and 0.970, the factor load of the three questions of the dimension teaching and examination dimension was between 0.803 and 0.865, and the factor load of parental expectations was between 0.692 and 0.911, and the specific scores are shown in Table 2. Table 2. Exploratory factor rotation matrix for cognitive components of academic burden attitude. Item Factor Load of coursework Teaching & examinations Difficulty of coursework Parental expectations V5 0.845 V6 0.967 V7 0.699 V12 0.849 V13 0.803 V15 0.865 V17 0.768 V18 0.970 145 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 8, No. 2: 141-155, 2024 DOI: 10.55214/25768484.v8i2.1763 © 2024 by the author; licensee Learning Gate Item Factor Load of coursework Teaching & examinations Difficulty of coursework Parental expectations V19 0.791 V23 0.692 V24 0.807 V25 0.911 Eigen- value 4.14 2.107 1.974 1.004 Contributi on rate 31.165% 19.221% 17.951% 5.367% (1) Emotional Component of Academic Burden Attitude. Through data analysis and extraction of common factors, combined with the gravel steep slope map, the emotional component of academic burden attitude showed three common factors with characteristic roots greater than 1, and the cumulative contribution rate of the scale was 73.094%, and the exploratory factor rotation matrix of the emotional component of academic burden attitude was 0.844 - 0.929, and the factor loading value of the three items in the dimension of academic learning interest was between 0.844 and 0.929. The factor loads of the 3 items in the learning anxiety dimension was between 0.788 and 0.834, and the factor load of the 4 items in the learning emotion dimension was between 0.713 and 0.844, and the specific scores are shown in Table 3. Table 3. Exploratory factor rotation matrix for the emotional component of academic burden attitude. Item Factor Learning interest Learning anxiety Learning emotion V27 0.929 V28 0.860 V29 0.844 V33 0.834 V34 0.822 V35 0.788 V39 0.713 V42 0.844 V43 0.823 V44 0.844 Eigen-value 4.034 2.278 1.198 Contribution rate 30.339 22.778 19.978 (2) Behavior components of academic burden attitude. Through the data analysis and extraction of common factors, combined with the gravel steep slope map, the academic burden attitude and behavior component presented a total of 4 common factors with characteristic roots greater than 1, and the cumulative contribution rate of the scale was 72.236%, and the exploratory factor rotation matrix for behavior components of academic burden attitude, and the factor loading values of the three items in the initiative dimension of the academic learning were between 0.571 and 0.907. The factor loads of the four items of learning were between 0.763 and 0.891, the factor loads of the three questions of the learning adaptability dimension were 0.724 and 0.896, and the factor loads of the three questions of the leaning affordability dimension were between 0.598 and 0.833, and the specific scores are shown in Table 4. 146 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 8, No. 2: 141-155, 2024 DOI: 10.55214/25768484.v8i2.1763 © 2024 by the author; licensee Learning Gate Table 4. Exploratory factor rotation matrix for behavior components of academic burden attitude. Factor Initiative Adaptability Plan Affordability V46 0.907 V47 0.886 V49 0.571 V51 0.763 V52 0.891 V53 0.795 V55 0.808 V57 0.724 V58 0.896 V59 0.892 V60 0.788 V61 0.833 V62 0.598 Eigen-value 4.538 3.194 1.919 1.054 Contribution rate 34.905% 24.572% 7.065% 5.804% 3.4. Results And Analysis of Confirmatory Factor Analysis (1) Results of confirmatory factor analysis of cognitive subscales. For example, the importance of the topics V5, V6, and V7 to the “load of coursework” in this study is 0.83, 0.77, and 0.81, respectively, and it can be seen that the path coefficients of each topic of the cognitive component of the karmic burden attitude are between 0.54 and 0.83, all of which are greater than the standard of 0.50. The number on the double arrow symbol in Figure 1 represents the correlation coefficient between the two constructive factors (i.e., dimensions), and according to the requirements of statistics, the correlation between each dimension should be less than 0.8 as much as possible, so as to indicate that the structure between dimensions is better. For example, the correlation coefficient between “load of coursework” and “parental expectations” was 0.58, The correlation coefficient with coursework difficulty is 0.59 (deleted). The correlation coefficient between the dimensions of this study was below 0.8, except for “load of coursework” and “teaching and examination”, which was 0.81. It shows that the structure of this dimension is ideal, and the statistical effect of each dimension and topic of the theoretical conception is better. 