Application of random forests regression in impact of family early disadvantaged risks on internalization problems in adolescents
10.3760/cma.j.cn371468-20210411-00195
- VernacularTitle:家庭早期风险因素对青少年内化问题影响的随机森林回归分析
- Author:
Wenming XU
1
;
Xiaoqiong HAN
;
Haiyan YU
;
Wenjie TANG
Author Information
1. 嘉应学院心理咨询中心,梅州 514015
- Keywords:
Disadvantage;
Internalization problem;
Adolescents;
Random forest regression
- From:
Chinese Journal of Behavioral Medicine and Brain Science
2021;30(7):627-633
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To investigate the relationship between the type of exposure to early family disadvantaged risks and internalization problems in adolescents.Methods:Totally 746 adolescents were assessed with family disadvantaged risk items and Chinese version of Achenbach youth self-report (YSR-CV). The Welch's ANOVA test and post-hoc test were used to compare the scores of three different risk factors on internalization problems.The independent sample t-test was adopted to compare the scores of internalization problems between experienced 5 or more kinds of risk factors and less than 5.Pearson correlation analysis was used to test the correlation between the number of family risk factors experienced and the scores of internalization problems.Random forest regression analysis was used to test the variable importance(VI) of the internalization problem. Results:The differences in anxiety/depression and withdrawal scores between higher or lower-level risks group were significant ( P<0.05). The scores of adolescents with five or more adverse experiences on anxiety/depression(7.7±3.5), withdrawal (8.8±4.0) and physical problems(4.1±3.7) were higher than those with fewer than five risk factors(5.5±3.8), (6.7±3.4), (2.6±3.6). The cumulative family disadvantaged risk items was positively correlated with anxiety/depression( r=0.29, P<0.01), withdrawal( r=0.29, P<0.01), and physical symptoms ( r=0.26, P<0.01). The most important factor associated with anxiety/depression(VI=0.84, P=0.002; VI=0.56, P=0.022), withdrawal(VI=0.58, P=0.013; VI=0.89, P=0.001), and physical symptoms was marital relationship of parents and health status of family members. Conclusion:The parents’ marital relationship and health status of family members are the specific factors that influence the internalization problems, and the higher the level of risk adolescents experience, the more likely they are to develop internalization problems.