Analysis Of Factors Influencing Belief In A Just World For Children And Adolescents Nationwide Based On Machine Learning

Research BackgroundBelief in a just world(BJW)is a major psychological resource for children and adolescents’ development and is especially important for students from low-income families to achieve vertical mobility.Late childhood and the onset of adolescence are critical periods for the development of BJW,and exploring the factors influencing children’s BJW at this stage is of great significance.However,most studies have focused less on the influencing factors of BJW,and the limited Western studies have focused more on the influencing factors of personal BJW and less on the influencing factors of general BJW,and even fewer studies have examined the influencing factors of personal BJW jointly with general BJW.In addition,studies that incorporate individual,family,school,and social factors simultaneously have not been seen published.Therefore,it is of great theoretical and practical importance to examine the influential factors of personal BJW,general BJW,and the joint index of the two,using large national data,in terms of macro-and meso-micro factors.Research ObjectiveTo ensure the mental health and future development of children and adolescents,this study examines the factors that have a significant impact on the healthy development of children and adolescents’ personal and general BJW,and explores the impact of various factors on the combination of both personal and general BJW by constructing the hope index(General BJW-Personal BJW/Personal BJW)Subject and methodsThis study used the social adjustment database(24,103 students in grades 4-9)from the Chinese Children and Adolescents Psychological Developmental Characteristics Survey Project to select 70 variables related to the development of personal,general BJW from four aspects:personal,family,school,and social.Because of the large sample size,variable characteristics,and complex relationships,the study used machine learning to build five nonlinear models,the random forest algorithm(RF),the adaptive boosting algorithm(AdaBoost),the extreme gradient boosting tree algorithm(XGBoost),the light gradient boosting algorithm(LightGBM),and the nearest neighbor algorithm(KNN);and two linear models Support vector regression(SVR)using linear kernel and multiple linear regressionfunctions,respectively,and tuning based on model evaluation metrics to obtain the feature importance of the optimal model.Results(1)The results of the ten-fold cross-validation show that the better performing regression model for personal BJW is the Light Gradient Boosting algorithm(LightGBM)with a coefficient of determination(R2)of 0.420 and a root mean square error(RMSE)of 0.761,and for general BJW,the better performing regression model is the Light Gradient Boosting algorithm(LightGBM)with a coefficient of determination(R2)of 0.345 and a root mean square error(RMSE)of 0.810.The better performing regression model for the hope index was the Random Forest algorithm(RFR)with a coefficient of determination(R2)of 0.345 and root mean square error(RMSE)of 0.810.(2)The results of feature importance showed that among the factors influencing personal BJW,authoritative parenting style took the first place with a relative importance of 0.046.Among the factors influencing general BJW,collectivism took the first place with a relative importance of 0.131,far exceeding the second age factor with a relative importance of 0.077.Among the factors influencing the hope index,the trust and support of parents and the trust and support of classmates for themselves are in the first and second place,with relative importance of 0.106 and 0.103,respectively.Among the three characteristic importance results,the role of tangible resources(annual family income,etc.)is observed only in the influencing factors of personal BJW,with a relative importance of 0.043.(3)The results of the weighting of characteristic importance in different environmental systems show that the influencing factors of personal BJW and hope index have similar patterns,i.e.,family>school>students’ own characteristics,with relative importance of 0.493(family),0.290(school),and 0.217(students’ personal factors)and 0.683(family),0.220(school),and 0.097(individual student factors).Among the factors influencing general BJW,personal characteristics,family factors,and school factors have more similar influence weights,with relative importance of 0.329,0.330,and 0.341,respectively.ConclusionsThis study found that family has a more significant influence on children and adolescents’ personal BJW and hope index,and factors outside the family have a more significant influence on children and adolescents’ general BJW.Among the three influencing factors,intangible resources were important.