147 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 8, No. 2: 141-155, 2024 DOI: 10.55214/25768484.v8i2.1763 © 2024 by the author; licensee Learning Gate Figure 1. Confirmatory analysis of the cognitive component structure of academic burden attitude. Table 5. Goodness fit indicators for confirmatory factor analysis of cognitive components of academic burden attitude. df χ2 RMR GFI NFI CFI RMSEA SRMR 48 110.533 0.045 0.938 0.928 0.958 0.068 0.042 As can be seen from Table 5, the fit index 2/df = 2.303<3.0, RMR = 0.045<0.05, GFI = 0.938 >0.90, NFI = 0.928 >0.90, CFI = 0.958>0.90, RMSEA = 0.068<0.08, SRMR=0.042<0.05. The results indicated that all indicators reached a good fitting index (Wu, 2013) [14], indicating that the model fitting index met the goodness fitting conditions and met the assumptions of the previous model. （2）Results of confirmatory factor analysis of emotion subscales. The number on the single arrow symbol in Figure 2 represents the factor load of each item, also known as the path coefficient, which represents the relative importance of the indicator variable to the construct factor. For example, the importance of the items V27, V28, and V29 to the " interest in learning" is 0.74, 0.83, and 0.77, respectively, and it can be seen from the figure that the path coefficients of each topic of the emotional component of the karmic burden attitude are between 0.62 and 0.83, all of which are greater than the standard of 0.50. The number on the double arrow symbol in Figure 2 represents the correlation coefficient between the two constructive factors (i.e., dimensions), and according to the requirements of statistics, the correlation between dimensions should be less than 0.8 to indicate that the structure between dimensions is better. For example, the correlation coefficient between the two constructive factors of learning interest and learning emotion was 0.76, The correlation coefficient with “learning anxiety” was 0.61 (deleted). The correlation coefficient between the dimensions of this study was between 0.61 and 0.76, and the correlation was below 0.8. It shows that the structure of this dimension is ideal, and the statistical effect of each dimension and topic of the theoretical conception is better. Load of coursework Teaching & examinations Difficulty of coursework Parental expectations 148 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 8, No. 2: 141-155, 2024 DOI: 10.55214/25768484.v8i2.1763 © 2024 by the author; licensee Learning Gate Figure 2. Confirmatory analysis of the emotional component structure of academic burden attitude. Table 6. Goodness fit indicators for confirmatory factor analysis of the emotional component of academic burden attitude. df χ2 RMR GFI NFI CFI RMSEA SRMR 32 53.7 0.041 0.962 954 0.981 0.071 0.039 At the same time, the absolute fitness index of 2, df, (2/df), RMR, GFI, NFI, CFI, RMSEA, SRMR and other absolute fitness indexes were used to evaluate the structural model of this study. As can be seen from Table 6, the fitness index 2/df = 1.678<3.0, RMR = 0.037<0.05, GFI = 0.962>0.90, NFI = 0.954>0.90, CFI = 0.981>0.90, RMSEA = 0.071<0.08, SRMR = 0.039<0.05. The results indicated that all indicators reached a good fitting index (Wu, 2013) [14], indicating that the model fitting index met the goodness fitting conditions and met the assumptions of the previous model. （3）Results of confirmatory factor analysis of behavior subscales. The number on the single arrow symbol in Figure 3 represents the factor load of each question, also known as the path coefficient, which represents the relative importance of the