Application In Analyzing Behavioral Problems Of Middle School Students In Hang Zhou City

These years, behavioral and emotional problems of children and adolescents have become increasingly prominent in the world. The situation of children and adolescents' psychological health is severe in both developed and developing countries. In developed countries, prevalence rates of behavioral and emotional problems among children and adolescents have been reported to be between 6.2% and 41.3%, and prevalence rates for behavioral and emotional among Chinese adolescents have been found to range 8% to 25%. And Behavioral problems have negative effect on children and adolescents' development, which may lead to study failure, adolescent delinquency, and psychological problems when they are adults. That is to say, the influence is long lasting. Therefore, how to prevent and interfere with their behavioral problems has become an international issue.Prevention and intervention should start from analyzing effects. At present, statistical methods often used in behavioral research include multiple linear regression, principal component analysis, factor analysis et al. However, these methods have their limits in analyzing data of behavioral problems. Structural equation modeling (SEM) is a combination of two important statistical techniques, i.e. factor analysis and path analysis. It makes observedvariables indicate latent variables, and allows measurement errors between them. It can study the manifest variables as well as latent variables;meanwhile, it is able to probe into both direct and indirect effects between these variables. Therefore, structural equation modeling has been adopted in research of behavioral problems and their effects. In this study, modern statistical modeling techniques were applied to show from different perspectives the results. Both direct and indirect effects of all influential factors of behavioral problems are analyzed in this paper, and strategies are suggested to prevent and intervene behavioral problems for middle school students in Hang Zhou City.ObjectivesThe main objectives of this thesis are: to find out the status quo of behavior problems of middle school students in Hang Zhou and explore differences between boys and girls;to build a cause -effect model on the basis of factor analysis;and then to modify and evaluate the model so as to get an optimal model;to make sure the influential factors of these students' behavioral problems;besides, to explore the direct and indirect effects of these factors, and to provide scientific evidences for prevention and intervention of behavioral problems of middle school students in Hang Zhou City.Material and MethodsIn this study, By random group sampling method, 2600 middle school students from eight districts and six counties in Hang Zhou City, their mean age is 14.9±1.7 years old, of the 2600 middle school students recruited, there are 1140 boys and 1100 girls. Then they were investigated by using Children behavioral Checklist and Youth Self-Report scales, raw data was input into computer by the Epidata3.1, and those questionnaires with more than eight items missing were eliminated (except for the second, fourth and fifty-sixth h) as well as those whose answers were illogical. Finally, 2240 effective questionnaires were obtained;valid respondentrate was 86.2%.SPSS 13.0 was used for statistical analysis, and data were analyzed by Descriptive statistic, Chi-square test, Reliability analysis and factor analysis. Data multi-normality was tested by using LISREL8.70, and non-normality continuous variables were transformed by using Box-Cox method in STATA 8.0. Structural equation model was analyzed by using AMOS4.0 produced by the company SmallWater.Main Results1. Prevalence of Behavioral and Emotional problemsThe overall prevalence rates of parent-and Youth self-reported behavioral problems were 20.3% (95% confidence interval [CI] =18.6-21.9%) and 17.7% (95% confidence interval [CI] =16.1-19.3%), respectively. Externalizing behavior was more prevalent in boys according to both parent and youth self-reports, however, Internalizing behavior was more prevalent in girls. Girls' prevalence rate was higher that of boys according to parent and youth self-reports, and parent report remained statistically significant (i><0.05), but Youth self-report did not (p>0.05). Of eight syndromes, Withdrawn, Somatic Complaints, Anxious/Depressed, Social Problems were more prevalence in girls than in boys;While Delinquent, AggressiveBehaviors and attention problem were more prevalence in boys than in girls, and both of the prevalence rates are statistically statistical significant (PO.05).2. Reliability and Validity testsCronbach's Alpha of eight syndromes, Externalizing and Internalizing and Total Problems for the CBCL and YSR reached to over 0.6 in all samplings, the two sexes, junior high school and senior high school students, only Alpha of girls' aggressive behavior was 0.58 according to YSR. It is generally considered that Alpha over 0.6 reflects good internal consistency, andthat assessment results can indicated that respondents' understanding of CBCL and YSR is what the devisers have expected. So Reliability of the two scales is rather.Confirmatory factor analysis was used to test the construct validity of the CBCL and YSR. Confirmatory factor analysis results showed that x2 RMSEAN and SRMR were both less than 0.05. These results indicate every one item of the two questionnaires has a good validity. But some subitem's factor loading was lower than 0.4, and t-test showed they did not have statistical significance.3. Factor analysisFactor analysis was used to analyze 29 variables, such as urban and rural, sex, age, parents' occupation and education, marriage status, family population, activities, social and school, and seven common factors were extracted. They were parent-report behavioral problems, Youth-self report behavioral problem, family economic status, parent qualities, individual situation, social competence and parental relationship. Factor loading all were over 0.5, and these common factors accounted for 85.1% of total variance, and thus can explain most of variance for these 29 variables.4. Result of Structural equation modeling analysisBased on theories of building structural equation modeling, six acceptable and competitive models were built in this study, and these models were evaluated by standard coefficient, %2fiFl, AGI, CFI, RMSEA and its 90% confidence interval et al, Then Maximal Likelihood method was used to compare these models. The comparison showed that model one was the best one of these six models and its GFI, AGFI, NFI, CFI, RMSEA and its 90% confidence interval were 0.956, 0.932, 0.984, 0.985, 0.031 and 0.018-0.040 respectively. These results indicated model one was fit well. Meanwhile, the final model results showed family economy status, parental relationship, parent qualities, social competence and individual situation had direct effect on behavioral problems. Of these direct effects, family economy status, parental relationship, parent quality and individual situations affected behavioral problems indirectly through social competence. Furthermore, parent quality also affected behavioral problems through family economic status. Finally, it was concluded that the main factors influencing behavioral problems were family economic status, parent quality and parental relationship. The standardized coefficients of theirtotal effects on behavioral problems were -0.47,0.38, -0.36 respectively.Conclusions1. The reliability for the CBCL and YSR scales indicated that the research results were reliable. Construct validity showed every one item of the two questionnaires had a good validity, but some subitems' factor loading was low. It implied that these subitems need to be revised or verified further in other studies.2. Prevalence rates of total behavioral problems for middle school students in Hang Zhou City were consistent with the result of previous studies in china;and total prevalence rates for CBCL and YSR were more prevalent in girls than in boys;Externalizing behavior was more prevalent in boys according to both parent and youth self-reports, while Internalizing behavior was more prevalent in girls.3. The main factors influencing behavioral problems of middle school students in Hang Zhou City were family economic status, parent quality, and parental relationship by using structural equation modeling. Furthermore.it reveals that these effects had direct and indirect influence.The final models fit well.4. Structural equation modeling is the organic combination with mathematics and speciality knowledge, which is relatively satisfactory method to deal with data of behavioral research, as it resolves some problems in methodology. It possesses some advantages compared with conventional statistical methods. But it has its own limitation and shortage. For example, the 6 theoretical models and definitions of the latent variables in this study are subject to some extent.