indicator variable to the construct factor. For example, the importance of the items V46, V47, and V49 to “initiative” is 0.83, 0.83, and 0.56, respectively, and it Learning interest Learning emotion Learning anxiety 149 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 8, No. 2: 141-155, 2024 DOI: 10.55214/25768484.v8i2.1763 © 2024 by the author; licensee Learning Gate can be seen from the figure that the path coefficients of each topic of the emotional component of karmic burden attitude are between 0.56 and 0.85, all of which are greater than the standard of 0.50. The number on the double arrow symbol in Figure 3 indicates the correlation coefficient between the two constructive factors (i.e., dimensions), and according to the requirements of statistics, the correlation between dimensions should be less than 0.8 to indicate that the structure between dimensions is better. For example, the correlation coefficient between “initiative” and “affordability” is 0.76,Correlation coefficient with “planned” is 0.07 (deleted). The correlation coefficients between each dimension were below 0.8, except for “planning” and “affordability”, which were 0.81. It shows that the structure of this dimension is ideal, and the statistical effect of each dimension and topic of the theoretical conception is better. Figure 3. Confirmatory analysis of the component structure of academic burden attitude and behavior. Table 7. Goodness fit indicators for confirmatory factor analysis of academic burden attitude and behavior components. df χ2 RMR GFI NFI CFI RMSEA SRMR 59 165.865 0.032 0.912 0.904 0.938 0.079 0.045 At the same time, the absolute fitness index of 2, df, (2/df), RMR, GFI, NFI, CFI, RMSEA, SRMR and other absolute fitness indexes were used to evaluate the structural model of this study. As can be seen from Table 7, the fitness index 2/df = 2.811<3.0, RMR = 0.047<0.05, GFI= 0.912 >0.90, NFI = Initiative Plan Affordability Adaptability 150 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 8, No. 2: 141-155, 2024 DOI: 10.55214/25768484.v8i2.1763 © 2024 by the author; licensee Learning Gate 0.904 >0.90, CFI = 0.938 >0.90, RMSEA = 0.079<0.08, SRMR = 0.045<0.05. The results indicated that all indicators reached a good fitting index (Wu, 2013) [14], indicating that the model fitting index met the goodness fitting conditions and met the assumptions of the previous model. 3.3.4. Differentiation of the Questions in the Revised Questionnaire “Critical ratio” is also used to analyze the discrimination of questionnaire questions. According to the statistical results of the mean equivalence t-test (two-tailed test), the discrimination of 35 items met the statistical requirements, indicating that each question had a good degree of discrimination. 3.3.5. Reliability And Validity of the Modified Questionnaire In this study, four indicators were mainly used to test the reliability of the internal consistency of the composite scale, namely the Cronbach’s α consistency coefficient, the halving coefficient Guttman half, the CITC coefficient, and the deleted Cronbach’s α value (Hassan, et al., 2020) [15]. The validity test was mainly performed by factors such as factor loading coefficients, correlation between dimensions and correlation with the total topic, AVE value, CR value, and comparison of the square root of AVE value with each correlation coefficient (Ryan, & Deci, 2020) [16].When the CITC value is less than 0.40 and the Cronbach's a value increases significantly after deletion, the internal consistency between the item and the overall scale is low and needs to be deleted (Churchill, 1979) [17]. （1）Reliability of the modified questionnaire. (1) Reliability of academic burden attitude cognitive components. The results of the reliability test of Cronbach's a coefficient and halving coefficient in this study show that the Cronbach's a is between 0.699-0.845 and the halving coefficient is 0.669 - 0.765 in each dimension, and the overall Cronbach's a is 0.898 and the overall halving coefficient is 0.755, indicating that the reliability of the measurement results of this study scale is high. When the CITC value is less than 0.40 and the Cronbach’s a value increases significantly after deletion, the item has low internal agreement with the overall scale and needs to be deleted (Churchill, 1979) [17]. From the CITC value of this study and the measurement results of the alpha reliability test after deletion, it can be seen that the CITC of each item is between 0.547 and 0.810, which is greater than the standard of 0.40, and the alpha value after deletion is less than the overall Cronbach's a value, so it can be seen that the 12 items have high internal consistency, and the specific data are shown in Table 8. Table 8. Reliability analysis of cognitive component of academic burden attitude. Dimension Topic CITC Cronbach’s α after deleting items Cronbach’s α Half factor Guttman half Amount of coursework A5 0.746 0.751 0.845 0.738 A6 0.701 0.794 A7 0.688 0.808 Teaching & Examinations A12 0.705 0.772 0.838 0.706 A13 0.735 0.742 A15 0.665 0.810 Difficulty of coursework A17 0.620 0.701 0.780 0.669 A18 0.656 0.659 A19 0.578 0.747 Parental expectations A23 0.563 0.547 0.699 0.765 A24 0.531 0.587 A25 0.454 0.683 Cognition subscales 0.898 0.775 151 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 8, No. 2: 141-155, 2024 DOI: 10.55214/25768484.v8i2.1763 © 2024 by the author; licensee Learning Gate (2) Academic burden, attitude, emotional component reliability. The results of the reliability test of Cronbach’s a coefficient and halving coefficient in this study show that the Cronbach's a is between 0.731 - 0.837 and the halving coefficient is 0.686-0.862 in each dimension, and the overall Cronbach's a is 0.858 and the overall halving coefficient is 0.801, indicating that the reliability of the measurement results of this study scale is high. When the CITC value is less than 0.40 and the Cronbach’s a value increases significantly after deletion, the intrinsic consistency of the item with the overall scale is low and needs to be deleted (Churchill, 1979) [17] (deleted). From the CITC value of this study and the measurement results of the alpha reliability test after deletion, it can be seen that the CITC of each item is between 0.555 and 0.732, which is greater than the standard of 0.40, and the alpha value of the deleted item is less than the overall Cronbach's a value, so it can be seen that the 10 items have high internal consistency, and the specific data are shown in Table 9. Table 9. Reliability analysis of emotional component of academic burden attitude. Dimension Topic CITC Cronbach’s α after deleting items Cronbach’s α Half factor Guttman half Learning interests A27 0.671 0.775 0.827 0.742 A28 0.692 0.754 A29 0.691 0.755 Learning anxiety A33 0.555 0.642 0.731 0.686 A34 0.613 0.569 A35 0.497 0.707 Learning emotions A39 0.658 0.798 0.837 0.862 A42 0.704 0.779 A43 0.732 0.764 A44 0.583 0.830 Emotion subscales 0.858 0.801 (3) Academic burden, attitude, and behavioral component reliability. The results of the reliability test of Cronbach’s a coefficient and halving coefficient in this study show that the Cronbach's a is between 0.760 - 0.863 and the halving coefficient is 0.653 - 0.856 in each dimension, and the overall Cronbach's a is 0.823 and the overall halving coefficient is 0.792, indicating that the reliability of the measurement results of this study scale is high. When the CITC value is less than 0.40 and the Cronbach’s a value increases significantly after deletion, the internal consistency of the item with the overall scale is low and needs to be deleted (Churchill, 1979) [17] (deleted). From the CITC value of this study and the measurement results of the alpha reliability test after deletion, it can be seen that the CITC of each item is between 0.443 and 0.719, which is greater than the standard of 0.40, and the alpha value of the deleted item is less than the overall Cronbach’s a value, so it can be seen that all 10 items have high internal consistency, and the specific data are shown in Table 10. Table 10. Reliability analysis of behavior component of academic burden attitude. Dimension Topic CITC After deleting the item Cronbach’s a Cronbach’s a Half factor Guttman half Initiative A46 0.660 0.598 0.838 0.654 A47 0.689 0.567 A49 0.443 0.738 Adaptability A51 0.733 0.816 0.863 0.856 A52 0.719 0.822 152 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 8, No. 2: 141-155, 2024 DOI: 10.55214/25768484.v8i2.1763 © 2024 by the author; licensee Learning Gate Dimension Topic CITC After deleting the item Cronbach’s a Cronbach’s a Half factor Guttman half A53 0.694 0.832 A55 0.696 0.831 Planned A57 0.700 0.656 0.795 0.653 A58 0.669 0.691 A59 0.554 0.815 Affordability A60 0.575 0.695 0.760 0.702 A61 0.596 0.673 A62 0.602 0.666 Behavior subscales 0.823 0.792 (2) Validity of the Modified Questionnaire Convergence validity reflects whether each indicator reflects the same construct. If the convergence validity is poor, it indicates that the constructs and connotations reflected by each index are different, so these indicators cannot be used to represent this factor together, and the score of the whole factor does not reflect a single connotation. Confirmatory factor analysis was used to calculate the factor load for each problem, and the larger the factor load value is, the better the convergence validity is. The AVE value reflects the average variance extraction of each dimension, and the larger the AVE value is, the better the convergence validity of the questionnaire is. Fornell, & Kohl (2013) [18] argue that when the AVE > 0.5, the structure of the questionnaire can be considered to have a high convergence validity. CR value is also an important indicator to judge convergence validity, which reflects the combined reliability of each dimension. It is generally accepted that a higher value indicates better convergence validity, and a CR value greater than 0.7 indicates a good convergence validity (Fornell, & Kohl, 2013) [18]. (3) Validity of Cognitive Components of Academic Burden Attitude The AVE value of each dimension of academic burden attitude cognition ranged from 0.545 to 0.646, and the CR value of each dimension ranged from 0.70 to 20.845, indicating that this scale met the statistical requirements of combined reliability. According to the factor load, AVE value and CR value of each topic, the comprehensive judgment indicates that the academic burden attitude cognitive subscale has high convergence validity, as shown in Table 11. Table11. Convergence validity analysis of cognitive components of academic burden attitude. Dimension Topic Factor loading AVE CR Load of coursework A5 0.830 0.646 0.845 A6 0.769 A7 0.811 Teaching & Examinations A12 0.765 0.633 0.838 A13 0.818 A15 0.803 Difficulty of coursework A17 0.776 0.546 0.782 A18 0.750 A19 0.687 Parental expectations A23 0.688 0.545 0.702 A24 0.758 A25 0.535 153 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 8, No. 2: 141-155, 2024 DOI: 10.55214/25768484.v8i2.1763 © 2024 by the author; licensee Learning Gate (4) Validity emotional component of academic burden attitude. According to the statistical structure of Table 12, the AVE value of each dimension of this scale is between 0.586 and 0.618, and the CR value of each dimension is between 0.736 and 0.839, indicating that this scale meets the statistical requirements of combined reliability. According to the factor load, AVE value and CR value of each topic, the comprehensive judgment indicates that the academic burden attitude cognitive subscale has a high convergence validity. Table 12. Convergence validity analysis of the emotional component of academic burden attitude Dimension Topic Factor loading AVE CR Learning interests A27 0.741 0.614 0.826 A28 0.833 A29 0.774 Learning anxiety A33 0.636 0.586 0.736 A34 0.822 A35 0.615 Learning emotions A39 0.742 0.618 0.829 A42 0.790 A43 0.824 (5) Validity of academic burden attitude behavior components. According to the statistical structure of Table 13, except for the questions V49 (0.561) and V59 (0.620), the factor load scores of the other questions were greater than 0.7. According to the results of this study, the knowledge subscale of academic burden attitude has good convergence validity. The AVE value of each dimension of this scale was between 0.514 and 0.610, and the CR value of each dimension was between 0.761 and 0.862, indicating that the scale met the statistical requirements of combined reliability. According to the factor load, AVE value and CR value of each topic, the comprehensive judgment indicates that the academic burden attitude cognitive subscale has a high convergence validity. Table 13. Convergence validity analysis of academic burden attitude behavior components. Dimension Topic Factor loading AVE CR Initiative A46 0.827 0.564 0.790 A47 0.833 A49 0.561 Adaptability A51 0.838 0.610 0.862 A52 0.771 A53 0.769 A55 0.744 Planned A57 0.851 0.586 0.806 A58 0.805 A59 0.620 Affordability A60 0.703 0.514 0.761 A61 0.703 A62 0.745 154 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 8, No. 2: 141-155, 2024 DOI: 10.55214/25768484.v8i2.1763 © 2024 by the author; licensee Learning Gate 4. Conclusion and Discussion Through the revision of the Academic Burden Attitude Scale, the Academic Burden Attitude Scale for middle school students was finally obtained (deleted) and was still composed of 3 subscales, and the number of items was reduced from 62 to 35. In the cognitive subscale of academic burden of middle school students, the test ranking and integration were integrated into teaching and examination. The emotions under the burden reduction and the homework volume emotions in the middle school students' academic burden emotion subscale were combined into learning emotions. The academic burden score is calculated, and the mean of the question scores of each subscale is the overall score of the scale. The score is between 1-5 points, with lower scores indicating lower academic load, and higher scores indicating higher academic load. The overall reliability of the Revised Academic Burden Attitude Scale was 0.929, the halving coefficient was 0.815, and the reliability of the questionnaire reached the standard of 0.8 or higher, with high reliability (Churchill, 1979) [17]. The confirmatory factor analysis of AMOS 22.0 was used in each subscale of the questionnaire, and the structural verification analysis values reached the significance standard (P <0.001), and the fitting indicators of each goodness reached the adaptation standard of 2/df <3.0, RMR <0.05, GFI >0.90, NFI >0.90, CFI ≥0.90, RMSEA<0.08, and SRMR< 0.05 (Fornell, & Kohl, 2013) [18], all questionnaire subscales passed the validity test. The results indicated that the reliability and validity of this scale reached the statistical standard after testing, and reached the statistical standard through confirmatory factor analysis and the fitting index of the structural model. Therefore, the Academic Burden Attitude Scale revised in this study can be used as an assessment tool for related studies (Wu, 2013) [14]. Xu Guoxing (2020) proposed that if there are too many items or a questionnaire is too long , it is easy to make the respondents bored in the process of filling in, and the students’ concentration and accuracy of answering questions will be reduced, resulting in the authenticity of the questionnaire is not high. The number of items on the revised scale was reduced to 35 before the revision of the 67 items on the Academic Burden Attitude Scale, which reduced the reliability of the scale due to too many items. At the same time, the efficiency of scholars’ use was improved, and the factors of each subscale were refined, so that the later scholars could intervene in the attitude of primary and secondary school students towards academic burden, and the goal was more accurate. This article can reinforce the importance of the revised scale by discussing its broader impact in improving student well-being, informing education policy, or promoting targeted interventions. 5. Study limitations Due to personal abilities, the study only included some middle school students from Sichuan Province, China as the research subjects, and it was not able to expand to the whole country. We look forward to further expanding the sample size in future research to make the data more representative. Copyright: © 2024 by the author. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/). References [1] Zhang Feng, Deng Chengqiong et al (2004). Development of Academic Burden Attitude Scale for Middle School Students. Psychological Science, 27(2), 449-452. [2] Lin Guozhen, Xu Xudong, et al (2010). Psychological Characteristics of Students with Learning Difficulties and Comprehensive Intervention. China Medical Herald, 7(14), 13-15.https://doi.org/10.3969/j.issn.1673- 7210.2010.14.008. [3] Dong Yan, Yu Guoliang (2010). The Effect of Adolescent Academic Emotions on Academic Achievement. Psychological Science (4), 934-937.https://doi.org/CNKI:SUN:XLKX.0.2010-04-046. [4] Epkins, C. C., & Seegan, P. L.(2015). Mother-reported and children's perceived social and academic competence in clinic- referred youth: unique relations to depression and/or social anxiety and the role of self-perceptions. Child Psychiatry & Human Development, 46(5), 656-670.https://doi.org/10.1007/s10578-014-0508-9. https://creativecommons.org/licenses/by/4.0/ 155 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 8, No. 2: 141-155, 2024 DOI: 10.55214/25768484.v8i2.1763 © 2024 by the author; licensee Learning Gate [5 ]Shi, R.H.(2022). Discussion on the elementary school mathematics homework design under the background of reducing burden. Foreign Edition: Educational Science, (4), 68-71. [6] Jin Yule, Zhao Ruixue (2023). The basic rationale of comprehensive management of academic burden. Inner Mongolia Social Sciences, 44(1), 186-192. [7] Li Yajun (2020). Suicidal psychological behavior of primary and secondary school students in Henan Province and its correlation with academic burden. Chinese Journal of School Health, 41(4), 4.https://doi.org/10.16835/j.cnki.1000- 9817.2020.04.016. [8] Andreu, J. M. P. (2020). Motivation, social anxiety, physical exercise practice and academic performance. Journal of Sport and Health Research, 12(1), 6. [9] Angelica, M., Katz, I., & Alesi, M. (2018). Scaffolding for motivation by parents, and child homework motivations and emotions: effects of a training programme. British Journal of Educational Psychology, 88(1),323-344. [10] Zhao Shushen (2005). Practical Sports Statistics. Beijing Sport University Press. [11]Elizabeth, F. A，Nicola, A. B.(2013). Child protection and out of home care: policy, practice, and research connections australia and new zealand. Psychosocial Intervention, 22(3), 175-184.https://doi.org/10.5093/in2013a21. [12] Fredricks, J. A., Reschly, A. L., Christenson, S. L.(2019). Handbook of Student Engagement Interventions. San Diego: AcademicPress, 6. [13] Du, Y., Dang, Q., Zhang, B., Wu, R., & Rasool, A.(2021). The effects of parenting differences on psychological resilience in adolescent students: the role of gratitude. Children and Youth Services Review, 130. https://doi.org/10.1016/j.childyouth.2021.106224. [14] Wu,M.L.(2013).Structural Equation Model—AMOS Practice Advanced. Chongqing: Chongqing University Press,2013: 22-24. [15] Hassan, A., Bashir, R., & Arshad, M. T.(2020). Learning engagement scale (les): development of a composite measure and analysis of psychometric properties for business schools. Journal of Business Strategies. https://doi.org/10.29270/JBS.14.2(20).03. [16]Ryan, R. M., & Deci, E. L.(2020). Intrinsic and extrinsic motivation from a self-determination theory perspective: definitions, theory, practices, and future directions. Contemporary Educational Psychology, 101860.https://doi.org/10.1016/j.cedpsych.2020.101860. [17]Churchill, G. A.(1979). A paradigm for developing better measures of marketing constructs. Journal of Marketing Research, 16(1), 12-27. [18]Fornell Cook, H. D. E., & Kohl92013), H. W. I. E. Educating the Student Body: Taking Physical Activity and Physical education to school. Washington, D. C: National Academies press. [19] Rong Taisheng (2010). AMOS and Research Methods. Chongqing University Press. [20] Liu A-Xiu, Feng L. (2024). Realistic Representation, Cause Analysis and Mitigation of the Utilitarian Tendency of Higher Education. Modern Communication (02), 95-103+124. [21]Wu Fang, Ren Yuanhang (2023). Alleviating the Learning Burnout of Engineering College Students: Improving Self- awareness from the Perspective of "Curriculum Ideology and Politics" and "Integration of Industry and Education". Journal of Chengdu University of Traditional Chinese Medicine (Educational Sciences) (04), 110-114. [22] Wu Minglong (2001). SPSS Statistical Application Practice. China Railway Press. [23] Xu Guoxing (2020). East China Normal University Press.